Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Bridging the gap between complexity science and clinical practice by formalizing idiographic theories: a computational model of functional analysis

Bridging the gap between complexity science and clinical practice by formalizing idiographic... Background: The past decades of research have seen an increase in statistical tools to explore the complex dynamics of mental health from patient data, yet the application of these tools in clinical practice remains uncommon. This is surprising, given that clinical reasoning, e.g., case conceptualizations, largely coincides with the dynamical system approach. We argue that the gap between statistical tools and clinical practice can partly be explained by the fact that current estimation techniques disregard theoretical and practical considerations relevant to psychotherapy. To address this issue, we propose that case conceptualizations should be formalized. We illustrate this approach by introducing a computational model of functional analysis, a framework commonly used by practitioners to formulate case conceptualizations and design patient-tailored treatment. Methods: We outline the general approach of formalizing idiographic theories, drawing on the example of a functional analysis for a patient suffering from panic disorder. We specified the system using a series of differential equations and simulated different scenarios; first, we simulated data without intervening in the system to examine the effects of avoidant coping on the development of panic symptomatic. Second, we formalized two interventions commonly used in cognitive behavioral therapy (CBT; exposure and cognitive reappraisal) and subsequently simulated their effects on the system. Results: The first simulation showed that the specified system could recover several aspects of the phenomenon (panic disorder), however, also showed some incongruency with the nature of panic attacks (e.g., rapid decreases were not observed). The second simulation study illustrated differential effects of CBT interventions for this patient. All tested interventions could decrease panic levels in the system. (Continued on next page) * Correspondence: j.burger@uva.nl University of Groningen, University Medical Center Groningen, University Center Psychiatry (UCP) Interdisciplinary Center Psychopathology and Emotion Regulation (ICPE), Hanzeplein 1, 9713 GZ Groningen, The Netherlands University of Amsterdam, Institute for Advanced Study, Amsterdam, The Netherlands Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Burger et al. BMC Medicine (2020) 18:99 Page 2 of 18 (Continued from previous page) Conclusions: Formalizing idiographic theories is promising in bridging the gap between complexity science and clinical practice and can help foster more rigorous scientific practices in psychotherapy, through enhancing theory development. More precise case conceptualizations could potentially improve intervention planning and treatment outcomes. We discuss applications in psychotherapy and future directions, amongst others barriers for systematic theory evaluation and extending the framework to incorporate interactions between individual systems, relevant for modeling social learning processes. With this report, we hope to stimulate future efforts in formalizing clinical frameworks. Keywords: Dynamical systems, Functional analysis, Computational modeling, Network analysis, Complex systems, Ordinary differential equations, Formalizing theories, Idiographic approach, Process-based psychotherapy, Theory development Background approach in mental health research, there has been spe- Complex system thinking is gaining increasing import- cific interest in implementing statistical tools to explore ance in understanding mental health [1–3]. In recent patient-specific symptom dynamics in clinical practice. It years, some clinicians have proposed a move away from is commonly assumed that successful implementation is the approach of treating mental illness as disorder cat- in part a question of providing technical trainings and egories towards a focus on processes and patient-specific accessible guidelines for clinicians [20]. However, merely mechanisms in psychotherapy [4]. These proposals call training clinicians in adopting tools provided by method- for a framework for thinking about mental illness in ologists does not guarantee that these tools also result in terms of systems, to understand the processes underlying models that map onto the language used by practi- psychopathology, and to apply this understanding to tioners. Indeed, an often-discussed barrier to implemen- patient-specific contexts. The network perspective to tation is the accurate translation of knowledge into the psychopathology [5–8], conceptualizing psychological relevant practice field [21]. That is, the language used to disorders as complex interactions of symptoms and re- discuss promising research findings and techniques does lated mental health factors, provides a framework to ad- not always match the targeted language of the dress this movement. Statistical procedures that allow practitioner. for the estimation of psychopathological networks have This issue applies to the estimation of personalized been developed [9–11] and applied across a wide range network models. At present, network estimation of mental disorders [12–15]. methods remain technical and do not account for poten- Furthermore, and arguably most relevant for psycho- tially relevant clinical considerations. For example, net- therapy, tools for idiographic network analysis have been work estimation methods identify “highly central” developed [16, 17], allowing us to explore patient- symptoms, given some assumptions, as promising targets specific symptom dynamics from data collected using of intervention [19], but these methods generally fail to the experience sampling method (ESM) [18]. This ap- account for the fact that symptoms differ in their amen- proach may be especially relevant for psychotherapy, as ability to psychological treatment or that some symp- it has the potential to be embedded within clinical prac- toms may have “low centrality” but remain critical tice through informing the formulation of idiographic targets for intervention because of their impact on psy- theories (i.e., case conceptualizations) and the identifica- chosocial functioning (e.g., suicidal thoughts and behav- tion of patient-tailored intervention targets [19]. Indeed, ior; [22, 23]). Further, currently available techniques to idiographic network analysis aligns well with the move- estimating personalized networks are primarily of ex- ment towards process-based psychotherapy [4]. It there- ploratory nature and do not allow clinicians to incorpor- fore seems surprising that, despite the availability of ate relevant a priori knowledge or clinical expertise. By supportive statistical tools and efforts to provide primers failing to see their ideas reflected in network models, for conducting idiographic research [20], the actual ap- practitioners might consider them as impractical and plication of personalized network modeling within psy- not in line with their clinical view, likely resulting in chotherapy is to date rare. hesitancy towards using personalized network models. Indeed, a recent study has shown that case conceptuali- zations greatly differ from temporal networks estimated From implementation barriers to a clinician’s wishlist from ESM data [24]. Implementation gaps between mental health research and clinical practice are a topic of enormous importance Based on these considerations, we argue that providing [21, 22]. With the emergence of the complex system trainings and guidelines is necessary, but not sufficient Burger et al. BMC Medicine (2020) 18:99 Page 3 of 18 in implementing the complex system approach in theories in psychiatry, including the relationship between clinical practice. For methods to be regarded clinically patient and therapist [31] and models of burnout [32, relevant, it is vital that tools have the flexibility to be 33], addiction [34], and panic disorder [35]. However, guided by clinical needs and allow practitioners to much remains unknown about precisely how such for- incorporate clinical considerations. mal theories should be developed and how they should be used in psychotherapy. The main objective of this Theories versus data models paper is to take a step towards addressing this gap in the In recent literature, special attention has been paid to literature by demonstrating the potential of formalizing disentangling conceptual aspects of data models and idiographic theories in clinical practice and illustrating theories. According to Haslbeck, Ryan, Robinaugh, an approach to formalizing such theories using the Waldorp, and Borsboom [25], data models (e.g., a mean, framework of functional analysis. correlation, or idiographic network model) are merely ways of representing or organizing data, often with the Approaches to constructing idiographic systems aim of establishing a phenomenon: a robust, We see two main ways of constructing personalized generalizable feature of the world identified through dynamical systems in psychopathology: First, modeling a empirical regularities [25, 26]. In contrast, the aim of a generic disorder model, and subsequently personalizing theory is to explain a phenomenon by representing those the model through estimating control parameters for the aspects of the real world that give rise to the equations in the system (top-down approach, cf. [35]), phenomenon. Whereas verbal theories are expressed in and second, modeling relations between specific variables language, formal theories are expressed in mathematical directly for and with each patient (bottom-up approach, equations or a computational programming language. cf. [36–39]). An advantage of the former approach is This level of specification allows formal theories to that it allows modeling individual differences between simulate theory-implied system behavior, and by observ- patients regarding the strength of shared relations (e.g., ing the effects of simulated interventions, we can draw person-specific tendencies to avoid when confronted conclusions about how the real-world system we are with fear), which consequently allows for examining for targeting would respond to a given treatment (a process instance tipping points in fear responses following referred to as “surrogative reasoning”, cf. [27]). maladaptive coping. An advantage of the latter approach In the following, we will refer to the approach of trans- is that it allows to flexibly model any psychological lating (verbal) case conceptualizations into mathematical hypotheses, as well as individual problems and resources systems as the formalization of idiographic theories. Al- [40]. though the term “theory” is commonly used to describe The method outlined in this paper is based on the phenomena on the nomothetic level, in this paper, we framework of functional analysis, and therefore utilizes are focused on the explaining phenomena at the level of elements of both approaches: On the one hand, func- the individual patient, and will use the term “idiographic tional analysis constitutes a generic framework for case theory” in respect to theorized relations within one formulation (top-down elements); on the other hand, it individual. also provides the flexibility to integrate patient-specific problems and resources (bottom-up elements). Formalizing idiographic theories To bridge the gap between methodological advances and The role of computational models in bridging the practical application of the complex system approach, scientist-practitioner gap we propose to derive dynamical system models directly We argue that formalizing idiographic theories provides from clinical theory, clinicians’ expertise and case- advantages for both, clinical practice and mental health specific knowledge. Formalizing patient systems tackles research, schematically displayed in Fig. 1, and is promis- the mismatch between technical tools and target ing in bridging the gap between the two. language as discussed above at its core; that is, rooting First, computational models of idiographic theories dynamical systems in the language of practitioners allows can be used to advance the current practice of a patient’s examining the patient’s system behavior based on clinic- case conceptualization. Sim, Gwee, and Bateman [40] ally relevant considerations. identified five key advantages associated with formulat- In other scientific disciplines like biology [28], ecology ing thorough case conceptualizations in clinical practice: [29], and political science [30], it is common to model (a) the integration/relation of multiple problems of a pa- dynamic processes based on theory and/or knowledge. tient, (b) the explanatory nature of the resulting model, Unfortunately, the application of formalized theories in (c) the prescription of interventions, (d) the prediction mental health research is to date extremely rare. of outcomes, and (e) the support for the therapeutic Recently, there have been efforts to propose formal relationship. Schiepek and colleagues [36, 37, 41] Burger et al. BMC Medicine (2020) 18:99 Page 4 of 18 Fig. 1 The role of computational modeling in bridging the scientist-practitioner gap. Schematic illustration of computational modeling (the product of formalizing a theory), at the intersection of clinical practice and mental health research. Computational models allow us to evaluate case conceptualizations in clinical practice (a–d), and bring clinical theories closer to empirical studies through guiding choices crucial to the estimation of and inferences drawn from data models (b, e–g) pioneered the integration of case formulation and are missing, or if included variables stem from theoretic- idiographic system modeling and argued that these key ally similar constructs, indicating topological overlap advantages could be strengthened through computa- [48]. Formalizing theories can provide useful informa- tional models. Clinicians are required to make more tion regarding the set-up of variables needed to retrieve rigorous decisions in specifying relations in the case clinical phenomena. Further, empirical research is often conceptualization, which makes the formalization of confronted with practical constraints to assessing idiographic theories a promising avenue to foster more psychological processes. Many clinically relevant psycho- scientific practices in designing patient-tailored treat- logical processes are difficult—sometimes even impos- ment. This reasoning is in line with a growing body of sible—to assess on their appropriate time-scale. For literature indicating the need for more rigorous theory practical reasons, variables are often measured within development in clinical and social sciences [25, 35, 42– the same time-scale (usually once a day or every few 45]. The left part of Fig. 1 illustrates how computational hours), potentially leading to biased estimates in dynam- modeling can inform case conceptualizations in clinical ical models. A recent simulation study suggests that practice: Formalizing a case conceptualization results in using the most commonly applied ESM time-intervals a computational model that allows the clinician to sub- results in data models that are largely unable to recover sequently simulate data, given the specified idiographic the micro dynamics of a system [49]. A stronger focus system. Based on these simulations, it is possible to com- on theory and the utilization of clinical knowledge could pare theoretical implications to phenomena observed in therefore be helpful in informing relationships in the es- clinical practice and to evaluate and adapt theory ac- timated model that cannot reasonably be captured by cordingly [25, 46, 47]. Theory formation can thus be commonly used ESM data. The right part of Fig. 1 illus- adapted by examining what a theory implies, and these trates how computational modeling can guide mental implications only become fully apparent once a theory is health research, resulting in data models that are formalized and data can be simulated. grounded in theory-based considerations. The resulting Second, computational models bring clinical theories data models can be compared against theory-implied closer to empirical research. For instance, prior to em- simulation results and guide further theory development pirically studying a patient’s systems, the researcher [25] as well as future research design planning. needs to determine variables to include into the analysis. This question is of great importance in network estima- Example of a computational model: functional analysis of tion, since parameters in partial correlation networks are patient with panic disorder heavily dependent on the set-up of variables. The choice In the remainder of this paper, we will introduce and of variables has a crucial impact on network estimation evaluate an example system based on functional analysis and inference, especially if clinically relevant variables (sometimes referred to as applied behavior analysis or Burger et al. BMC Medicine (2020) 18:99 Page 5 of 18 SORKC model [50]), a framework commonly used by cli- panic disorder [35]. While Robinaugh et al. focus on the nicians to formulate case conceptualizations in CBT. generic approach described above, we also include per- Functional analysis explains maladaptive behavior in sonal factors as components in the model, in accordance terms of classical and operant conditioning processes: a with the principles of functional analysis. As will be dis- discriminant stimulus (Sd) evokes specific emotional, cussed later on, patient-specific reinforcing factors can cognitive and behavioral responses in the patient (Re, Rc, be modeled through both, extending equations and and Rb, respectively). Persistent dysfunctional coping is altering parameters in the system. explained through the presence of reinforcing stimuli. In the short term, dysfunctional coping mostly yields posi- tive effects (perceived benefits), while on the long term, Methods negative effects (perceived costs) are accumulating. In the following, we describe the general approach to To illustrate, we are modeling the case formalizing idiographic theories, using the functional conceptualization of a hypothetical patient suffering analysis of our hypothetical patient. To facilitate from panic disorder. This example patient experiences readability, we focus on introducing the process on a unusual bodily sensations (arousal) in the cinema and conceptual basis. We advise the reader interested in concludes that she will have a heart attack and that there technical detail to consider the supplementary material is no chance she can get medical assistance on time. The (see Additional file 1: Mathematical Background). Note experience of heart racing in the cinema constitutes her that the simulation results and the discussion can be discriminant stimulus (Sd). Her emotional response is followed without having read the mathematical back- panic (Re), due to catastrophic interpretations of the ground section. heart racing (“I am having a heart-attack”; cognitive re- In many formal theories, including the one that will be sponse, Rc). In order to cope with the aversiveness of presented here, every component of the system is this situation, she leaves the cinema (Rb). This behavior expressed as a differential equation, precisely explicating yields benefits: The patient manages to decrease the in- the specific influences of system variables on one an- tense fear she felt in the cinema (perceived benefits). other. Intuitively, differential equations can be under- However, constant avoidance also leads to costs: The pa- stood as specifying the rate of change in a given variable tient withdraws herself socially and experiences prob- (i.e., how a given variable will change over time), as a lems at work due to her avoidant coping in panic- function of itself and other causally related variables. For evoking situations (perceived costs). Further, she is faced instance, in the simplest case of a first-order derivative, with a lack of falsification possibilities, increasing the the differential equation of the variable avoidance cap- credibility of her catastrophic thoughts in confrontation tures the extent to which avoidance behavior will in- with experiencing heart racing while not being able to crease or decrease moving forward from a given time get medical assistance. point. Since our system predicts that avoidance is Figure 2 shows a schematic summary of the main fac- employed as a consequence of anxiety, the correspond- tors involved in the patient’s functional analysis, as typic- ing differential equation would encode that high levels ally documented in psychotherapy. Robinaugh and of anxiety increase the first-order derivative (the mo- colleagues recently proposed a computational model for mentary change) of avoidance. Fig. 2 Functional analysis of hypothetical patient suffering from panic disorder. Case conceptualization of our example patient using the framework of functional analysis, as commonly documented in clinical practice. A discriminant stimulus leads to cognitive, emotional, and behavioral reactions (Rc, Re, Rb, respectively). The behavioral reaction has perceived benefits and costs, reinforcing or inhibiting the behavior Burger et al. BMC Medicine (2020) 18:99 Page 6 of 18 Note that in this paper, we primarily focus on model- instance social withdrawal or potential problems at ing linear differential equations. Extending the frame- work. These detriments (perceived costs) are theorized to work to including non-linear equations would be a have an inhibiting effect on the patient’s avoidance relevant step for future research, given that prior litera- behavior. Third, persistent application of avoidance be- ture found that psychotherapeutic processes are often havior comes with a lack of opportunities to falsify the chaotic, a feature that is characteristic for non-linear catastrophic interpretation. Therefore, we modeled in- dynamics [41, 51]. For the sake of implementation, how- creasing credibility of the catastrophic interpretation as a ever, we decided to focus on linear equations, since consequence of avoidant coping. The credibility of the many aspects of non-linear dynamics require an exten- catastrophic interpretation increases the patient’s sive mathematical understanding. We will discuss the tendency to catastrophize in confrontation with the difference between these approaches and the impact on discriminant stimulus. predictions in systems later on. Step 3: Formalizing interventions Procedure of formalizing idiographic theories One of the main advantages of computational modeling Step 1: Schematic representation in clinical practice is that interventions on a system can Prior to formulating differential equations, we recom- be examined in silico, and their effects evaluated on the mend visualizing the system schematically. This facili- basis of a case conceptualization. Note that the simu- tates specifying relations in the equations later on. A lated effects are dependent on the accuracy of the model, graphical depiction of the relations in the patient’s highlighting the importance of theory evaluation [25]. functional analysis, including the target nodes of the in- We will discuss future avenues for systematic evalua- terventions introduced below, is presented in Fig. 3. This tions later on. is a crucial step, since it opens the search horizon Similar to step 2, interventions need to be formalized. beyond the given boundaries of functional analysis (i.e., We modeled two commonly used interventions in CBT: allowing to incorporate person-specific elements into exposure therapy and cognitive reappraisal. First, we im- the system, such as competencies and resources), and re- plemented exposure through setting avoidant coping to quires the clinician to explicate relations between the 0. Second, cognitive reappraisal was implemented variables. through formalizing another system variable, capturing the credibility of an alternative functional interpretation Step 2: Deriving differential equations of heart racing. The credibility of the functional inter- Based on the schematic representation of the patient’s pretation was theorized to “compete” with the credibility functional analysis, we formulated differential equations of the catastrophic interpretation, and we thus formal- for each component in the system. Practical guidelines ized the former as an inverse function of the latter; if the for defining dynamical systems from both theory and functional interpretation of the stimulus increases, the data have been recently described elsewhere [52]. dysfunctional interpretation decreases and vice versa. As a starting point, we modeled catastrophic interpre- This change in interpretation of the stimulus influences tations (Rc) of the discriminant stimulus (Sd) as input the extent to which the patient catastrophizes. We there- for the occurrence of panic (Re); heart racing in the cin- fore extended the equation for catastrophizing with an ema leads to the catastrophic idea that this is a sign of inhibitive term; increasing the credibility of the func- an upcoming heart attack, and the patient consequently tional interpretation (e.g., “I simply had too much experiences panic symptoms. In turn, the patient copes coffee”) leads to less catastrophic interpretations of the through avoidance behavior. We modeled coping behav- discriminant stimulus. ior using equations commonly applied to model the dynamics between prey and predator populations in Step 4: Choosing initial values of system variables and ecology [53]. In our model, panic (Re) is analogous to parameters “prey” and avoidance (Rb) is analogous to “predator”. Prior to conducting simulations, initial values of each Thus, increases in panic give rise to increases in avoid- system variable and parameters need to be defined. In ance behavior, while increases in avoidance behavior contrast to many data-driven approaches of estimating lead to lower panic. networks, these values are difficult to interpret numeric- Avoidant coping is modulated through the presence of ally. This is because formalizing idiographic theories reinforcing/inhibiting factors. First, if the patient per- does not require the clinician to operationalize variables, ceives avoidance to be effective in decreasing panic (i.e., since these will not (necessarily) be measured. The units experiencing relief; perceived benefits), her tendency to of system variables are therefore not meaningful. We will cope through avoidance increases. Second, avoidance be- discuss advantages and disadvantages of aligning theory havior comes with detriments for the patient, for components with the measurement procedure later on. Burger et al. BMC Medicine (2020) 18:99 Page 7 of 18 Fig. 3 Schematic representation of the functional analysis. Theoretical relations adapted from the patient’s functional analysis as a basis for deriving the system equations. Anxiety (Re) is reduced through applying avoidance behavior (Rb). In addition, avoidance behavior is reinforced through perceived benefits and inhibited through perceived costs. Persistent avoidant behavior increases the credibility of catastrophic interpretations, in turn leading to more catastrophizing during exposure. We formalized and tested three interventions, exposure, cognitive reappraisal, and their combination, represented through the red boxes In contrast to common parameter estimation tech- For our example model, we chose parameters and ini- niques in data models, the approach outlined in this tial values of the variables according to a qualitative paper treats parameters in formalized theories as “tun- examination of the system behavior, i.e., through adjust- ing-knobs” to tailor the relations towards the patient’s ing parameters until the system resembled behavior to case until theory-implied behavior resembles phenomena be expected given the information on the case of our of interest. For instance, one can increase the parameter hypothetical patient. The choice of parameters and ini- encoding the extent to which avoidance behavior follows tial values can be found in the mathematical appendix panic, if it is known that the patient has a strong ten- (see Additional file 1: Mathematical Background), along- dency to employ avoidance behavior as coping. Further, side all differential equations used in the simulations. one can vary values of parameters to examine differential effects of unknown relations; for instance, clinician and Step 5: Simulating and visualizing theory-implied data patient can collaboratively examine the effects of differ- Following the system specification, we can simulate and ent parameter choices for catastrophizing leading to visualize data. We provide the code to reproduce our panic. This allows patients to experimentally examine analysis and plots in R (Additional file 2: Code to repro- the responses of their system towards alterations. duce analyses). System data is commonly visualized in Burger et al. BMC Medicine (2020) 18:99 Page 8 of 18 time-series plots and phase portraits. Time-series plots avoidant coping, accompanied by a decrease in catastro- indicate the time trajectories of all system variables, with phizing and credibility of the catastrophic interpretation, time on the x-axis and variable levels on the y-axis. demonstrating the effectiveness of behavioral therapy for Phase portraits are useful to display the relationship be- our patient. With the introduction of exposure therapy, tween two or three variables over time. Each variable is the perceived benefits of avoidance behavior disap- represented on an axis, and following the trajectory in peared, e.g., the patient could not experience relief the phase portrait gives us information regarding the through avoidance anymore, and the associated costs time course of the displayed variables. To illustrate, we decayed over time. used the example of three-dimensional phase portraits, indicating the relationship between panic, avoidant cop- Scenario 3: Cognitive therapy (cognitive reappraisal) ing,and the credibility of the catastrophic interpretation. Figure 6 a–c show the time-series plots and phase portrait when applying cognitive reappraisal. While Step 6: Evaluating case conceptualizations functional interpretations of the discriminant stimulus In a last step, the simulated (“theory-implied”) data can could help decreasing panic tendencies, avoidance be- be compared to phenomena observed in clinical practice. havior only decreased after the functional interpretation Differences between simulated data and observed gained sufficient credibility. Additionally, catastrophizing patterns can be an indication that specific system rela- and the credibility of the dysfunctional cognition de- tions need to be adapted or that important variables are creased, while avoidance behavior gave rise to both, the missing in the system [46, 47]. As illustrated in Fig. 1, perceived costs and benefits. these considerations can be important pointers for set- ting up empirical investigations of symptom dynamics Scenario 4: Cognitive behavioral therapy (exposure + (e.g., which variables to include in an ESM study). We cognitive reappraisal) will address formal aspects of theory evaluation in the Figure 7 a–c show the time-series plots and phase por- “Discussion” section. trait when applying exposure and cognitive reappraisal simultaneously. Similar to scenario 2, this combination Results of interventions led to an increase in panic tendencies in Scenario 1: System behavior without intervention the short term. The introduction of the functional inter- Figure 4 a–c show time-series plots and phase portraits pretation of the discriminant stimulus was accompanied for the simulated system behavior without intervention. by a decrease in catastrophic interpretation and its Being confronted with the discriminant stimulus led to a credibility, ultimately leading to a decrease in panic ten- rapid increase in catastrophizing, followed by panic. dencies. Similar to scenario 2, confrontation led the as- Over time, avoidance behavior gradually built up as a sociated benefits of the behavior to disappear and the coping mechanism. While this was associated with a costs to decay over time. momentary decrease in panic, persistent avoidance was also accompanied by increasing credibility of the cata- Discussion strophic interpretation, in turn leading the patient to Current movements in psychotherapy strongly align with catastrophize even more when confronted with the technical advances in dynamical modeling tools—yet discriminant stimulus. In the short term, avoidance be- their implementation in clinical practice is rather scarce. havior was mainly associated with benefits, while in the To bridge this gap, we call for a stronger focus on tools long term, the perceived costs built up. The three- that make use of frameworks and theories embedded in dimensional phase portrait shows that persistent avoid- clinical practice. In this paper, we discussed the ance behavior did not allow the patient to decrease panic formalization of idiographic theories, through the use of states in the long term. Instead, panic tendencies mani- differential equations, as an alternative to data-driven fested as a function of the credibility of the catastrophic network modeling approaches. Our main objective for interpretation. A clinical interpretation could be that the promoting the use of formalized idiographic theories is patient was not able to falsify catastrophic interpreta- that data models cannot always account for consider- tions due to the lack of exposure to the discriminant ations relevant to clinical practice. In consequence, even stimulus. though techniques seem to be promising in analyzing patient data, their implementation might be hampered Scenario 2: Behavioral therapy (exposure) due to the lack of options to incorporate theoretical and Figure 5 a–c show the time-series plots and phase practical considerations. This barrier can be addressed portrait when applying exposure. This intervention led through grounding dynamical systems in the theories of to a sudden increase in panic states in the short term. In practitioners. Differential equations are commonly used the long term, panic decayed even under absence of in a variety of other scientific fields to describe systems, Burger et al. BMC Medicine (2020) 18:99 Page 9 of 18 Fig. 4 (See legend on next page.) Burger et al. BMC Medicine (2020) 18:99 Page 10 of 18 (See figure on previous page.) Fig. 4 Simulation results of scenario 1 (no intervention). The top and middle parts show the simulated time-series for the discriminant stimulus, panic, and avoidant coping along with a catastrophizing and the credibility of the catastrophic interpretation and b perceived benefits and costs. The bottom part of the figure (c) shows the three-dimensional phase portrait for panic, avoidant coping, and the credibility of the catastrophic interpretation, where the white box indicates the start and the black box the end of the trajectory and are a promising avenue for formalizing theories of should be specified. For instance, patient A might have mental health. more exposure to their discriminant stimulus in their To illustrate this approach, we formulated a computa- everyday life compared to patient B, or patient C has tional model based on dynamics of the functional ana- stronger avoidance tendencies than patient D. These lysis for a patient suffering from panic disorder and considerations can be reflected in altering the parame- examined implications for the case conceptualization ters in the system, aligning this approach with the idea and the effects of commonly applied CBT interventions. of idiographic modeling. Further, specific components in The results of the simulations are largely congruent with the system can be added/removed, if applicable for a phenomena observed in clinical practice and in line with given individual. The framework of functional analysis is predictions of other theoretical frameworks. In the fol- transdiagnostic in nature and can be applied to a broad lowing, we discuss further benefits for clinical practice, range of disorders that involve dysfunctional coping, for concrete examples for theory adaptation, and future example, substance abuse, post-traumatic stress disorder, directions. obsessive-compulsive disorder, and depression. Benefits for clinical practice and clinical relevance Explanation We identify at least five benefits from formalizing case Functional analysis provides a framework that allows conceptualizations in respect to challenges faced in clin- explaining the function of maladaptive behavior and ical practice. helps understanding symptom maintenance. The explanatory character of these verbal theories can be ad- Scientific rigor vanced through formalization, since case conceptualiza- One of the main advances in mental health care over the tions can subsequently be evaluated in respect to how past decades is its increasing focus on scientific prac- well they can reproduce clinical phenomena [46, 47]. If a tices. The introduction of the scientist-practitioner case conceptualization fails to explain relevant phenom- model [54] was an attempt to strengthen scientific prac- ena, this will more easily be detected if data is simulated tices in psychotherapy, for instance through theory- from a formalized case conceptualization, compared to a guided hypothesis testing. It became vital for designing verbal theory. patient-tailored psychotherapy to formulate a testable theory regarding intervention effects. The case Prediction conceptualization is an example of a framework for such While computational modeling can foster the develop- scientific theories in clinical practice. However, if a the- ment of theoretical relations, it is also a useful tool for ory is vague, the resulting hypotheses, predictions, and predicting theory-implied system behavior under given tests become scientifically questionable [45]. Especially interventions. Most relevant for clinical practice, this al- in the current landscape of replicability issues [55, 56], lows the clinician to examine the effects of formalized we see value in enhancing theory development through clinical intervention in silico. Testing interventions in formalizing idiographic systems in clinical practice. As computational models offers efficient insight into inter- became evident in this report, especially when compar- vention effects without having to collect data. ing the initial verbal theory in Fig. 2 to the system of differential equations, the process of formalizing idio- Didactics graphic theories is mostly a process of increasing specifi- Simulation outcomes of a formalized idiographic theory city, in which clinicians need to thoroughly reflect on can be beneficial for didactics in clinical practice. First, and justify all relations between system variables. visualizing the simulation results allows the clinician to collaboratively examine symptom dynamics with the Idiography patient. This can be used in the process of psychoeduca- While the model used in this paper uses concepts that tion, and communicating a treatment rationale, espe- are relevant for a broader range of patients suffering cially for interventions that might be aversive for the from panic disorder (generic approach), there are many patient (e.g., exposure). Second, in the long term, we see individual differences in how exactly these relations potential in implementing formalized idiographic Burger et al. BMC Medicine (2020) 18:99 Page 11 of 18 Fig. 5 (See legend on next page.) Burger et al. BMC Medicine (2020) 18:99 Page 12 of 18 (See figure on previous page.) Fig. 5 Simulation results of scenario 2 (exposure; behavioral therapy). The top and middle part show the simulated time-series for the discriminant stimulus, panic, and avoidant coping along with catastrophizing and the credibility of the catastrophic interpretation (a) and perceived benefits and costs (b). The bottom part of the figure (c) shows the three-dimensional phase portrait for panic, avoidant coping, and the credibility of the catastrophic interpretation, where the white box indicates the start and the black box the end of the trajectory theories to enhance more concise communication be- variable as a tendency to experience panic in the pres- tween clinicians through more rigorous documentation ence of the discriminant stimulus, rather than the actual and visualization. experience of panic itself. Theory evaluation of the example model Future directions A main benefit to formalizing idiographic theories is that The approach of formalizing idiographic theories is still simulated data can directly be compared against fairly new to clinical psychology, and there is a lot of re- expected/reported behavior in the patient. One potential search that needs to be conducted to help implementing interpretation of discrepancies between simulated data it in clinical practice. In this section, we aim to give and clinical phenomena is that the case some directions for future research. conceptualization in its current form cannot account for potentially relevant clinical phenomena, for instance, if Systematic theory evaluation and testing important relations or variables are missing. If this is the A crucial barrier for implementation is that the explana- case, the clinician might want to adapt specific theoret- tions and predictions provided by a theory need to be as ical relations until the simulated data adequately repre- accurate as possible, especially if the aim is to test for- sents clinical phenomena. This is crucial when testing malized clinical interventions; such interventions will de- formalized interventions in a patient’s system. pend heavily on the accuracy of the model. We outlined In some aspects, the computational model presented that through comparisons of theory-implied and empir- in this paper is congruent with clinical phenomena, ical data, systems can be evaluated to increase accuracy. while in other aspects theory adaptation might be Notably, any systematic comparison between theory- needed. Note that the set-up of the simulation repre- implied and empirical data models would require that sents panic-symptomatology experienced by one hypo- variables used in data collection either directly map on thetical individual. Phenomena observed in simulations to components in the theory, or that they can be pre- might differ if parameters are altered, which allows cisely derived from those components. As outlined capturing individual differences in experiencing panic above, there are many elements in idiographic systems symptoms, and differences in treatment response. First, that are difficult to capture in common forms of data the simulations showed that for this patient, persistent collection (e.g., ESM data), suggesting direct mapping of avoidance behavior is accompanied by increasing theory components to variables in empirical data may be tendencies to catastrophize and increasing credibility of difficult. Accordingly, it will be necessary for researchers the catastrophic interpretation. This finding highlights to not only formalize theories, but also the auxiliary the role of falsification in fear disorders; avoidant coping hypotheses about measurement that link the theory is associated with a lack of opportunities to falsify the components to the variables in empirical data. In this catastrophic interpretation, subsequently leading to in- paper, we opted for modeling idiographic systems with- creasing tendencies to experience panic in confrontation out restrictions to what can be operationalized and com- with discriminant stimuli. Second, the simulations pared how well theory-implied data qualitatively indicate that all interventions (exposure, cognitive re- resembles clinical phenomena based on expert discus- appraisal, and combination) are effective in decreasing sions, but did not go through the process of formalizing panic tendencies for this patient, which is in line with our assumptions about measurement or deriving what empirical studies testing the efficacy of CBT interven- should be expected in any given empirical data model. tions for panic disorder [57]. Third, the simulation re- Second, it needs to be noted that the origin of a poten- sults showed that panic manifests in the long term, if no tial mismatch between theory-implied and empirical intervention is applied. This finding does not seem to data remains unknown. Such discrepancies can have a adequately represent the experience of panic attacks, multitude of sources and can be ascribed to either short- since these usually emerge rapidly and decline after a comings in the structure of the theory (e.g., missing cru- short amount of time. To account for this feature of cial variables in the theory, mis-specified or missing panic attacks, we propose to model stronger decay of relations between present elements of the theory), the panic. Alternatively, one could conceptualize this set-up of the simulation (e.g., exact initial conditions, Burger et al. BMC Medicine (2020) 18:99 Page 13 of 18 Fig. 6 (See legend on next page.) Burger et al. BMC Medicine (2020) 18:99 Page 14 of 18 (See figure on previous page.) Fig. 6 Simulation results of scenario 3 (cognitive reappraisal; cognitive therapy). The top and middle part show the simulated time-series for the discriminant stimulus, panic, and avoidant coping along with catastrophizing and the credibility of the catastrophic interpretation (a) and perceived benefits and costs (b). The bottom part of the figure (c) shows the three-dimensional phase portrait for panic, avoidant coping, and the credibility of the catastrophic interpretation, where the white box indicates the start and the black box the end of the trajectory valid parameter values, input and boundary conditions), from idiographic data models. Further, recent studies or shortcomings in empirical data collection and suggest that there is little incremental information in modeling (e.g., inappropriate modeling assumptions, time-series measures beyond mean levels and general measurement issues). Further, estimating parameters variability [60] and that time-series effects show largely from non-linear time-series data is often difficult and unacceptable reliability after partialling out redundancies undergoes strong limitations [58]. We call for future re- with mean and variability [61]. It is important to note search to investigate systematic ways of identifying the that these findings pertain to the utility of idiographic core of such discrepancies. data models. As discussed above, these data models face several challenges in the clinical context (e.g., insufficient Technical expertise and effort number of observations, time-scaling, measurement arti- Another barrier to implementation is that, in the current facts, modeling assumptions), offering a potential practice of formalizing idiographic theories, constructing explanation for the questionable performance of time- a series of differential equations to formalize a patient’s series measures. system can be immensely challenging and requires Formalized idiographic theories, on the other hand, technical expertise that is not part of psychotherapy aim to explain phenomena that can be observed in the trainings. To address this issue, we propose that meth- patient. They do so by representing the system posited odologists elaborate on a set of functions relevant to re- to give rise to the phenomenon. We outlined how for- lations between clinical variables that can readily be malizing such systems can foster theory development used by clinicians to formalize idiographic theories. To and therefore potentially help clinicians gaining insight enhance accessibility, this set of functions could be im- into the effects of (formalized) clinical interventions. plemented in an interactive tool to visualize variable in- Valid inferences from such intervention simulations re- teractions. Clinicians could then pick from this set and quire clinicians to thoroughly evaluate their theories, construct formalized systems without the need for and formalizing theories can help in doing so. We argue understanding the mathematical background in depth. that, if proof-of-principle studies can support the Further, implementation would greatly benefit from a hypothesis that formalizing idiographic theories improve procedure that allows clinicians to formalize idiographic treatment planning, this could greatly benefit clinical theories using graphical tools. Such tools could incorp- practice. However, to facilitate implementation, future orate a simple three-step procedure: In a first step, clin- research should conduct surveys with practitioners to ician and patient collaboratively specify variables and understand potential barriers of implementing formal- sketch relations between the variables. Second, they se- ized idiographic theories. lect the qualitative nature of these specified relationships from the aforementioned list. This step encompasses the Linear versus non-linear dynamics derivation of differential equations adapted to clinical We introduced two perspectives in constructing idio- practice. Third, simulations are conducted and patient graphic systems: First, a top-down approach in which and therapist can interpret and explore symptom dy- generic factors are modeled and subsequently personal- namics given the case conceptualization and the differ- ized through adapting parameters, and second, a ential effects of interventions. bottom-up approach in which personalized factors are modeled directly—extending the search horizon to in- Clinicians’ skepticism and utility corporate any factor that can be related to the patient’s Recent investigations suggest that clinicians are skeptical system. In the present paper, we formalized a case regarding the utility of idiographic assessment ap- conceptualization within the generic framework of func- proaches, specifically regarding ESM data collection and tional analysis, using (primarily) linear equations. It is modeling techniques [24, 59]. While these surveys sug- important to note that, especially when following the gest that clinicians find idiographic data models to be bottom-up approach of constructing idiographic systems generally intuitive and aligning well with their clinical for and with each patient, system dynamics should en- reasoning, it was also found that clinicians are not compass not only linear, but also non-linear dynamics. always convinced that they can learn something new Indeed, prior research examining the quality of system Burger et al. BMC Medicine (2020) 18:99 Page 15 of 18 Fig. 7 (See legend on next page.) Burger et al. BMC Medicine (2020) 18:99 Page 16 of 18 (See figure on previous page.) Fig. 7 Simulation results of scenario 4 (exposure and cognitive reappraisal; CBT). The top and middle parts show the simulated time-series for the discriminant stimulus, panic, and avoidant coping along with catastrophizing and the credibility of the catastrophic interpretation (a) and perceived benefits and costs (b). The bottom part of the figure (c) shows the three-dimensional phase portrait for panic, avoidant coping, and the credibility of the catastrophic interpretation, where the white box indicates the start and the black box the end of the trajectory dynamics found that processes in therapy are often non- Conclusion linear and chaotic [41, 62]. Such dynamics are, by defin- Complexity models are of great relevance for psycho- ition, hard to predict and are heavily dependent on the therapy. Case conceptualizations, even if only incorpor- specific set-up of the simulation; slight changes in the ating a small set of variables, can produce highly set-up of initial conditions and parameters might have complex behavior. We present the formalization of idio- dramatic effects on the simulated behavior. In such graphic theories through differential equations as an cases, it may only be possible to make broad predictions approach to align the movement of process-based psy- about expected behavior, for example, not when a panic chotherapy to dynamical system methodology. Simula- attack will occur, but rather whether a system is vulner- tion results based on formalized theories can account for able to such attacks. We encourage future research to considerations that are vital to clinical practice. Further- further investigate how such dynamics should precisely more, the process of formalizing a system promotes be incorporated in the formalization of theories. more scientific rigor in clinical practice and could help in improving explanatory and predictive precision of Incorporating social and contextual dynamics case conceptualizations, as well as treatment planning. Computational models, as the one presented in this paper, can account for processes that occur within an in- Supplementary information Supplementary information accompanies this paper at https://doi.org/10. dividual, and explain psychopathology on the basis of re- 1186/s12916-020-01558-1. inforcing factors. However, it seems unrealistic that these processes occur in isolation, independent from a Additional file 1: Mathematical background. This file includes the social context. Indeed, clinical reasoning often includes mathematical background, including differential equations for the system the influence of the social environment on certain psy- variables and interventions, as well as parameter choices and initial values to conduct the simulations. chological processes, for instance, the link between avoi- Additional file 2: Code to reproduce analyses. This file provides the dant coping tendencies and a certain attachment style, code to reproduce all analyses discussed in this report in the open- or the influence of peers in substance use. Incorporating source software R. All materials are made available in the open-science- interactions between different systems could open doors framework repository: https://osf.io/spb37/. to model these clinical phenomena. Future research could use methods from agent-based modeling to simu- Abbreviations Sd: Discriminant stimulus; Cat (Rc): Catastrophizing (cognitive reaction); Av late social interactions between patient-specific compu- (Rb): Avoidance (behavioral reaction); Pan (Re): Panic (emotional reaction); tational models and investigate how these interactions Ben: Perceived benefits of avoidance behavior; Cost: Perceived costs of can inform parameters or variables in the patient’s avoidance behavior; Cred: Credibility of dysfunctional interpretation; FunCog: Credibility of functional interpretation system. Acknowledgements Proof-of-principle We would like to thank the Institute for Advanced Study Amsterdam In order for new techniques to be considered relevant to (https://ias.uva.nl), which greatly supported the interdisciplinary exchange on this project. Further, we want to thank the Society for the Improvement of clinical practice, they should provide practitioners with a Psychological Science (SIPS), specifically Eiko Fried and Don Robinaugh for clear incentive, and a main incentive for psychotherapy organizing a workshop on the value of formalizing theories in psychology is to improve treatment outcomes. For many health care (materials can be found on OSF: https://osf.io/5czsn/), as well as Kimberly Quinn and Leonid Tiokhin for organizing a Hackathon on formalizing verbal systems, case conceptualizations form the starting point models (materials can be found on OSF: https://osf.io/6vx8b/). These for hypothesis-driven intervention planning and execu- workshops inspired great discussions that added to this report, and we hope tion. We expect that formalizing idiographic theories to see more of these in the future. can improve the precision of intervention predictions, Funding through enhancing explanatory and predictive precision The research was in part funded by the research talent grant no. 406.18.542, in formulating case conceptualizations; however, this awarded by the Netherlands Organization for Scientific Research (NWO), and idea needs empirical support. We hope that future re- further in part by a National Institute of Mental Health Career Development Award (1K23MH113805-01A1) awarded to D. Robinaugh. search will follow up on this hypothesis and provide us with proof-of-principle studies validating the utility of Availability of data and materials formal theories in enhancing predictive precision of case The code to reproduce all analyses conducted in this report, including conceptualizations. generated data, can be found in the OSF repository, https://osf.io/spb37/. Burger et al. BMC Medicine (2020) 18:99 Page 17 of 18 Authors’ contributions 16. Epskamp S, Waldorp LJ, Mõttus R, Borsboom D. The Gaussian Graphical JB performed simulation studies and was responsible for the main writing of Model in Cross-Sectional and Time-Series Data. Multivariate Behav Res. 2018; the manuscript, under close supervision of SE. JB, SE, RQ, and DJR were 53(4):453–80. involved in deriving the differential equations and the set-up of the simula- 17. Fisher AJ, Reeves JW, Lawyer G, Medaglia JD, Rubel JA. Exploring the tions. DCV, DJR, HR, and RAS were involved in discussing clinical aspects of idiographic dynamics of mood and anxiety via network analysis. J Abnorm the model and the applicability of the approach to clinical practice in gen- Psychol. 2017;126(8):1044–56. eral. All authors provided feedback on the manuscript and approved of its 18. Stone AA, Shiffman S. Ecological momentary assessment (Ema) in final version. behavioral medicine. Ann Behav Med. 1994;16(3):199–202. 19. Epskamp S, van Borkulo CD, van der Veen DC, Servaas MN, Isvoranu AM, Ethics approval and consent to participate Riese H, et al. Personalized network modeling in psychopathology: the Not applicable (this project did not involve human participants, human data, importance of contemporaneous and temporal connections. Clin Psychol human tissue or animals.) Sci. 2018;6:416–27. 20. Piccirillo ML, Beck ED, Rodebaugh TL. A clinician’s primer for idiographic research: considerations and recommendations. Behav Ther. 2019;50(5):938–51. Consent for publication 21. Wensing M, Grol R. Knowledge translation in health: how implementation Not applicable (this report does not contain any individual person’sdata in science could contribute more. BMC Med. 2019 ;17(1):88. Available from: any form.) https://doi.org/10.1186/s12916-019-1322-9. 22. Proctor EK, Landsverk J, Aarons G, Chambers D, Glisson C, Mittman B. Competing interests Implementation research in mental health services: an emerging science The authors declare that they have no competing interests. with conceptual, methodological, and training challenges. Adm Policy Ment Heal Ment Heal Serv Res. 2009;36(1):24–34. Author details 23. Fried EI, Nesse RM. The impact of individual depressive symptoms on University of Groningen, University Medical Center Groningen, University impairment of psychosocial functioning. PLoS One. 2014;9(2):e90311. Center Psychiatry (UCP) Interdisciplinary Center Psychopathology and 24. Frumkin M, Piccirillo M, Beck E, Grossman J, Rodebaugh T. Feasibility and Emotion Regulation (ICPE), Hanzeplein 1, 9713 GZ Groningen, The utility of idiographic models in the clinic: a pilot study. 2019. Netherlands. University of Amsterdam, Institute for Advanced Study, 25. Haslbeck J, Ryan O, Robinaugh D, Waldorp L, Borsboom D. Modeling Amsterdam, The Netherlands. Harvard University, Department of Psychiatry, psychopathology: from data models to formal theories. 2019. Massachusetts General Hospital, .Cambridge, MA, USA. 26. Bogen J, Woodward J. Saving the phenomena. Philos Rev. 1988;97(3):303–52. 27. Swoyer C. Structural representation and surrogative reasoning. Synthese. Received: 11 October 2019 Accepted: 16 March 2020 1991;87:449–508. 28. Furusawa C, Kaneko K. A dynamical-systems view of stem cell biology. Science. 2012;338(6104):215–17. References 29. Scheffer M, Hosper SH, Meijer ML, Moss B, Jeppesen E. Alternative equilibria 1. Cramer AOJ, Van Borkulo CD, Giltay EJ, Van Der Maas HLJ, Kendler KS, in shallow lakes. Trends Ecol Evol. 1993;8(8):275–79. Scheffer M, et al. Major depression as a complex dynamic system. PLoS 30. Coleman PT, Vallacher RR, Bartoli A, Nowak A, Bui-Wrzosinska L. Navigating One. 2016;11(12):e0167490. the landscape of conflict: applications of dynamical systems theory to 2. Borsboom D, Cramer AOJ, Kalis A. Brain disorders? Not really: why network addressing protracted conflict. In: The non-linearity of peace processes – structures block reductionism in psychopathology research. Behav Brain Sci. theory and practice of systemic conflict transformation; 2011. 2019;42:1–54. 31. Liebovitch LS, Peluso PR, Norman MD, Su J, Gottman JM. Mathematical model 3. Fried EI, van Borkulo CD, Cramer AOJ, Boschloo L, Schoevers RA, Borsboom of the dynamics of psychotherapy. Cogn Neurodyn. 2011;5(3):265–275. D. Mental disorders as networks of problems: a review of recent insights. 32. von Kentzinsky H, Wijtsma S, Treur J. A temporal-causal modelling approach Soc Psychiatry Psychiatr Epidemiol. 2017;52(1):1–10. to analyse the dynamics of burnout and the effects of sleep. 2019;. 4. Hofmann SG, Hayes SC. The future of intervention science: process-based 33. Dujmić Z, Machielse E, Treur J. A temporal-causal modeling approach to the therapy. Clin Psychol Sci. 2019;7(1):37–50. dynamics of a burnout and the role of physical exercise. In: Biologically 5. Borsboom D. A network theory of mental disorders. World Psychiatry. 2017; Inspired Cognitive Architectures Meeting; 2018. p. 88–100. 16(1):5–13. 34. Grasman J, Grasman RPPP, Van Der Maas HLJ. The dynamics of addiction: 6. Borsboom D, Cramer AOJ, Schmittmann VD, Epskamp S, Waldorp LJ. The craving versus self-control. PLoS One. 2016;11(6):e0158323. small world of psychopathology. PLoS One. 2011;6(11):e27407. 35. Robinaugh DJ, Haslbeck JMB, Waldorp, LJ, Kossakowski JJ, Fried EI, Millner 7. Borsboom D, Cramer AOJ. Network analysis: an integrative approach to the AJ, McNally RJ, van Nes EH, Scheffer M, Kendler KS BD. Advancing the structure of psychopathology. SSRN. 2013;9:91–121. network theory of mental disorders: a computational model of panic 8. Wichers M. The dynamic nature of depression: a new micro-level disorder. 2019. perspective of mental disorder that meets current challenges. Psychol Med. 36. Schiepek G. A dynamic systems approach to clinical case formulation. Eur J 2014;44(7):1349–60. Psychol Assess. 2003;19(3):175–84. 9. Epskamp S, Cramer AOJ, Waldorp LJ, Schmittmann VD, Borsboom D. 37. Schiepek G. A dynamic systems approach to clinical case formulation. Eur J qgraph: network visualizations of relationships in psychometric data. J Stat Psychol Assess. 2003;. Softw. 2012;48(4). 38. Schaub H, Schiepek G. Simulation of psychological processes: basic 10. Epskamp S, Fried EI. A tutorial on regularized partial correlation networks. issues and an illustration within the etiology of a depressive disorder. Psychol Methods. 2018;23(4):617–34. 11. Epskamp S, Borsboom D, Fried EI. Estimating psychological networks and 39. Strunk G, Schiepek G. Systemische Psychologie: eine Einführung in die their accuracy: a tutorial paper. Behav Res Methods. 2018;50:195–212. komplexen Grundlagen menschlichen Verhaltens. Heidelberg: Elsevier, 12. David S, Marshall A, Evanovich E, Mumma G. Intraindividual dynamic Spektrum Akad. Verlag; 2006. network analysis - implications for clinical assessment. J Psychopathol Behav 40. Sim K, Gwee KP, Bateman A. Case formulation in psychotherapy: revitalizing Assess. 2017;40:235–248. its usefulness as a clinical tool. Acad Psychiatry. 2005;29(3):289–92. 13. Dotterer H, Beltz A, Foster K, Simms L, Wright A. Personalized models of personality disorders: using a temporal network method to understand 41. Schiepek GK, Viol K, Aichhorn W, Hütt MT, Sungler K, Pincus D, et al. Psychotherapy is chaotic- (not only) in a computational world. Front symptomatology and daily functioning in a clinical sample. Psychol Med. 2019;1–9. Psychol. 2017;8:379. 14. Fisher A. Toward a dynamic model of psychological assessment: 42. Borsboom D, van der Maas H, Dalege J, Kievit R, Haig B. Theory construction implications for personalized care. J Consult Clin Psychol. 2015;83(4):825–36. methodology: a practical framework for theory formation in psychology. 15. Lutz W, Schwartz B, Hofmann S, Fisher A, Husen K, Rubel J. Using network 2020;. analysis for the prediction of treatment dropout in patients with mood and 43. van Rooij I, Baggio G. Theory before the test: how to build high- anxiety disorders: a methodological proof-of-concept study. Sci Rep. 2018;8. verisimilitude explanatory theories in psychological science. 2020;. Burger et al. BMC Medicine (2020) 18:99 Page 18 of 18 44. Guest O, Martin AE. How computational modeling can force theory building in psychological science. 2020;. 45. Fried E. Lack of theory building and testing impedes progress in the factor and network literature. 2020. 46. Societies A, Simulation S. Why model? J Artif Soc Soc Simul. 2008;11(4):12. 47. Smaldino PE. Models are stupid, and we need more of them. In: Computational Social Psychology; 2017. 48. Fried EI, Cramer AOJ. Moving forward: challenges and directions for psychopathological network theory and methodology. Perspect Psychol Sci. 2017;12(6):999–1020. 49. Haslbeck J, Ryan O. Recovering bistable systems from psychological time series. 2019;. 50. Pierce WD, Cheney CD, Pierce WD, Cheney CD. Applied behavior analysis. In: Behavior Analysis and Learning; 2018. 51. Schiepek G. Complexity and nonlinear dynamics in psychotherapy. Eur Rev. 2009;17(2):331–56. 52. Chow SM. Practical tools and guidelines for exploring and fitting linear and nonlinear dynamical systems models. Multivariate Behav Res. 2019;54(5): 690–718. 53. Wangersky PJ. Lotka-Volterra Population Models. Annu Rev Ecol Syst. 1978;9: 189–218. 54. Baker DB, Benjamin LT. The affirmation of the scientist-practitioner: a look back at Boulder. Am Psychol. 2000;55(2):241–7. 55. Pashler H, Wagenmakers EJ. Editors’ introduction to the special section on replicability in psychological science: a crisis of confidence? Perspect Psychol Sci. 2012;7(6):528–30. 56. Simmons JP, Nelson LD, Simonsohn U. False-positive psychology: undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychol Sci. 2011;22(11):1359–66. 57. Barlow DH. Cognitive-behavioral therapy for panic disorder: current status. J Clin Psychiatry. 1997;58(Suppl 2):32–36. 58. Gábor A, Banga J. Robust and efficient parameter estimation in dynamic models of biological systems. BMC Syst Biol. 2015;9:74. 59. Zimmermann J, Woods W, Ritter S, Happel M, Masuhr O, Jaeger U, et al. Integrating structure and dynamics in personality assessment: first steps toward the development and validation of a personality dynamics diary. Psychol Assess. 2019;31:516–31. 60. Dejonckheere E, Mestdagh M, Houben M, Rutten I, Sels L, Kuppens P, et al. Complex affect dynamics add limited information to the prediction of psychological well-being. Nat Hum Behav. 2019;3:1. 61. Wendt L, Wright A, Pilkonis P, Woods W, Denissen J, Kühnel A, et al. Indicators of affect dynamics: structure, test-retest reliability, and personality correlates. 2019. 62. Schiepek G, Fartacek C, Sturm J, Kralovec K, Fartacek R, Plöderl M. Nonlinear dynamics: theoretical perspectives and application to suicidology. Suicide Life Threat Behav. 2011;41(6):661–75. [cited 2019 May 28]. Available from: http://doi.wiley.com/10.1111/j.1943-278X.2011.00062.x. Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Medicine Springer Journals

Bridging the gap between complexity science and clinical practice by formalizing idiographic theories: a computational model of functional analysis

Loading next page...
 
/lp/springer-journals/bridging-the-gap-between-complexity-science-and-clinical-practice-by-gaBLHA7T7o

References (64)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2020
eISSN
1741-7015
DOI
10.1186/s12916-020-01558-1
Publisher site
See Article on Publisher Site

Abstract

Background: The past decades of research have seen an increase in statistical tools to explore the complex dynamics of mental health from patient data, yet the application of these tools in clinical practice remains uncommon. This is surprising, given that clinical reasoning, e.g., case conceptualizations, largely coincides with the dynamical system approach. We argue that the gap between statistical tools and clinical practice can partly be explained by the fact that current estimation techniques disregard theoretical and practical considerations relevant to psychotherapy. To address this issue, we propose that case conceptualizations should be formalized. We illustrate this approach by introducing a computational model of functional analysis, a framework commonly used by practitioners to formulate case conceptualizations and design patient-tailored treatment. Methods: We outline the general approach of formalizing idiographic theories, drawing on the example of a functional analysis for a patient suffering from panic disorder. We specified the system using a series of differential equations and simulated different scenarios; first, we simulated data without intervening in the system to examine the effects of avoidant coping on the development of panic symptomatic. Second, we formalized two interventions commonly used in cognitive behavioral therapy (CBT; exposure and cognitive reappraisal) and subsequently simulated their effects on the system. Results: The first simulation showed that the specified system could recover several aspects of the phenomenon (panic disorder), however, also showed some incongruency with the nature of panic attacks (e.g., rapid decreases were not observed). The second simulation study illustrated differential effects of CBT interventions for this patient. All tested interventions could decrease panic levels in the system. (Continued on next page) * Correspondence: j.burger@uva.nl University of Groningen, University Medical Center Groningen, University Center Psychiatry (UCP) Interdisciplinary Center Psychopathology and Emotion Regulation (ICPE), Hanzeplein 1, 9713 GZ Groningen, The Netherlands University of Amsterdam, Institute for Advanced Study, Amsterdam, The Netherlands Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Burger et al. BMC Medicine (2020) 18:99 Page 2 of 18 (Continued from previous page) Conclusions: Formalizing idiographic theories is promising in bridging the gap between complexity science and clinical practice and can help foster more rigorous scientific practices in psychotherapy, through enhancing theory development. More precise case conceptualizations could potentially improve intervention planning and treatment outcomes. We discuss applications in psychotherapy and future directions, amongst others barriers for systematic theory evaluation and extending the framework to incorporate interactions between individual systems, relevant for modeling social learning processes. With this report, we hope to stimulate future efforts in formalizing clinical frameworks. Keywords: Dynamical systems, Functional analysis, Computational modeling, Network analysis, Complex systems, Ordinary differential equations, Formalizing theories, Idiographic approach, Process-based psychotherapy, Theory development Background approach in mental health research, there has been spe- Complex system thinking is gaining increasing import- cific interest in implementing statistical tools to explore ance in understanding mental health [1–3]. In recent patient-specific symptom dynamics in clinical practice. It years, some clinicians have proposed a move away from is commonly assumed that successful implementation is the approach of treating mental illness as disorder cat- in part a question of providing technical trainings and egories towards a focus on processes and patient-specific accessible guidelines for clinicians [20]. However, merely mechanisms in psychotherapy [4]. These proposals call training clinicians in adopting tools provided by method- for a framework for thinking about mental illness in ologists does not guarantee that these tools also result in terms of systems, to understand the processes underlying models that map onto the language used by practi- psychopathology, and to apply this understanding to tioners. Indeed, an often-discussed barrier to implemen- patient-specific contexts. The network perspective to tation is the accurate translation of knowledge into the psychopathology [5–8], conceptualizing psychological relevant practice field [21]. That is, the language used to disorders as complex interactions of symptoms and re- discuss promising research findings and techniques does lated mental health factors, provides a framework to ad- not always match the targeted language of the dress this movement. Statistical procedures that allow practitioner. for the estimation of psychopathological networks have This issue applies to the estimation of personalized been developed [9–11] and applied across a wide range network models. At present, network estimation of mental disorders [12–15]. methods remain technical and do not account for poten- Furthermore, and arguably most relevant for psycho- tially relevant clinical considerations. For example, net- therapy, tools for idiographic network analysis have been work estimation methods identify “highly central” developed [16, 17], allowing us to explore patient- symptoms, given some assumptions, as promising targets specific symptom dynamics from data collected using of intervention [19], but these methods generally fail to the experience sampling method (ESM) [18]. This ap- account for the fact that symptoms differ in their amen- proach may be especially relevant for psychotherapy, as ability to psychological treatment or that some symp- it has the potential to be embedded within clinical prac- toms may have “low centrality” but remain critical tice through informing the formulation of idiographic targets for intervention because of their impact on psy- theories (i.e., case conceptualizations) and the identifica- chosocial functioning (e.g., suicidal thoughts and behav- tion of patient-tailored intervention targets [19]. Indeed, ior; [22, 23]). Further, currently available techniques to idiographic network analysis aligns well with the move- estimating personalized networks are primarily of ex- ment towards process-based psychotherapy [4]. It there- ploratory nature and do not allow clinicians to incorpor- fore seems surprising that, despite the availability of ate relevant a priori knowledge or clinical expertise. By supportive statistical tools and efforts to provide primers failing to see their ideas reflected in network models, for conducting idiographic research [20], the actual ap- practitioners might consider them as impractical and plication of personalized network modeling within psy- not in line with their clinical view, likely resulting in chotherapy is to date rare. hesitancy towards using personalized network models. Indeed, a recent study has shown that case conceptuali- zations greatly differ from temporal networks estimated From implementation barriers to a clinician’s wishlist from ESM data [24]. Implementation gaps between mental health research and clinical practice are a topic of enormous importance Based on these considerations, we argue that providing [21, 22]. With the emergence of the complex system trainings and guidelines is necessary, but not sufficient Burger et al. BMC Medicine (2020) 18:99 Page 3 of 18 in implementing the complex system approach in theories in psychiatry, including the relationship between clinical practice. For methods to be regarded clinically patient and therapist [31] and models of burnout [32, relevant, it is vital that tools have the flexibility to be 33], addiction [34], and panic disorder [35]. However, guided by clinical needs and allow practitioners to much remains unknown about precisely how such for- incorporate clinical considerations. mal theories should be developed and how they should be used in psychotherapy. The main objective of this Theories versus data models paper is to take a step towards addressing this gap in the In recent literature, special attention has been paid to literature by demonstrating the potential of formalizing disentangling conceptual aspects of data models and idiographic theories in clinical practice and illustrating theories. According to Haslbeck, Ryan, Robinaugh, an approach to formalizing such theories using the Waldorp, and Borsboom [25], data models (e.g., a mean, framework of functional analysis. correlation, or idiographic network model) are merely ways of representing or organizing data, often with the Approaches to constructing idiographic systems aim of establishing a phenomenon: a robust, We see two main ways of constructing personalized generalizable feature of the world identified through dynamical systems in psychopathology: First, modeling a empirical regularities [25, 26]. In contrast, the aim of a generic disorder model, and subsequently personalizing theory is to explain a phenomenon by representing those the model through estimating control parameters for the aspects of the real world that give rise to the equations in the system (top-down approach, cf. [35]), phenomenon. Whereas verbal theories are expressed in and second, modeling relations between specific variables language, formal theories are expressed in mathematical directly for and with each patient (bottom-up approach, equations or a computational programming language. cf. [36–39]). An advantage of the former approach is This level of specification allows formal theories to that it allows modeling individual differences between simulate theory-implied system behavior, and by observ- patients regarding the strength of shared relations (e.g., ing the effects of simulated interventions, we can draw person-specific tendencies to avoid when confronted conclusions about how the real-world system we are with fear), which consequently allows for examining for targeting would respond to a given treatment (a process instance tipping points in fear responses following referred to as “surrogative reasoning”, cf. [27]). maladaptive coping. An advantage of the latter approach In the following, we will refer to the approach of trans- is that it allows to flexibly model any psychological lating (verbal) case conceptualizations into mathematical hypotheses, as well as individual problems and resources systems as the formalization of idiographic theories. Al- [40]. though the term “theory” is commonly used to describe The method outlined in this paper is based on the phenomena on the nomothetic level, in this paper, we framework of functional analysis, and therefore utilizes are focused on the explaining phenomena at the level of elements of both approaches: On the one hand, func- the individual patient, and will use the term “idiographic tional analysis constitutes a generic framework for case theory” in respect to theorized relations within one formulation (top-down elements); on the other hand, it individual. also provides the flexibility to integrate patient-specific problems and resources (bottom-up elements). Formalizing idiographic theories To bridge the gap between methodological advances and The role of computational models in bridging the practical application of the complex system approach, scientist-practitioner gap we propose to derive dynamical system models directly We argue that formalizing idiographic theories provides from clinical theory, clinicians’ expertise and case- advantages for both, clinical practice and mental health specific knowledge. Formalizing patient systems tackles research, schematically displayed in Fig. 1, and is promis- the mismatch between technical tools and target ing in bridging the gap between the two. language as discussed above at its core; that is, rooting First, computational models of idiographic theories dynamical systems in the language of practitioners allows can be used to advance the current practice of a patient’s examining the patient’s system behavior based on clinic- case conceptualization. Sim, Gwee, and Bateman [40] ally relevant considerations. identified five key advantages associated with formulat- In other scientific disciplines like biology [28], ecology ing thorough case conceptualizations in clinical practice: [29], and political science [30], it is common to model (a) the integration/relation of multiple problems of a pa- dynamic processes based on theory and/or knowledge. tient, (b) the explanatory nature of the resulting model, Unfortunately, the application of formalized theories in (c) the prescription of interventions, (d) the prediction mental health research is to date extremely rare. of outcomes, and (e) the support for the therapeutic Recently, there have been efforts to propose formal relationship. Schiepek and colleagues [36, 37, 41] Burger et al. BMC Medicine (2020) 18:99 Page 4 of 18 Fig. 1 The role of computational modeling in bridging the scientist-practitioner gap. Schematic illustration of computational modeling (the product of formalizing a theory), at the intersection of clinical practice and mental health research. Computational models allow us to evaluate case conceptualizations in clinical practice (a–d), and bring clinical theories closer to empirical studies through guiding choices crucial to the estimation of and inferences drawn from data models (b, e–g) pioneered the integration of case formulation and are missing, or if included variables stem from theoretic- idiographic system modeling and argued that these key ally similar constructs, indicating topological overlap advantages could be strengthened through computa- [48]. Formalizing theories can provide useful informa- tional models. Clinicians are required to make more tion regarding the set-up of variables needed to retrieve rigorous decisions in specifying relations in the case clinical phenomena. Further, empirical research is often conceptualization, which makes the formalization of confronted with practical constraints to assessing idiographic theories a promising avenue to foster more psychological processes. Many clinically relevant psycho- scientific practices in designing patient-tailored treat- logical processes are difficult—sometimes even impos- ment. This reasoning is in line with a growing body of sible—to assess on their appropriate time-scale. For literature indicating the need for more rigorous theory practical reasons, variables are often measured within development in clinical and social sciences [25, 35, 42– the same time-scale (usually once a day or every few 45]. The left part of Fig. 1 illustrates how computational hours), potentially leading to biased estimates in dynam- modeling can inform case conceptualizations in clinical ical models. A recent simulation study suggests that practice: Formalizing a case conceptualization results in using the most commonly applied ESM time-intervals a computational model that allows the clinician to sub- results in data models that are largely unable to recover sequently simulate data, given the specified idiographic the micro dynamics of a system [49]. A stronger focus system. Based on these simulations, it is possible to com- on theory and the utilization of clinical knowledge could pare theoretical implications to phenomena observed in therefore be helpful in informing relationships in the es- clinical practice and to evaluate and adapt theory ac- timated model that cannot reasonably be captured by cordingly [25, 46, 47]. Theory formation can thus be commonly used ESM data. The right part of Fig. 1 illus- adapted by examining what a theory implies, and these trates how computational modeling can guide mental implications only become fully apparent once a theory is health research, resulting in data models that are formalized and data can be simulated. grounded in theory-based considerations. The resulting Second, computational models bring clinical theories data models can be compared against theory-implied closer to empirical research. For instance, prior to em- simulation results and guide further theory development pirically studying a patient’s systems, the researcher [25] as well as future research design planning. needs to determine variables to include into the analysis. This question is of great importance in network estima- Example of a computational model: functional analysis of tion, since parameters in partial correlation networks are patient with panic disorder heavily dependent on the set-up of variables. The choice In the remainder of this paper, we will introduce and of variables has a crucial impact on network estimation evaluate an example system based on functional analysis and inference, especially if clinically relevant variables (sometimes referred to as applied behavior analysis or Burger et al. BMC Medicine (2020) 18:99 Page 5 of 18 SORKC model [50]), a framework commonly used by cli- panic disorder [35]. While Robinaugh et al. focus on the nicians to formulate case conceptualizations in CBT. generic approach described above, we also include per- Functional analysis explains maladaptive behavior in sonal factors as components in the model, in accordance terms of classical and operant conditioning processes: a with the principles of functional analysis. As will be dis- discriminant stimulus (Sd) evokes specific emotional, cussed later on, patient-specific reinforcing factors can cognitive and behavioral responses in the patient (Re, Rc, be modeled through both, extending equations and and Rb, respectively). Persistent dysfunctional coping is altering parameters in the system. explained through the presence of reinforcing stimuli. In the short term, dysfunctional coping mostly yields posi- tive effects (perceived benefits), while on the long term, Methods negative effects (perceived costs) are accumulating. In the following, we describe the general approach to To illustrate, we are modeling the case formalizing idiographic theories, using the functional conceptualization of a hypothetical patient suffering analysis of our hypothetical patient. To facilitate from panic disorder. This example patient experiences readability, we focus on introducing the process on a unusual bodily sensations (arousal) in the cinema and conceptual basis. We advise the reader interested in concludes that she will have a heart attack and that there technical detail to consider the supplementary material is no chance she can get medical assistance on time. The (see Additional file 1: Mathematical Background). Note experience of heart racing in the cinema constitutes her that the simulation results and the discussion can be discriminant stimulus (Sd). Her emotional response is followed without having read the mathematical back- panic (Re), due to catastrophic interpretations of the ground section. heart racing (“I am having a heart-attack”; cognitive re- In many formal theories, including the one that will be sponse, Rc). In order to cope with the aversiveness of presented here, every component of the system is this situation, she leaves the cinema (Rb). This behavior expressed as a differential equation, precisely explicating yields benefits: The patient manages to decrease the in- the specific influences of system variables on one an- tense fear she felt in the cinema (perceived benefits). other. Intuitively, differential equations can be under- However, constant avoidance also leads to costs: The pa- stood as specifying the rate of change in a given variable tient withdraws herself socially and experiences prob- (i.e., how a given variable will change over time), as a lems at work due to her avoidant coping in panic- function of itself and other causally related variables. For evoking situations (perceived costs). Further, she is faced instance, in the simplest case of a first-order derivative, with a lack of falsification possibilities, increasing the the differential equation of the variable avoidance cap- credibility of her catastrophic thoughts in confrontation tures the extent to which avoidance behavior will in- with experiencing heart racing while not being able to crease or decrease moving forward from a given time get medical assistance. point. Since our system predicts that avoidance is Figure 2 shows a schematic summary of the main fac- employed as a consequence of anxiety, the correspond- tors involved in the patient’s functional analysis, as typic- ing differential equation would encode that high levels ally documented in psychotherapy. Robinaugh and of anxiety increase the first-order derivative (the mo- colleagues recently proposed a computational model for mentary change) of avoidance. Fig. 2 Functional analysis of hypothetical patient suffering from panic disorder. Case conceptualization of our example patient using the framework of functional analysis, as commonly documented in clinical practice. A discriminant stimulus leads to cognitive, emotional, and behavioral reactions (Rc, Re, Rb, respectively). The behavioral reaction has perceived benefits and costs, reinforcing or inhibiting the behavior Burger et al. BMC Medicine (2020) 18:99 Page 6 of 18 Note that in this paper, we primarily focus on model- instance social withdrawal or potential problems at ing linear differential equations. Extending the frame- work. These detriments (perceived costs) are theorized to work to including non-linear equations would be a have an inhibiting effect on the patient’s avoidance relevant step for future research, given that prior litera- behavior. Third, persistent application of avoidance be- ture found that psychotherapeutic processes are often havior comes with a lack of opportunities to falsify the chaotic, a feature that is characteristic for non-linear catastrophic interpretation. Therefore, we modeled in- dynamics [41, 51]. For the sake of implementation, how- creasing credibility of the catastrophic interpretation as a ever, we decided to focus on linear equations, since consequence of avoidant coping. The credibility of the many aspects of non-linear dynamics require an exten- catastrophic interpretation increases the patient’s sive mathematical understanding. We will discuss the tendency to catastrophize in confrontation with the difference between these approaches and the impact on discriminant stimulus. predictions in systems later on. Step 3: Formalizing interventions Procedure of formalizing idiographic theories One of the main advantages of computational modeling Step 1: Schematic representation in clinical practice is that interventions on a system can Prior to formulating differential equations, we recom- be examined in silico, and their effects evaluated on the mend visualizing the system schematically. This facili- basis of a case conceptualization. Note that the simu- tates specifying relations in the equations later on. A lated effects are dependent on the accuracy of the model, graphical depiction of the relations in the patient’s highlighting the importance of theory evaluation [25]. functional analysis, including the target nodes of the in- We will discuss future avenues for systematic evalua- terventions introduced below, is presented in Fig. 3. This tions later on. is a crucial step, since it opens the search horizon Similar to step 2, interventions need to be formalized. beyond the given boundaries of functional analysis (i.e., We modeled two commonly used interventions in CBT: allowing to incorporate person-specific elements into exposure therapy and cognitive reappraisal. First, we im- the system, such as competencies and resources), and re- plemented exposure through setting avoidant coping to quires the clinician to explicate relations between the 0. Second, cognitive reappraisal was implemented variables. through formalizing another system variable, capturing the credibility of an alternative functional interpretation Step 2: Deriving differential equations of heart racing. The credibility of the functional inter- Based on the schematic representation of the patient’s pretation was theorized to “compete” with the credibility functional analysis, we formulated differential equations of the catastrophic interpretation, and we thus formal- for each component in the system. Practical guidelines ized the former as an inverse function of the latter; if the for defining dynamical systems from both theory and functional interpretation of the stimulus increases, the data have been recently described elsewhere [52]. dysfunctional interpretation decreases and vice versa. As a starting point, we modeled catastrophic interpre- This change in interpretation of the stimulus influences tations (Rc) of the discriminant stimulus (Sd) as input the extent to which the patient catastrophizes. We there- for the occurrence of panic (Re); heart racing in the cin- fore extended the equation for catastrophizing with an ema leads to the catastrophic idea that this is a sign of inhibitive term; increasing the credibility of the func- an upcoming heart attack, and the patient consequently tional interpretation (e.g., “I simply had too much experiences panic symptoms. In turn, the patient copes coffee”) leads to less catastrophic interpretations of the through avoidance behavior. We modeled coping behav- discriminant stimulus. ior using equations commonly applied to model the dynamics between prey and predator populations in Step 4: Choosing initial values of system variables and ecology [53]. In our model, panic (Re) is analogous to parameters “prey” and avoidance (Rb) is analogous to “predator”. Prior to conducting simulations, initial values of each Thus, increases in panic give rise to increases in avoid- system variable and parameters need to be defined. In ance behavior, while increases in avoidance behavior contrast to many data-driven approaches of estimating lead to lower panic. networks, these values are difficult to interpret numeric- Avoidant coping is modulated through the presence of ally. This is because formalizing idiographic theories reinforcing/inhibiting factors. First, if the patient per- does not require the clinician to operationalize variables, ceives avoidance to be effective in decreasing panic (i.e., since these will not (necessarily) be measured. The units experiencing relief; perceived benefits), her tendency to of system variables are therefore not meaningful. We will cope through avoidance increases. Second, avoidance be- discuss advantages and disadvantages of aligning theory havior comes with detriments for the patient, for components with the measurement procedure later on. Burger et al. BMC Medicine (2020) 18:99 Page 7 of 18 Fig. 3 Schematic representation of the functional analysis. Theoretical relations adapted from the patient’s functional analysis as a basis for deriving the system equations. Anxiety (Re) is reduced through applying avoidance behavior (Rb). In addition, avoidance behavior is reinforced through perceived benefits and inhibited through perceived costs. Persistent avoidant behavior increases the credibility of catastrophic interpretations, in turn leading to more catastrophizing during exposure. We formalized and tested three interventions, exposure, cognitive reappraisal, and their combination, represented through the red boxes In contrast to common parameter estimation tech- For our example model, we chose parameters and ini- niques in data models, the approach outlined in this tial values of the variables according to a qualitative paper treats parameters in formalized theories as “tun- examination of the system behavior, i.e., through adjust- ing-knobs” to tailor the relations towards the patient’s ing parameters until the system resembled behavior to case until theory-implied behavior resembles phenomena be expected given the information on the case of our of interest. For instance, one can increase the parameter hypothetical patient. The choice of parameters and ini- encoding the extent to which avoidance behavior follows tial values can be found in the mathematical appendix panic, if it is known that the patient has a strong ten- (see Additional file 1: Mathematical Background), along- dency to employ avoidance behavior as coping. Further, side all differential equations used in the simulations. one can vary values of parameters to examine differential effects of unknown relations; for instance, clinician and Step 5: Simulating and visualizing theory-implied data patient can collaboratively examine the effects of differ- Following the system specification, we can simulate and ent parameter choices for catastrophizing leading to visualize data. We provide the code to reproduce our panic. This allows patients to experimentally examine analysis and plots in R (Additional file 2: Code to repro- the responses of their system towards alterations. duce analyses). System data is commonly visualized in Burger et al. BMC Medicine (2020) 18:99 Page 8 of 18 time-series plots and phase portraits. Time-series plots avoidant coping, accompanied by a decrease in catastro- indicate the time trajectories of all system variables, with phizing and credibility of the catastrophic interpretation, time on the x-axis and variable levels on the y-axis. demonstrating the effectiveness of behavioral therapy for Phase portraits are useful to display the relationship be- our patient. With the introduction of exposure therapy, tween two or three variables over time. Each variable is the perceived benefits of avoidance behavior disap- represented on an axis, and following the trajectory in peared, e.g., the patient could not experience relief the phase portrait gives us information regarding the through avoidance anymore, and the associated costs time course of the displayed variables. To illustrate, we decayed over time. used the example of three-dimensional phase portraits, indicating the relationship between panic, avoidant cop- Scenario 3: Cognitive therapy (cognitive reappraisal) ing,and the credibility of the catastrophic interpretation. Figure 6 a–c show the time-series plots and phase portrait when applying cognitive reappraisal. While Step 6: Evaluating case conceptualizations functional interpretations of the discriminant stimulus In a last step, the simulated (“theory-implied”) data can could help decreasing panic tendencies, avoidance be- be compared to phenomena observed in clinical practice. havior only decreased after the functional interpretation Differences between simulated data and observed gained sufficient credibility. Additionally, catastrophizing patterns can be an indication that specific system rela- and the credibility of the dysfunctional cognition de- tions need to be adapted or that important variables are creased, while avoidance behavior gave rise to both, the missing in the system [46, 47]. As illustrated in Fig. 1, perceived costs and benefits. these considerations can be important pointers for set- ting up empirical investigations of symptom dynamics Scenario 4: Cognitive behavioral therapy (exposure + (e.g., which variables to include in an ESM study). We cognitive reappraisal) will address formal aspects of theory evaluation in the Figure 7 a–c show the time-series plots and phase por- “Discussion” section. trait when applying exposure and cognitive reappraisal simultaneously. Similar to scenario 2, this combination Results of interventions led to an increase in panic tendencies in Scenario 1: System behavior without intervention the short term. The introduction of the functional inter- Figure 4 a–c show time-series plots and phase portraits pretation of the discriminant stimulus was accompanied for the simulated system behavior without intervention. by a decrease in catastrophic interpretation and its Being confronted with the discriminant stimulus led to a credibility, ultimately leading to a decrease in panic ten- rapid increase in catastrophizing, followed by panic. dencies. Similar to scenario 2, confrontation led the as- Over time, avoidance behavior gradually built up as a sociated benefits of the behavior to disappear and the coping mechanism. While this was associated with a costs to decay over time. momentary decrease in panic, persistent avoidance was also accompanied by increasing credibility of the cata- Discussion strophic interpretation, in turn leading the patient to Current movements in psychotherapy strongly align with catastrophize even more when confronted with the technical advances in dynamical modeling tools—yet discriminant stimulus. In the short term, avoidance be- their implementation in clinical practice is rather scarce. havior was mainly associated with benefits, while in the To bridge this gap, we call for a stronger focus on tools long term, the perceived costs built up. The three- that make use of frameworks and theories embedded in dimensional phase portrait shows that persistent avoid- clinical practice. In this paper, we discussed the ance behavior did not allow the patient to decrease panic formalization of idiographic theories, through the use of states in the long term. Instead, panic tendencies mani- differential equations, as an alternative to data-driven fested as a function of the credibility of the catastrophic network modeling approaches. Our main objective for interpretation. A clinical interpretation could be that the promoting the use of formalized idiographic theories is patient was not able to falsify catastrophic interpreta- that data models cannot always account for consider- tions due to the lack of exposure to the discriminant ations relevant to clinical practice. In consequence, even stimulus. though techniques seem to be promising in analyzing patient data, their implementation might be hampered Scenario 2: Behavioral therapy (exposure) due to the lack of options to incorporate theoretical and Figure 5 a–c show the time-series plots and phase practical considerations. This barrier can be addressed portrait when applying exposure. This intervention led through grounding dynamical systems in the theories of to a sudden increase in panic states in the short term. In practitioners. Differential equations are commonly used the long term, panic decayed even under absence of in a variety of other scientific fields to describe systems, Burger et al. BMC Medicine (2020) 18:99 Page 9 of 18 Fig. 4 (See legend on next page.) Burger et al. BMC Medicine (2020) 18:99 Page 10 of 18 (See figure on previous page.) Fig. 4 Simulation results of scenario 1 (no intervention). The top and middle parts show the simulated time-series for the discriminant stimulus, panic, and avoidant coping along with a catastrophizing and the credibility of the catastrophic interpretation and b perceived benefits and costs. The bottom part of the figure (c) shows the three-dimensional phase portrait for panic, avoidant coping, and the credibility of the catastrophic interpretation, where the white box indicates the start and the black box the end of the trajectory and are a promising avenue for formalizing theories of should be specified. For instance, patient A might have mental health. more exposure to their discriminant stimulus in their To illustrate this approach, we formulated a computa- everyday life compared to patient B, or patient C has tional model based on dynamics of the functional ana- stronger avoidance tendencies than patient D. These lysis for a patient suffering from panic disorder and considerations can be reflected in altering the parame- examined implications for the case conceptualization ters in the system, aligning this approach with the idea and the effects of commonly applied CBT interventions. of idiographic modeling. Further, specific components in The results of the simulations are largely congruent with the system can be added/removed, if applicable for a phenomena observed in clinical practice and in line with given individual. The framework of functional analysis is predictions of other theoretical frameworks. In the fol- transdiagnostic in nature and can be applied to a broad lowing, we discuss further benefits for clinical practice, range of disorders that involve dysfunctional coping, for concrete examples for theory adaptation, and future example, substance abuse, post-traumatic stress disorder, directions. obsessive-compulsive disorder, and depression. Benefits for clinical practice and clinical relevance Explanation We identify at least five benefits from formalizing case Functional analysis provides a framework that allows conceptualizations in respect to challenges faced in clin- explaining the function of maladaptive behavior and ical practice. helps understanding symptom maintenance. The explanatory character of these verbal theories can be ad- Scientific rigor vanced through formalization, since case conceptualiza- One of the main advances in mental health care over the tions can subsequently be evaluated in respect to how past decades is its increasing focus on scientific prac- well they can reproduce clinical phenomena [46, 47]. If a tices. The introduction of the scientist-practitioner case conceptualization fails to explain relevant phenom- model [54] was an attempt to strengthen scientific prac- ena, this will more easily be detected if data is simulated tices in psychotherapy, for instance through theory- from a formalized case conceptualization, compared to a guided hypothesis testing. It became vital for designing verbal theory. patient-tailored psychotherapy to formulate a testable theory regarding intervention effects. The case Prediction conceptualization is an example of a framework for such While computational modeling can foster the develop- scientific theories in clinical practice. However, if a the- ment of theoretical relations, it is also a useful tool for ory is vague, the resulting hypotheses, predictions, and predicting theory-implied system behavior under given tests become scientifically questionable [45]. Especially interventions. Most relevant for clinical practice, this al- in the current landscape of replicability issues [55, 56], lows the clinician to examine the effects of formalized we see value in enhancing theory development through clinical intervention in silico. Testing interventions in formalizing idiographic systems in clinical practice. As computational models offers efficient insight into inter- became evident in this report, especially when compar- vention effects without having to collect data. ing the initial verbal theory in Fig. 2 to the system of differential equations, the process of formalizing idio- Didactics graphic theories is mostly a process of increasing specifi- Simulation outcomes of a formalized idiographic theory city, in which clinicians need to thoroughly reflect on can be beneficial for didactics in clinical practice. First, and justify all relations between system variables. visualizing the simulation results allows the clinician to collaboratively examine symptom dynamics with the Idiography patient. This can be used in the process of psychoeduca- While the model used in this paper uses concepts that tion, and communicating a treatment rationale, espe- are relevant for a broader range of patients suffering cially for interventions that might be aversive for the from panic disorder (generic approach), there are many patient (e.g., exposure). Second, in the long term, we see individual differences in how exactly these relations potential in implementing formalized idiographic Burger et al. BMC Medicine (2020) 18:99 Page 11 of 18 Fig. 5 (See legend on next page.) Burger et al. BMC Medicine (2020) 18:99 Page 12 of 18 (See figure on previous page.) Fig. 5 Simulation results of scenario 2 (exposure; behavioral therapy). The top and middle part show the simulated time-series for the discriminant stimulus, panic, and avoidant coping along with catastrophizing and the credibility of the catastrophic interpretation (a) and perceived benefits and costs (b). The bottom part of the figure (c) shows the three-dimensional phase portrait for panic, avoidant coping, and the credibility of the catastrophic interpretation, where the white box indicates the start and the black box the end of the trajectory theories to enhance more concise communication be- variable as a tendency to experience panic in the pres- tween clinicians through more rigorous documentation ence of the discriminant stimulus, rather than the actual and visualization. experience of panic itself. Theory evaluation of the example model Future directions A main benefit to formalizing idiographic theories is that The approach of formalizing idiographic theories is still simulated data can directly be compared against fairly new to clinical psychology, and there is a lot of re- expected/reported behavior in the patient. One potential search that needs to be conducted to help implementing interpretation of discrepancies between simulated data it in clinical practice. In this section, we aim to give and clinical phenomena is that the case some directions for future research. conceptualization in its current form cannot account for potentially relevant clinical phenomena, for instance, if Systematic theory evaluation and testing important relations or variables are missing. If this is the A crucial barrier for implementation is that the explana- case, the clinician might want to adapt specific theoret- tions and predictions provided by a theory need to be as ical relations until the simulated data adequately repre- accurate as possible, especially if the aim is to test for- sents clinical phenomena. This is crucial when testing malized clinical interventions; such interventions will de- formalized interventions in a patient’s system. pend heavily on the accuracy of the model. We outlined In some aspects, the computational model presented that through comparisons of theory-implied and empir- in this paper is congruent with clinical phenomena, ical data, systems can be evaluated to increase accuracy. while in other aspects theory adaptation might be Notably, any systematic comparison between theory- needed. Note that the set-up of the simulation repre- implied and empirical data models would require that sents panic-symptomatology experienced by one hypo- variables used in data collection either directly map on thetical individual. Phenomena observed in simulations to components in the theory, or that they can be pre- might differ if parameters are altered, which allows cisely derived from those components. As outlined capturing individual differences in experiencing panic above, there are many elements in idiographic systems symptoms, and differences in treatment response. First, that are difficult to capture in common forms of data the simulations showed that for this patient, persistent collection (e.g., ESM data), suggesting direct mapping of avoidance behavior is accompanied by increasing theory components to variables in empirical data may be tendencies to catastrophize and increasing credibility of difficult. Accordingly, it will be necessary for researchers the catastrophic interpretation. This finding highlights to not only formalize theories, but also the auxiliary the role of falsification in fear disorders; avoidant coping hypotheses about measurement that link the theory is associated with a lack of opportunities to falsify the components to the variables in empirical data. In this catastrophic interpretation, subsequently leading to in- paper, we opted for modeling idiographic systems with- creasing tendencies to experience panic in confrontation out restrictions to what can be operationalized and com- with discriminant stimuli. Second, the simulations pared how well theory-implied data qualitatively indicate that all interventions (exposure, cognitive re- resembles clinical phenomena based on expert discus- appraisal, and combination) are effective in decreasing sions, but did not go through the process of formalizing panic tendencies for this patient, which is in line with our assumptions about measurement or deriving what empirical studies testing the efficacy of CBT interven- should be expected in any given empirical data model. tions for panic disorder [57]. Third, the simulation re- Second, it needs to be noted that the origin of a poten- sults showed that panic manifests in the long term, if no tial mismatch between theory-implied and empirical intervention is applied. This finding does not seem to data remains unknown. Such discrepancies can have a adequately represent the experience of panic attacks, multitude of sources and can be ascribed to either short- since these usually emerge rapidly and decline after a comings in the structure of the theory (e.g., missing cru- short amount of time. To account for this feature of cial variables in the theory, mis-specified or missing panic attacks, we propose to model stronger decay of relations between present elements of the theory), the panic. Alternatively, one could conceptualize this set-up of the simulation (e.g., exact initial conditions, Burger et al. BMC Medicine (2020) 18:99 Page 13 of 18 Fig. 6 (See legend on next page.) Burger et al. BMC Medicine (2020) 18:99 Page 14 of 18 (See figure on previous page.) Fig. 6 Simulation results of scenario 3 (cognitive reappraisal; cognitive therapy). The top and middle part show the simulated time-series for the discriminant stimulus, panic, and avoidant coping along with catastrophizing and the credibility of the catastrophic interpretation (a) and perceived benefits and costs (b). The bottom part of the figure (c) shows the three-dimensional phase portrait for panic, avoidant coping, and the credibility of the catastrophic interpretation, where the white box indicates the start and the black box the end of the trajectory valid parameter values, input and boundary conditions), from idiographic data models. Further, recent studies or shortcomings in empirical data collection and suggest that there is little incremental information in modeling (e.g., inappropriate modeling assumptions, time-series measures beyond mean levels and general measurement issues). Further, estimating parameters variability [60] and that time-series effects show largely from non-linear time-series data is often difficult and unacceptable reliability after partialling out redundancies undergoes strong limitations [58]. We call for future re- with mean and variability [61]. It is important to note search to investigate systematic ways of identifying the that these findings pertain to the utility of idiographic core of such discrepancies. data models. As discussed above, these data models face several challenges in the clinical context (e.g., insufficient Technical expertise and effort number of observations, time-scaling, measurement arti- Another barrier to implementation is that, in the current facts, modeling assumptions), offering a potential practice of formalizing idiographic theories, constructing explanation for the questionable performance of time- a series of differential equations to formalize a patient’s series measures. system can be immensely challenging and requires Formalized idiographic theories, on the other hand, technical expertise that is not part of psychotherapy aim to explain phenomena that can be observed in the trainings. To address this issue, we propose that meth- patient. They do so by representing the system posited odologists elaborate on a set of functions relevant to re- to give rise to the phenomenon. We outlined how for- lations between clinical variables that can readily be malizing such systems can foster theory development used by clinicians to formalize idiographic theories. To and therefore potentially help clinicians gaining insight enhance accessibility, this set of functions could be im- into the effects of (formalized) clinical interventions. plemented in an interactive tool to visualize variable in- Valid inferences from such intervention simulations re- teractions. Clinicians could then pick from this set and quire clinicians to thoroughly evaluate their theories, construct formalized systems without the need for and formalizing theories can help in doing so. We argue understanding the mathematical background in depth. that, if proof-of-principle studies can support the Further, implementation would greatly benefit from a hypothesis that formalizing idiographic theories improve procedure that allows clinicians to formalize idiographic treatment planning, this could greatly benefit clinical theories using graphical tools. Such tools could incorp- practice. However, to facilitate implementation, future orate a simple three-step procedure: In a first step, clin- research should conduct surveys with practitioners to ician and patient collaboratively specify variables and understand potential barriers of implementing formal- sketch relations between the variables. Second, they se- ized idiographic theories. lect the qualitative nature of these specified relationships from the aforementioned list. This step encompasses the Linear versus non-linear dynamics derivation of differential equations adapted to clinical We introduced two perspectives in constructing idio- practice. Third, simulations are conducted and patient graphic systems: First, a top-down approach in which and therapist can interpret and explore symptom dy- generic factors are modeled and subsequently personal- namics given the case conceptualization and the differ- ized through adapting parameters, and second, a ential effects of interventions. bottom-up approach in which personalized factors are modeled directly—extending the search horizon to in- Clinicians’ skepticism and utility corporate any factor that can be related to the patient’s Recent investigations suggest that clinicians are skeptical system. In the present paper, we formalized a case regarding the utility of idiographic assessment ap- conceptualization within the generic framework of func- proaches, specifically regarding ESM data collection and tional analysis, using (primarily) linear equations. It is modeling techniques [24, 59]. While these surveys sug- important to note that, especially when following the gest that clinicians find idiographic data models to be bottom-up approach of constructing idiographic systems generally intuitive and aligning well with their clinical for and with each patient, system dynamics should en- reasoning, it was also found that clinicians are not compass not only linear, but also non-linear dynamics. always convinced that they can learn something new Indeed, prior research examining the quality of system Burger et al. BMC Medicine (2020) 18:99 Page 15 of 18 Fig. 7 (See legend on next page.) Burger et al. BMC Medicine (2020) 18:99 Page 16 of 18 (See figure on previous page.) Fig. 7 Simulation results of scenario 4 (exposure and cognitive reappraisal; CBT). The top and middle parts show the simulated time-series for the discriminant stimulus, panic, and avoidant coping along with catastrophizing and the credibility of the catastrophic interpretation (a) and perceived benefits and costs (b). The bottom part of the figure (c) shows the three-dimensional phase portrait for panic, avoidant coping, and the credibility of the catastrophic interpretation, where the white box indicates the start and the black box the end of the trajectory dynamics found that processes in therapy are often non- Conclusion linear and chaotic [41, 62]. Such dynamics are, by defin- Complexity models are of great relevance for psycho- ition, hard to predict and are heavily dependent on the therapy. Case conceptualizations, even if only incorpor- specific set-up of the simulation; slight changes in the ating a small set of variables, can produce highly set-up of initial conditions and parameters might have complex behavior. We present the formalization of idio- dramatic effects on the simulated behavior. In such graphic theories through differential equations as an cases, it may only be possible to make broad predictions approach to align the movement of process-based psy- about expected behavior, for example, not when a panic chotherapy to dynamical system methodology. Simula- attack will occur, but rather whether a system is vulner- tion results based on formalized theories can account for able to such attacks. We encourage future research to considerations that are vital to clinical practice. Further- further investigate how such dynamics should precisely more, the process of formalizing a system promotes be incorporated in the formalization of theories. more scientific rigor in clinical practice and could help in improving explanatory and predictive precision of Incorporating social and contextual dynamics case conceptualizations, as well as treatment planning. Computational models, as the one presented in this paper, can account for processes that occur within an in- Supplementary information Supplementary information accompanies this paper at https://doi.org/10. dividual, and explain psychopathology on the basis of re- 1186/s12916-020-01558-1. inforcing factors. However, it seems unrealistic that these processes occur in isolation, independent from a Additional file 1: Mathematical background. This file includes the social context. Indeed, clinical reasoning often includes mathematical background, including differential equations for the system the influence of the social environment on certain psy- variables and interventions, as well as parameter choices and initial values to conduct the simulations. chological processes, for instance, the link between avoi- Additional file 2: Code to reproduce analyses. This file provides the dant coping tendencies and a certain attachment style, code to reproduce all analyses discussed in this report in the open- or the influence of peers in substance use. Incorporating source software R. All materials are made available in the open-science- interactions between different systems could open doors framework repository: https://osf.io/spb37/. to model these clinical phenomena. Future research could use methods from agent-based modeling to simu- Abbreviations Sd: Discriminant stimulus; Cat (Rc): Catastrophizing (cognitive reaction); Av late social interactions between patient-specific compu- (Rb): Avoidance (behavioral reaction); Pan (Re): Panic (emotional reaction); tational models and investigate how these interactions Ben: Perceived benefits of avoidance behavior; Cost: Perceived costs of can inform parameters or variables in the patient’s avoidance behavior; Cred: Credibility of dysfunctional interpretation; FunCog: Credibility of functional interpretation system. Acknowledgements Proof-of-principle We would like to thank the Institute for Advanced Study Amsterdam In order for new techniques to be considered relevant to (https://ias.uva.nl), which greatly supported the interdisciplinary exchange on this project. Further, we want to thank the Society for the Improvement of clinical practice, they should provide practitioners with a Psychological Science (SIPS), specifically Eiko Fried and Don Robinaugh for clear incentive, and a main incentive for psychotherapy organizing a workshop on the value of formalizing theories in psychology is to improve treatment outcomes. For many health care (materials can be found on OSF: https://osf.io/5czsn/), as well as Kimberly Quinn and Leonid Tiokhin for organizing a Hackathon on formalizing verbal systems, case conceptualizations form the starting point models (materials can be found on OSF: https://osf.io/6vx8b/). These for hypothesis-driven intervention planning and execu- workshops inspired great discussions that added to this report, and we hope tion. We expect that formalizing idiographic theories to see more of these in the future. can improve the precision of intervention predictions, Funding through enhancing explanatory and predictive precision The research was in part funded by the research talent grant no. 406.18.542, in formulating case conceptualizations; however, this awarded by the Netherlands Organization for Scientific Research (NWO), and idea needs empirical support. We hope that future re- further in part by a National Institute of Mental Health Career Development Award (1K23MH113805-01A1) awarded to D. Robinaugh. search will follow up on this hypothesis and provide us with proof-of-principle studies validating the utility of Availability of data and materials formal theories in enhancing predictive precision of case The code to reproduce all analyses conducted in this report, including conceptualizations. generated data, can be found in the OSF repository, https://osf.io/spb37/. Burger et al. BMC Medicine (2020) 18:99 Page 17 of 18 Authors’ contributions 16. Epskamp S, Waldorp LJ, Mõttus R, Borsboom D. The Gaussian Graphical JB performed simulation studies and was responsible for the main writing of Model in Cross-Sectional and Time-Series Data. Multivariate Behav Res. 2018; the manuscript, under close supervision of SE. JB, SE, RQ, and DJR were 53(4):453–80. involved in deriving the differential equations and the set-up of the simula- 17. Fisher AJ, Reeves JW, Lawyer G, Medaglia JD, Rubel JA. Exploring the tions. DCV, DJR, HR, and RAS were involved in discussing clinical aspects of idiographic dynamics of mood and anxiety via network analysis. J Abnorm the model and the applicability of the approach to clinical practice in gen- Psychol. 2017;126(8):1044–56. eral. All authors provided feedback on the manuscript and approved of its 18. Stone AA, Shiffman S. Ecological momentary assessment (Ema) in final version. behavioral medicine. Ann Behav Med. 1994;16(3):199–202. 19. Epskamp S, van Borkulo CD, van der Veen DC, Servaas MN, Isvoranu AM, Ethics approval and consent to participate Riese H, et al. Personalized network modeling in psychopathology: the Not applicable (this project did not involve human participants, human data, importance of contemporaneous and temporal connections. Clin Psychol human tissue or animals.) Sci. 2018;6:416–27. 20. Piccirillo ML, Beck ED, Rodebaugh TL. A clinician’s primer for idiographic research: considerations and recommendations. Behav Ther. 2019;50(5):938–51. Consent for publication 21. Wensing M, Grol R. Knowledge translation in health: how implementation Not applicable (this report does not contain any individual person’sdata in science could contribute more. BMC Med. 2019 ;17(1):88. Available from: any form.) https://doi.org/10.1186/s12916-019-1322-9. 22. Proctor EK, Landsverk J, Aarons G, Chambers D, Glisson C, Mittman B. Competing interests Implementation research in mental health services: an emerging science The authors declare that they have no competing interests. with conceptual, methodological, and training challenges. Adm Policy Ment Heal Ment Heal Serv Res. 2009;36(1):24–34. Author details 23. Fried EI, Nesse RM. The impact of individual depressive symptoms on University of Groningen, University Medical Center Groningen, University impairment of psychosocial functioning. PLoS One. 2014;9(2):e90311. Center Psychiatry (UCP) Interdisciplinary Center Psychopathology and 24. Frumkin M, Piccirillo M, Beck E, Grossman J, Rodebaugh T. Feasibility and Emotion Regulation (ICPE), Hanzeplein 1, 9713 GZ Groningen, The utility of idiographic models in the clinic: a pilot study. 2019. Netherlands. University of Amsterdam, Institute for Advanced Study, 25. Haslbeck J, Ryan O, Robinaugh D, Waldorp L, Borsboom D. Modeling Amsterdam, The Netherlands. Harvard University, Department of Psychiatry, psychopathology: from data models to formal theories. 2019. Massachusetts General Hospital, .Cambridge, MA, USA. 26. Bogen J, Woodward J. Saving the phenomena. Philos Rev. 1988;97(3):303–52. 27. Swoyer C. Structural representation and surrogative reasoning. Synthese. Received: 11 October 2019 Accepted: 16 March 2020 1991;87:449–508. 28. Furusawa C, Kaneko K. A dynamical-systems view of stem cell biology. Science. 2012;338(6104):215–17. References 29. Scheffer M, Hosper SH, Meijer ML, Moss B, Jeppesen E. Alternative equilibria 1. Cramer AOJ, Van Borkulo CD, Giltay EJ, Van Der Maas HLJ, Kendler KS, in shallow lakes. Trends Ecol Evol. 1993;8(8):275–79. Scheffer M, et al. Major depression as a complex dynamic system. PLoS 30. Coleman PT, Vallacher RR, Bartoli A, Nowak A, Bui-Wrzosinska L. Navigating One. 2016;11(12):e0167490. the landscape of conflict: applications of dynamical systems theory to 2. Borsboom D, Cramer AOJ, Kalis A. Brain disorders? Not really: why network addressing protracted conflict. In: The non-linearity of peace processes – structures block reductionism in psychopathology research. Behav Brain Sci. theory and practice of systemic conflict transformation; 2011. 2019;42:1–54. 31. Liebovitch LS, Peluso PR, Norman MD, Su J, Gottman JM. Mathematical model 3. Fried EI, van Borkulo CD, Cramer AOJ, Boschloo L, Schoevers RA, Borsboom of the dynamics of psychotherapy. Cogn Neurodyn. 2011;5(3):265–275. D. Mental disorders as networks of problems: a review of recent insights. 32. von Kentzinsky H, Wijtsma S, Treur J. A temporal-causal modelling approach Soc Psychiatry Psychiatr Epidemiol. 2017;52(1):1–10. to analyse the dynamics of burnout and the effects of sleep. 2019;. 4. Hofmann SG, Hayes SC. The future of intervention science: process-based 33. Dujmić Z, Machielse E, Treur J. A temporal-causal modeling approach to the therapy. Clin Psychol Sci. 2019;7(1):37–50. dynamics of a burnout and the role of physical exercise. In: Biologically 5. Borsboom D. A network theory of mental disorders. World Psychiatry. 2017; Inspired Cognitive Architectures Meeting; 2018. p. 88–100. 16(1):5–13. 34. Grasman J, Grasman RPPP, Van Der Maas HLJ. The dynamics of addiction: 6. Borsboom D, Cramer AOJ, Schmittmann VD, Epskamp S, Waldorp LJ. The craving versus self-control. PLoS One. 2016;11(6):e0158323. small world of psychopathology. PLoS One. 2011;6(11):e27407. 35. Robinaugh DJ, Haslbeck JMB, Waldorp, LJ, Kossakowski JJ, Fried EI, Millner 7. Borsboom D, Cramer AOJ. Network analysis: an integrative approach to the AJ, McNally RJ, van Nes EH, Scheffer M, Kendler KS BD. Advancing the structure of psychopathology. SSRN. 2013;9:91–121. network theory of mental disorders: a computational model of panic 8. Wichers M. The dynamic nature of depression: a new micro-level disorder. 2019. perspective of mental disorder that meets current challenges. Psychol Med. 36. Schiepek G. A dynamic systems approach to clinical case formulation. Eur J 2014;44(7):1349–60. Psychol Assess. 2003;19(3):175–84. 9. Epskamp S, Cramer AOJ, Waldorp LJ, Schmittmann VD, Borsboom D. 37. Schiepek G. A dynamic systems approach to clinical case formulation. Eur J qgraph: network visualizations of relationships in psychometric data. J Stat Psychol Assess. 2003;. Softw. 2012;48(4). 38. Schaub H, Schiepek G. Simulation of psychological processes: basic 10. Epskamp S, Fried EI. A tutorial on regularized partial correlation networks. issues and an illustration within the etiology of a depressive disorder. Psychol Methods. 2018;23(4):617–34. 11. Epskamp S, Borsboom D, Fried EI. Estimating psychological networks and 39. Strunk G, Schiepek G. Systemische Psychologie: eine Einführung in die their accuracy: a tutorial paper. Behav Res Methods. 2018;50:195–212. komplexen Grundlagen menschlichen Verhaltens. Heidelberg: Elsevier, 12. David S, Marshall A, Evanovich E, Mumma G. Intraindividual dynamic Spektrum Akad. Verlag; 2006. network analysis - implications for clinical assessment. J Psychopathol Behav 40. Sim K, Gwee KP, Bateman A. Case formulation in psychotherapy: revitalizing Assess. 2017;40:235–248. its usefulness as a clinical tool. Acad Psychiatry. 2005;29(3):289–92. 13. Dotterer H, Beltz A, Foster K, Simms L, Wright A. Personalized models of personality disorders: using a temporal network method to understand 41. Schiepek GK, Viol K, Aichhorn W, Hütt MT, Sungler K, Pincus D, et al. Psychotherapy is chaotic- (not only) in a computational world. Front symptomatology and daily functioning in a clinical sample. Psychol Med. 2019;1–9. Psychol. 2017;8:379. 14. Fisher A. Toward a dynamic model of psychological assessment: 42. Borsboom D, van der Maas H, Dalege J, Kievit R, Haig B. Theory construction implications for personalized care. J Consult Clin Psychol. 2015;83(4):825–36. methodology: a practical framework for theory formation in psychology. 15. Lutz W, Schwartz B, Hofmann S, Fisher A, Husen K, Rubel J. Using network 2020;. analysis for the prediction of treatment dropout in patients with mood and 43. van Rooij I, Baggio G. Theory before the test: how to build high- anxiety disorders: a methodological proof-of-concept study. Sci Rep. 2018;8. verisimilitude explanatory theories in psychological science. 2020;. Burger et al. BMC Medicine (2020) 18:99 Page 18 of 18 44. Guest O, Martin AE. How computational modeling can force theory building in psychological science. 2020;. 45. Fried E. Lack of theory building and testing impedes progress in the factor and network literature. 2020. 46. Societies A, Simulation S. Why model? J Artif Soc Soc Simul. 2008;11(4):12. 47. Smaldino PE. Models are stupid, and we need more of them. In: Computational Social Psychology; 2017. 48. Fried EI, Cramer AOJ. Moving forward: challenges and directions for psychopathological network theory and methodology. Perspect Psychol Sci. 2017;12(6):999–1020. 49. Haslbeck J, Ryan O. Recovering bistable systems from psychological time series. 2019;. 50. Pierce WD, Cheney CD, Pierce WD, Cheney CD. Applied behavior analysis. In: Behavior Analysis and Learning; 2018. 51. Schiepek G. Complexity and nonlinear dynamics in psychotherapy. Eur Rev. 2009;17(2):331–56. 52. Chow SM. Practical tools and guidelines for exploring and fitting linear and nonlinear dynamical systems models. Multivariate Behav Res. 2019;54(5): 690–718. 53. Wangersky PJ. Lotka-Volterra Population Models. Annu Rev Ecol Syst. 1978;9: 189–218. 54. Baker DB, Benjamin LT. The affirmation of the scientist-practitioner: a look back at Boulder. Am Psychol. 2000;55(2):241–7. 55. Pashler H, Wagenmakers EJ. Editors’ introduction to the special section on replicability in psychological science: a crisis of confidence? Perspect Psychol Sci. 2012;7(6):528–30. 56. Simmons JP, Nelson LD, Simonsohn U. False-positive psychology: undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychol Sci. 2011;22(11):1359–66. 57. Barlow DH. Cognitive-behavioral therapy for panic disorder: current status. J Clin Psychiatry. 1997;58(Suppl 2):32–36. 58. Gábor A, Banga J. Robust and efficient parameter estimation in dynamic models of biological systems. BMC Syst Biol. 2015;9:74. 59. Zimmermann J, Woods W, Ritter S, Happel M, Masuhr O, Jaeger U, et al. Integrating structure and dynamics in personality assessment: first steps toward the development and validation of a personality dynamics diary. Psychol Assess. 2019;31:516–31. 60. Dejonckheere E, Mestdagh M, Houben M, Rutten I, Sels L, Kuppens P, et al. Complex affect dynamics add limited information to the prediction of psychological well-being. Nat Hum Behav. 2019;3:1. 61. Wendt L, Wright A, Pilkonis P, Woods W, Denissen J, Kühnel A, et al. Indicators of affect dynamics: structure, test-retest reliability, and personality correlates. 2019. 62. Schiepek G, Fartacek C, Sturm J, Kralovec K, Fartacek R, Plöderl M. Nonlinear dynamics: theoretical perspectives and application to suicidology. Suicide Life Threat Behav. 2011;41(6):661–75. [cited 2019 May 28]. Available from: http://doi.wiley.com/10.1111/j.1943-278X.2011.00062.x. Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Journal

BMC MedicineSpringer Journals

Published: Apr 8, 2020

There are no references for this article.