Day-to-day variation in perceived causal networks: a feasibility study in two psychiatric settings

Day-to-day variation in perceived causal networks: a feasibility study in two psychiatric settings | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Day-to-day variation in perceived causal networks: a feasibility study in two psychiatric settings Julian Reichert, Tessa F. Blanken, Therese Anderbro, Ämma Tangen, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9504631/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 6 You are reading this latest preprint version Abstract Purpose The network approach conceptualizes psychopathology as systems of causally interacting problems of living, more or less unstable over time. Longitudinal Perceived Causal Networks (L-PECAN) is a self-report method designed to capture day-to-day variability in patients’ perceived causal relations between problems. We examined the feasibility of an L-PECAN procedure in two clinical settings and explored within-person variability in structures. Methods Patients in two psychiatric clinics first completed a training week using a generic item list. Based on this, personalized items were formulated and used for at least three additional weeks. Feasibility was evaluated via recruitment rates and adherence. Within-person variability was examined using saturation and drift analyses. Results Eight patients completed the protocol. As expected, daily assessments were completed quickly, and participants largely perceived their personalized nodes as being caused by the other nodes. Some causal relations were perceived across days, but overall network structures showed substantial day-to-day variability. Conclusions An L-PECAN procedure, with a training week and personalized nodes, appears feasible in higher-functioning psychiatric patients and offers a structured way to examine within-person variability in perceived causal problem networks. Such variability may be clinically informative when investigating cyclical (e.g. menstrual) or context-dependent patterns. Further research is needed to evaluate validity and clinical utility. case conceptualization psychotherapy idiographic networks within-person dynamics perceived causal relations symptom fluctuations Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Background Diagnostic categories in psychiatry are broad constructs that encompass substantial heterogeneity across individuals (Allsopp et al., 2019). Precision psychiatry aims to identify clinically meaningful differences between patients and to use this information to tailor treatment to the specific problems and needs of the individual, with the goal of improving treatment outcomes (Fernandes et al., 2017). According to the network approach, psychopathology can be understood as a stable state in a network of causally entangled variables of behavioral, emotional, somatic and contextual variables (Borsboom, 2017 ). The network approach is particularly compatible with the goals of precision psychiatry. Rather than assuming that symptoms reflect a single latent disorder that is similar across individuals, the network approach conceptualizes psychopathology as a system of directly interacting symptoms and other factors that may differ substantially between individuals (Robinaugh et al., 2020). It is assumed that the relevant variables, i.e. nodes in the network, should cover symptoms irrespective of current diagnostic categories, and that these networks are to some extent person-specific (Vogel et al, 2025 ). Following this conceptualisation, the network of a particular person can be seen as shifting between different distinct states, between healthy and pathological. Although originating from different intellectual backgrounds, the network approach is similar to case conceptualizations in psychotherapy. In case conceptualizations, therapists and patients try to figure out how the context, behaviors and feelings of the patient are related to one another, in order to change central processes to maximize downstream effects (Kuyken, Padesky, & Dudley; 2008 ). In other words, the therapist and patient work together to capture both the day-to-day variability and which factors are consistently causing other problems. Such analysis is an important clinical skill for therapists, but typically relies on somewhat unsystematic data collection and qualitative analysis (Haynes et al, 2018 ). The network approach might contribute with better methods for both data collection and analysis, i.e. for personalizing treatment (Hofmann, 2025 ; Lavefjord et al, 2026 ). The network approach is primarily a conceptual model of psychopathology, but it has also given rise to a growing set of empirical methods for assessing, estimating, and visualizing individual networks (Scheffer et al, 2024 ). Networks are visualized as “nodes” (variables) connected by “edges” that reflect statistical or perceived relations between variables (Borsboom, 2017 ). Edges can be undirected (“these nodes tend to co-occur”), or directed (“variability in this node predicts or is perceived to cause variability in this other node”). A concept that follows from this is idea of differing importance or influence of nodes within a network: i.e. “centrality”. Different centrality metrics are used to summarize how much a given node is connected to other nodes, for example by quantifying how many and how strong connections it has to other parts of the network (Bringmann et al, 2022 ). Such central nodes are often interpreted as potentially influential points for intervention, under the assumption that changes in central nodes may have downstream effects on other problems (Vogel et al, 2025 ). Note however that this has also been critiqued (e.g. Bringmann et al, 2019 ). The network approach does not rely on a specific method for data collection, or a particular analysis. Rather, the approach is a way of thinking about psychopathology as a system of interacting problems, where the structure of causal relations may vary across persons and over time. From a clinical perspective, this offers a formalized language for ideas that are already familiar in psychotherapy, and opening up possibilities to investigate them in a more rigorous manner. Person-specific networks are typically estimated using dense repeated measures over several weeks (Ecological momentary assessment; EMA), which can be analysed using e.g. VAR (Vector Autoregression; for an overview see Bringmann et al, 2022 ). In a network estimated from such data, the directed edges between nodes in the network represent how much variation in one node predicts variation in the other nodes at the next assessment point (typically 2–3 hours later). This method has been used widely to map between-person heterogeneity and inform intervention foci, for a wide range of patient populations, e.g. depression (Siepe et al, 2024 ), eating disorders (Levinson et al, 2021 ) and post-traumatic stress (Bridges-Curry, 2025 ). Some studies use idiosyncratic nodes (for an overview, see Andreoli et al, 2025 ). However, time-series analyses comes with risk of missing relations because processes operate on multiple timescales within the same network, which makes alignment with sampling densities difficult (Bringmann et al, 2022 ). Further, such time-series analyses are necessarily very data-hungry, limiting their clinical feasibility (Mansueto et al, 2023 : Hall et al, 2025 ). Another technique, Perceived Causal Networks (PECAN; Klintwall et al, 2023 ; Vogel et al, 2025 ), creates a person specific network by simply asking the patient which causal relations he or she perceives between the nodes. This has the benefit of being able to capture processes of varying speed, but is obviously reliant on patient self insight and memory. The PECAN has been used in clinical settings (Andreasson et al, 2023 ), and in both self-guided questionnaire formats (Bångstad et al, 2022 ) and as a structured interview (Kaariniemi et al, 2025 ; Klintwall et al, in preparation; Reichert et al, 2025). Both these methods assume that the network structure of a particular patient is stable across time, or at least across the time period of investigation. Indeed, there are empirical papers that find that at least for a majority of patients, this is true (e.g., van der Tuin et al, 2023 ). Other studies by contrast find that networks change quite quickly (e.g., Cusack et al, 2025 ; Hulsman et al, 2024; Siepe et al, 2024 ). Some of this variation might plausibly be considered “noise,” i.e., random daily events affecting mood and behaviors, but e.g. for women some variation is likely due to hormonal cycles (see for example Tauseef et al, 2024 ) or important contextual factors such as weekdays vs weekends. A method that bridges EMA and PECAN is Longitudinal PECAN (L-PECAN). In this method, the patient is asked each day which symptoms were experienced that day, and how these symptoms were perceived to causally influence one another. This approach avoids both the timescale misalignment of EMA and decreases the recall bias of PECAN. Daily assessments can be aggregated into an averaged network to highlight the most frequently perceived causal relations. Crucially, this method moreover allows an analysis of how the perceived causal links fluctuate from day to day. This offers insight into both the variability and the stability of node interactions. By examining these two aspects together, we can identify nodes that are not only consistently central across time but also understand when and how their causal influence changes. This information may be useful for tailoring interventions to individual patients (Rubel et al, 2018 ; Hall et al, 2025 ). In the study by Burger et al. ( 2024 ), participants recruited through social media completed four weeks of daily assessments in which they selected the problems they had experienced that day from a predefined list (covering common symptoms, behaviors, and emotions) and rated how these symptoms influenced one another (i.e., both their causes and effects). For each participant, a core set of nodes was defined as those reported on at least one third of the days, and an aggregated network was created to identify consistent causal links within this core. For each participant, the rate of causal saturation was computed, defined as the number of assessed days needed to aggregate a stable network (i.e., when collecting more assessment days no longer changed the aggregated network). Distance plots were also visualized for each participant, showing whether days tended to be more alike the closer they were to each other in time. While this study demonstrated that individuals can indeed provide daily reports on perceived causal relations over extended periods, there were several limitations noted. The self-selected, non-clinical sample limited generalizability to clinical settings. The predefined node-list was experienced as limiting by respondents, and excluded contextual factors altogether (e.g., financial problems ; an issue pointed out previously, e.g. von Klipstein et al, 2025 ). Interestingly, only about half the sample reached causal saturation , that is, consistent causal links over days, raising the question whether a stable network is to be expected, and whether L-PECAN could be used to investigate this within-person variability further. The aim of the present study was to build on the previous L-PECAN study by improving on the method (e.g., individualized items), and evaluate the clinical feasibility with actual psychiatric patients of differing severity. Importantly, we also aim to make the within-person variability the focus, in order to investigate whether L-PECAN can be used to analyze such day-to-day fluctuations. We report how we determined our sample size, all data exclusions, all manipulations, and all measures in the study. Methods Design and Setting We use a within-subject observational design in which patients provide repeated daily assessment of perceived causal relations between personalized nodes over a three-week period. Each patient serves as their own unit of analysis, allowing both analysis of day-to-day variability and aggregation of person-specific networks. Data were collected in two clinical settings aimed at adult patients, but differing in symptom severity. The first setting is a university psychotherapy training clinic, i.e., outside of the public health care system, offering cognitive-behavioral therapy for adults with mild to moderate psychological problems such as anxiety, depression, or stress. The second setting is a specialist psychiatric outpatient clinic, part of routine secondary care, treating patients with severe and often comorbid conditions, including affective and neurodevelopmental disorders. Due to differing organizational requirements, L-PECAN was implemented in Qualtrics for the university clinic setting, and in RedCAP for the specialist psychiatry setting. Data collections has correspondingly two separate ethical approvals from the national Ethical Review Authority (reg. nrs. 2022–03141 and 2024–04165 respectively). To ensure patients are impossible to identify, we report no data on gender or age. The university clinic part of the study was preregistered at https://aspredicted.org/CY9_7WN . All data and code can be found at: https://osf.io/hma3e/overview?view_only=2fa0f1f341434be98dca8774bc930950 . No funding was received for the conduct of this study or the preparation of this manuscript. Procedure Recruitment Recruitment was done during a phase of 6 months at the two participating clinics. During this time-period, all patients filling inclusion criteria were approached with information about the study, which stressed that participation was both voluntary and that findings could be used to personalize their subsequent care. Interested patients were sent a brief psychoeducational video introducing the purpose of L-PECAN, explaining the purpose of the method and how to interpret the causal question in the daily assessments. The video illustrated concrete examples to ensure the patients understood the purpose and setup of the method before they consented to participation, thus hopefully reducing the dropouts observed in the Burger et al ( 2024 ) study. Patients still interested were given written study information and give their verbal and written consent. Training Week Patients completed one week of daily assessments using a generic list of 18 common problems (e.g., overthinking, sleep problems, financial issues, see appendix A). This predefined list was an updated version of the list used in Burger et al ( 2024 ), expanded to include contextual factors. Each evening, patients received a text message at a timepoint decided beforehand (between 18:00 and 20:30). The text message contained a link to the L-PECAN survey. Participants could complete the assessment the following morning, but they were instructed that answering at the end of the day in question was to be preferred (the proportion of responses that were in fact made the next morning is seen in Table 2 ). They reported which items they had experienced that day, and described the selected items in their own words (e.g., “You selected overthinking , what did you think about?”). Importantly, respondents picked which of the other items (whether experienced on that day or not) had contributed to the experienced problem (i.e., ingoing edges). Patients were also allowed to write causes other than the other items. Collaborative Item Individualization At the end of the training week, each patient met with a clinician (authors K.L. or G.A. in the university clinic, or Ä.T. or R.F. in the specialist clinic) to review their week-one data. The clinician had prepared a summary of the freetext answers (item specifications and causes) and quantitative data (i.e., item frequencies and out-degree centralities), and prepared a suggestion for a shortened item-list with not more than ten items. This was collaboratively edited so that each item was relevant and made sense to the patient. In practice, this procedure worked by both combining and splitting items to find a good level of abstraction, and add idiosyncratic information provided by the patient in free-text boxes. We were particularly careful to avoid content overlap by formulating items on a more abstract level (e.g., combining “pain” and “tired” into one item if these were hard to distinguish). Items were prioritized if they were perceived to be causing other items, or were experienced as important by the patient. This finalized list was used to create a personalized for the L-PECAN for the remainder of the study for each patient. Finally, the patient could choose to change the timepoint each evening the text-message with a link to the survey was sent out. L-PECAN phase Patients in the university clinic completed at least three weeks of daily assessments using the L-PECAN. Patients from the specialist clinic could respond for a longer period, depending on their planned sessions. In each assessment, the patient was first shown the individualized list of items and asked to select those that were experienced that day (note that for item “insomnia”, the respondent was asked to answer about the previous night; meaning that for responses delayed to the next morning, the patient was answering about a night two nights distant). Next, contributing causes for each selected item was asked about (e.g., “Today, what contributed to ?”) with the other individualized items as potential contributing causes. The patient could select as many contributing causes as he or she wished, including none. An option “other causes” was also available. As compared to the Burger et al ( 2024 ) version, the present study asked only about the causes (i.e., edges going into the node), not their effects (i.e., edges going out from the node). Another difference was that items were available as causes even though that item had not been reported as experienced that day (i.e. all items were always available as causes). The latter was changed to allow causal relations for time-scales longer than that day (e.g., being tired due to drinking alcohol the previous day). For each item, pictograms were created and used throughout to reduce cognitive load and make the repeated assessments less tedious. Note that responses were binary: an item was either selected as a cause or not (i.e., there was no grading of the causal strength or certainty of the perception). For the university sample, some extra questions were asked in the L-PECAN, described in appendix B. Other measures Depressive symptoms were assessed with the Patient Health Questionnaire–9 (PHQ-9; Kroenke, Spitzer, & Williams, 2001 ). The PHQ-9 consists of nine items corresponding to the diagnostic criteria for major depressive disorder. Each item is rated on a 4-point scale from 0 ( not at all ) to 3 ( nearly every day ), yielding a total score between 0 and 27. Higher scores indicate greater depressive symptom severity. Anxiety symptoms were assessed with the Generalized Anxiety Disorder–7 (GAD-7; Spitzer, Kroenke, Williams, & Löwe, 2006 ). The GAD-7 contains seven items measuring the frequency of core anxiety symptoms during the past two weeks, rated from 0 ( not at all ) to 3 ( nearly every day ). Total scores range from 0 to 21, with higher scores reflecting higher levels of anxiety. Participants University Psychotherapy Clinic Patients self-referred to the university training clinic by completing an online intake form, which included the PHQ-9 (Kroenke et al., 2001 ) and the GAD-7 (Spitzer et al., 2006 ) for symptom screening. The university clinic routinely excludes individuals with psychosis, acute suicidality, substance dependence, or other conditions judged unsuitable for student-provided therapy. All patients are required to have sufficient proficiency in for psychotherapy. Our pre-registered criteria (requiring patients to have PHQ9 and GAD7 scores both over 5) were not applied, as inclusion was slower than we had expected. Nine patients were approached with information about the study. Two declined participation after having been given written information about the study. Another two declined participation after having been given verbal information about the study and seen the inclusion video. Thus, five out of nine approached patients were included in this setting. Descriptives in Table 1 . Specialist Psychiatric Clinic Patients were recruited by clinicians during the early stages of their psychiatric evaluation before intervention initiation. Twelve patients were approached with study information. Four declined participation after receiving detailed information, primarily because the daily ratings were perceived as demanding. Two were excluded due to unclear technical difficulties that prevented completion of any of the daily assessments (i.e., could not get the link to work). Thus, six patients initially consented participation in the study. Of these, three dropped out during data-collection for unclear reasons. Thus, three out of twelve patients were included from this setting. Descriptives in Table 1 . Table 1 Descriptives of included patients Patient PHQ-9 GAD-7 Given reason for seeking therapy University clinic Patient 101 7 5 Anxiety, relationship issues Patient 102 10 18 Anxiety, depression Patient 103 5 4 Performance anxiety, social anxiety Patient 104 1 6 Obsessive thoughts and compulsive behaviors Patient 105 16 11 Anxiety, depression Specialist psychiatric clinic Patient 106 17 15 Depression, relationship issues Patient 107 13 13 Anxiety, anger Patient 108 missing Trouble focusing, restlessness Data Analysis Feasibility and adherence metrics For each participant, we calculated the percentage of completed daily assessments (i.e., percentage of days answered in between the first and last assessments), and the median time needed to complete the daily assessments. The typical time of completion (type response hour) was derived from time stamps of submitted surveys. Finally, we calculated the proportion of causal questions on which the “other causes” response option was used. All of the above are presented individually for each patient. Saturation analysis With the aim of quantifying day-to-day variation and to what extent assessing more days would eventually saturate a stable aggregated network for each patient, this analysis was a slight adaptation to the same analysis in Burger et al ( 2024 ). For each participant, we constructed one directed daily adjacency matrix (0/1) per assessment day, where each cell was coded as 1 if the perceived causal relation between the corresponding ordered pair of nodes was reported that day. To quantify how many assessment days were required for a participant’s network to stabilize, we repeatedly drew random sets of a assessment days from the participant’s available data without replacement and randomly partitioned each set into two equal-sized subsets of days. For each subset, daily adjacency matrices were aggregated to form a weighted network reflecting the proportion of days on which each causal relation was reported. Similarity between the two aggregated networks was quantified using the Spearman correlation between their edge weights. This procedure was repeated 1,000 times for each value of a , with a ranging from four days to the participant’s total number of available days in steps of two. The resulting distributions of correlations were summarized across values of a to obtain a saturation curve. In contrast to Burger et al. ( 2024 ), these analyses were not restricted to a core set of nodes, because nodes could be indicated as causes even if they were not reported as experienced on a given day; accordingly, all personalized nodes were retained for each participant. Also, pilot studies using individualized nodes indicated that this approach ensured nodes with adequate day-to-day frequency, avoiding the problem seen in Burger et al ( 2024 ), with most nodes having zero frequency. Causal drift analysis As in Burger et al ( 2024 ), we aimed to identify whether individual patients exhibited trends in their daily network structures. This was assessed by quantifying the extent to which similarity between daily perceived causal networks changed as a function of distance between assessment days. For each participant, all unique pairs of daily adjacency matrices were identified. Similarity between each pair was computed using the phi coefficient, reflecting the binary nature of the daily adjacency matrices (Burger et al 2024 used spearman correlations). Temporal distance was defined as the absolute difference in assessment day indices. For each participant, day to day trend was summarized descriptively as the regression slope obtained by regressing network similarity on temporal distance, with more negative slopes indicating greater divergence of network structure with larger distance between assessed days. Slopes were not subjected to inferential statistical testing. Network visualization and centrality measures Personalized causal networks were visualized using aggregated networks constructed across all available assessment days for each participant, with edge weights representing the proportion of days on which a directed causal relation was reported. To examine temporal variability in perceived causal structure, daily networks were also visualized for each participant, presented as a table with each cell representing the network for that day. To characterize the relative importance of individual nodes within aggregated networks, we computed outstrength centrality, defined as the sum of outgoing edge weights for each node. Outstrength values were standardized by the number of possible outgoing edges for each node to account for differences in number of nodes in the individual networks. Centrality measures were computed on the aggregated networks and interpreted descriptively as indicators of nodes with relatively strong perceived causal influence. Data were analyzed with R (version: 4.3.0, R Core Team 2023) in R-Studio (version: 2023.06.1, Posit team 2023 ) using the following packages: PECAN2 (v0.1.0, Reichert & Vogel. 2024), visNetwork (v2.1.2, Almende & Thieurmel, 2022 ), ggplot2 (v3.5.1, Wickham 2016 ), chromote (v0.5.1, Aden-Buie et al., 2025 ), htmlwidgets (v1.6.4, Vaidyanathan et al., 2023 ), webshot (v0.5.5, Chang et al., 2023), jaccard (v0.1.0, Chung et al., 2018 ), stringr (v1.5.1, Wickham, 2023 ), psych (v2.4.6.26, Revelle, 2024), car (v3.1-3, Fox et al., 2012 ), tidyr (v1.3.1, Wickham et al., 2024 )), janitor (v2.2.0, Firke, 2023 ), dplyr (v1.1.4, Wickham et al., 2023 ), and readxl (v1.4.3, Wickham & Bryan, 2023). Results Feasibility and adherence As can be seen in Table 2 , all five patients from the university clinic completed the 21-day L-PECAN protocol with good adherence. The number of completed assessments ranged from 15 to 21, corresponding to 71–100%. Median response duration per day ranged from 2 to 6 minutes. The “other causes” response option was used on 11 to 28% of causal questions, indicating that most perceived causal relations were captured within the network. The three patients from the specialist psychiatric clinic showed lower levels of adherence, ranging from answering 60% to 83% of days (note that the length of the L-PECAN phase varied for these patients). Response durations were similar to the university sample: ranging from 1 to 8 minutes. Also similar was the use of the “other causes” response alternative, which varied from 7% to 28%. Proportion of responses that were made the next day varied substantially, from none to almost three quarters. Table 2 Feasibility and adherence metrics Patient ID 101 102 103 104 105 106 = HS 107 = PJ 108 = DS Days responded 18 15 21 21 21 15 29 25 Response rate 86% 71% 100% 100% 100% 83% 79% 60% Response duration (median minutes) 3.6 2.1 5.1 5.9 2.7 3.0 1.0 8.0 Response hour (type value) 9 pm 1 am 8 pm 10 pm 8 pm 9 pm 12am 8 pm Responses next morning 6% 13% 0% 0% 19% 0% 3% 72% Average “other causes” 28% 13% 17% 11% 12% 28% 7% 20% Networks Individualized items can be found in Appendix C. Two example aggregated networks, one from the university clinic sample, and one from the specialized psychiatric clinic, can be seen in Fig. 1 (all aggregated patient networks can be found in Appendix D). Outdegree centralities for these two networks can be found in Fig. 2 . Saturation and drift As expected, networks tended to saturate (i.e., stabilize) the more days were included in aggregations. As expected, how quickly and to what extent networks saturated varied from patient to patient. Two “saturation plots” for two patients are shown in Fig. 3 . Note that causal saturation was calculated using all nodes for each participant (i.e., not limited to the core set, as in Burger et al, 2024 ). The plots of day to day trends (causal drift) for these two patients can be found in Fig. 4 . The day-by-day variability for patient 103 can be seen in Fig. 5 . In this figure, note how initially only a few nodes are selected each day as seen in red. Starting day 11, more nodes are selected each day, and with them more perceived causal relations are reported. One interpretation is that the increase of experienced nodes is maintained by a direct loop between “stuck worrying” and “negative emotions”. After a week, on day 18, the network winds down. Clinical utility Although clinical utility was not assessed systematically, anecdotal observations from patients and clinicians are worth noting. One patient in the university clinic sample commented: “The first week, the assessments made me feel worse. But that was probably because I had to accept the bad state I’m in. It made it more tangible how I actually felt. But after week one, it got better.” The same patient also noted that while the aggregated network itself was not particularly useful, the process of answering the questions was helpful: “Thinking about the causes did me good. It helped me understand that things are in fact connected. So when I saw the network, it was not a big surprise.” Similar reactions were reported by another patient in the university clinic, who described the network as largely consistent with her expectations and gave her no new insights. Therapists described the method as a potentially useful complement to clinical assessment, while also noting the trade-off between the burden of daily monitoring and help with identifying consistently central problems. Discussion Main findings We examined the feasibility of using daily perceived causal ratings to construct idiographic symptom networks before the start of psychiatric treatment in two clinical settings: five patients in a university clinic, and three from a specialist psychiatric clinic. Whereas the original demonstration of the method (Burger et al, 2024 ) relied on a self-selected, online-recruited sample which showed substantial attrition (making it impossible to know how representative the final analyzed sample was), we tested the method in routine clinical contexts with a known recruitment process, and patients who knew their responses might be used for their own care planning. We also updated the method by adding a training week to ensure the respondents understood the purpose and setup of the method, and adding a data-informed method for formulating idiosyncratic nodes. Further, we simplified the assessments by asking only about edges going into the node (not edges going out from the node). The results suggest that the updated method is feasible in a university clinic with relatively high-functioning patients, but more challenging to implement in specialist psychiatric care. In the university clinic, roughly half of those approached about participation in the study agreed to participate, and none discontinued after enrolling. Compliance was high, with three of five patients completing all assessment days and the lowest completion rate being 71%. By contrast, recruitment in the specialist setting was more difficult: of the patients approached about participation, only a quarter completed the full protocol, and adherence among these completers was still lower than in the university sample (60–79%). These findings suggest that, at least in its current form, daily causal self-ratings may not yet be realistic for patients with more severe or complex psychiatric presentations. For patients who did complete the protocol, the method functioned as intended. Daily assessments were short, between one and eight minutes on average. Patients largely stayed within their personalized item sets when assigning causes, suggesting that patients did view the constructed items as causally entangled (in line with the network approach to psychopathology). The within-person aggregated networks suggested that influence was unevenly distributed across problems: some nodes showed higher outgoing centrality (outstrength), indicating that they were perceived as affecting many other nodes on many days. These centrality patterns are descriptive and should be treated as hypothesis-generating rather than evidence of causal effects, but they may still be clinically useful for prioritizing potential treatment targets when planning interventions. Day-to-day variability in perceived causal relations was large. This is illustrated by the example in Fig. 3 , and by the slow causal saturation observed for most participants. Even with many days of observations, no patient had an aggregated network that was representative of all days. In other words, while some idiosyncratic problems were consistently present for a patient, the perceived causal structure connecting them varied across time, and there was little evidence that networks for patients converged toward a stable network with many days of data. These findings suggest that perceived causal structures are variable across time within individuals. From a methodological perspective, this highlights the value of explicitly examining fast acting temporal change in networks. As showcased in the present study, the causal drift metric can be used detect trends or phase shifts over time, e.g. over a menstrual cycle or important contextual changes. L-PECAN may be particularly well suited to an idiographic approach in which variability in problem networks is treated as information rather than noise. Finally, preliminary patient feedback in the university sample suggested that the procedure was acceptable but that the resulting network summary was not experienced as particularly informative. Patients engaged with the daily questions without difficulty, yet seeing the aggregated network did not seem to add insight beyond what they had already reflected on during the rating period. Limitations This study was small and designed for feasibility rather than generalizability, so the results should be viewed as preliminary. We did not evaluate whether the networks actually informed differential diagnostics, case formulation or treatment decisions, and importantly we also do not know whether the perceived causal relations correspond to actual causal relations. The method therefore remains unvalidated. The added value of the longitudinal procedure, with repeated small assessments, over a simpler single-timepoint PECAN interview is unknown. The initial training week may be helpful for learning the task and providing data for individualizing items, but simpler procedures might be sufficient. Importantly, the problems with using the method in a specialist psychiatric clinic is a notable limitation, since patients with more complex presentations are exactly those where idiographic networks may be most useful. Finally, the finding that perceived edges varied from day to day. This could reflect genuine fluctuation (i.e., true causal relations), but it could also arise from a tendency to report contrasts rather than absolute causes: one can easily imagine a patient who is always sad due to rumination, but instead of consistently reporting that cause, the patient each day reports causes that differ from day to day (e.g., one day conflicts at work, another day financial issues). Without external validation, it is difficult to know how to interpret this variability. Another issue is to what extent the observed within-person variability in network structures can be explained by variability in the symptoms, not the edges (e.g. worry is infrequently reported, but when it is the perceived contributing cause might consistently be conflicts with spouse). This could be explicitly analysed. Future directions Most importantly, clinical utility needs to be evaluated. Future studies should ask patients and clinicians how findings from a person-specific L-PECAN can be integrated with current practices of case conceptualization, either as data to be discussed together (e.g. Hall et al, 2025 ) or used as part of an algorithm to choose treatment foci (e.g. Rubel et al, 2018 ). Validity can be assessed after some treatment has taken place, i.e., “retrospective validity” (Andreasson et al, 2023 ). Ultimately, the utility should be measured in terms of improved or faster treatment outcomes (for a planned such study, see Lai et al, 2025 ). The substantial day-to-day variation in perceived edges should be explored more systematically. It may be useful to distinguish between stable influences and day-specific fluctuations, for example by prompting patients to indicate whether a reported edge reflects a general pattern or something unique about that day. Importantly, L-PECAN could be used to investigate cyclical patterns (e.g., symptoms linked to the menstrual cycle) or important contextual factors (e.g., different network structures on weekdays as opposed to weekends). Another possible extension would be to better capture the time-scales of different perceived causal relations. For every causal relation reported, the system could ask something like “Your said X contributed to Y today, how long after X did Y occur?” (phrased differently whether X and Y are discrete events or longer phenomena). Another possibility is that when a respondent indicates that X contributed to Y today, the patient could be asked when X occured. Possibly, the L-PECAN could use information about when X last occurred, so that the system could ask targeted follow-up questions. For example, if a respondent reports rumination today and selects a cause that was not experienced on the same day, the prompt could refer to the most recent occurrence of that symptom (e.g., “You reported symptom X four days ago. Is this what contributed to your rumination today?”). Such an approach could help distinguish short- from longer-term perceived influences while minimizing respondent burden, and may provide useful guidance for setting EMA sampling schedules aimed at estimating the same relations more objectively. The collaborative co-creation of nodes by patients and a clinician might seem atheoretical, but was obviously influenced by the predefined items used in the training week, and by the theoretical leanings of the researchers and clinicians who were all trained in CBT. This can be seen in for example in the university clinic sample, where all five patients had some variant of the three nodes “negative thoughts”, “physically inactive” and “tired” (i.e., a classic CBT triad of cognitions, behaviors and emotion/somatic). Since patients likely need guidance to formulate distinct nodes, it might make more sense to make the theoretical framework explicit. Indeed several network studies have used the framework of process based therapy (Hoffmann, 2025; Ong et al, 2025 ). Another interesting avenue might be to combine L-PECAN with a diagnosis specific model. An example of this can be found in Hertz-Palmor et al (under review), a study in which patients diagnosed with PTSD completed daily assessments about perceived causal relations between nodes which were personalized from a predefined model of PTSD. Only causal relations “allowed” by the model were asked about, thus decreasing the burden on participants at the cost of limiting the network to a specific diagnosis (e.g. precluding psychiatric comorbidities). In other words, focusing the data-collection or the analysis using priors about the specific patient (Burger et al, 2022 ) or the patient population (Bellander et al, under review) might decrease participant burden. Conclusion L-PECAN with personalized nodes is a structured method for assessing day-to-day variability in perceived causal networks. In this study, patients undergoing psychiatric evaluation were carefully involved in setting up their own data collection by first making sure they were on board with the purpose and idea of the study, e.g. by use of a psychoeducational video, and then trying out the method during an initial “training week”. Based on this, patients collaboratively formulated their personalized items. Evaluated under these relatively robust conditions, both some recurring causal relations and day-to-day variability were observed, pointing to the limitations of aggregated (i.e. assumed stable) person-level networks. Identifying both stability and fluctuation may be helpful when deciding on intervention targets and possibly timing. Declarations Author Contribution JR ran data analyses and edited manuscript. TB supervised and edited manuscript. TA supervised. ÄT, RF , KL and GA collected data. NJL and JB edited manuscript. LK initiated project and wrote first draft of maniscript. All authors reviewed the manuscript. Acknowledgement Frida Koernig, Alexandra Lönnroos, Olivia Wideroth, Jonas Holgersson and all participating patients and clinicians. Data Availability https://github.com/JR-psych/R_code_publications/tree/main/code%20and%20data%20lpecan References Aden-Buie, G., Chang, W., & Schloerke, B. (2025). chromote: Headless Chrome web browser interface (R package version 0.5.1).https://CRAN.R-project.org/package=chromote Almende B.V., & Thieurmel, B. (2022). visNetwork: Network visualization using the “vis.js” library (R package version 2.1.2).https://CRAN.R-project.org/package=visNetwork Andreasson, M., Schenström, J., Bjureberg, J., & Klintwall, L. (2023). Perceived causal networks: Clinical utility evaluated by therapists and patients. Journal for Person-Oriented Research, 9 (1), 29. Andreoli, G., Rafanelli, C., Hofmann, S. G., & Casu, G. (2025). A Systematic Scoping Review of Fully Idiographic Network Analysis in Mental Health. Cognitive Therapy and Research , 1-23. Bångstad, A., Fellman, J., Rosendahl, C., Bellander, M., Cervin, M., Bjureberg, J., & Klintwall, L. (2022). Perceived causal symptom network of adolescent mental health issues. Journal of Child & Adolescent Mental Health, 34 (1–3), 101–114. Bellander, M., Toftgård, A., Wedberg, R., Deserno, M. K., & Klintwall, L. (under review). Integrating expert expectations and patient perceptions in case conceptualizations . Manuscript under review. Borsboom, D. (2017). A network theory of mental disorders. World psychiatry , 16 (1), 5-13. Bridges-Curry, Z. An Idiographic Network Approach to Posttraumatic Stress Disorder: Symptom Heterogeneity and Implications for Treatment. Cogn Ther Res 49, 1045–1055 (2025). https://doi.org/10.1007/s10608-025-10575-4 Bringmann, L. F., Albers, C., Bockting, C., Borsboom, D., Ceulemans, E., Cramer, A., … Wichers, M. (2022). Psychopathological networks: Theory, methods and practice. Behaviour Research and Therapy, 149 , 104011. Bringmann, L. F., Elmer, T., Epskamp, S., Krause, R. W., Schoch, D., Wichers, M., … Snippe, E. (2019). What do centrality measures measure in psychological networks? Journal of Abnormal Psychology, 128 (8), 892–903. Burger, J., Andikkhash, V., Jäger, N., Anderbro, T., Blanken, T. F., & Klintwall, L. (2024). A novel approach for constructing personalized networks from longitudinal perceived causal relations. Behaviour Research and Therapy, 173 , 104456. Burger, J., Epskamp, S., van der Veen, D. C., Dablander, F., Schoevers, R. A., Fried, E. I., & Riese, H. (2022). A clinical PREMISE for personalized models: Toward a formal integration of case formulations and statistical networks. Journal of Psychopathology and Clinical Science, 131 (8), 906–918. Chang, W. (2023). webshot: Take screenshots of web pages (R package version 0.5.5).https://CRAN.R-project.org/package=webshot Chung, N. C., Miasojedow, B., Startek, M., & Gambin, A. (2018). jaccard: Test similarity between binary data using Jaccard/Tanimoto coefficients (R package version 0.1.0).https://CRAN.R-project.org/package=jaccard Cusack, C. E., Sandoval-Araujo, L. E., Hernández, J. C., Pennesi, J. L., Lazarus, G., Levinson, C. A., & Fisher, A. J. (2025). What’s strength centrality got to do with it? Examining the stability of central symptoms across symptom ensembles and time in idiographic networks. Journal of Psychopathology and Clinical Science . Firke, S. (2023). janitor: Simple tools for examining and cleaning dirty data (R package version 2.2.0).https://CRAN.R-project.org/package=janitor Fox, J., Weisberg, S., Adler, D., Bates, D., Baud-Bovy, G., Ellison, S., … Heiberger, R. (2012). car: Companion to applied regression (R package). R Foundation for Statistical Computing. Hall, M., Lappenbusch, L. M., Wiegmann, E., & Rubel, J. A. (2025). To Use or Not to Use: Exploring Therapists' Experiences with Pre-Treatment EMA-Based Personalized Feedback in the TheraNet Project. Administration and policy in mental health , 52 (1), 41–58. https://doi.org/10.1007/s10488-023-01333-3 Haynes, S. N., Smith, G. T., & Hunsley, J. D. (2018). Scientific foundations of clinical assessment . Routledge. Hertz-Palmor, N., Brown, G., Lazarov, A., Porat, J., Bevan, A., & Dalgleish, T. (under review). Momentary Structured Networks (MSN): A theory-guided approach to patient-elicited causal formulation . Manuscript under review. Hofmann, S. G. (2025). A network control theory of dynamic systems approach to personalize therapy. Behavior Therapy, 56 (1), 199–212. Hulsmans, D. H., Oude Maatman, F. J., Otten, R., Poelen, E. A., & Lichtwarck-Aschoff, A. (2024). Idiographic personality networks: Stability, variability and when they become problematic. Journal of Research in Personality, 109 , 104468. Kaariniemi, E., Bosund, V., Reichert, J., Bjureberg, J., & Klintwall, L. (2025). Perceived causal networks created using structured interviews: Feasibility and reliability. Cognitive Behaviour Therapy , 1–23. Klintwall, L., Bellander, M., & Cervin, M. (2023). Perceived causal problem networks: Reliability, central problems, and clinical utility for depression. Assessment, 30 (1), 73–83. Klintwall, L., Bellander, M., Stenius, L., Kenabi, S., Ebrahimi, O., Morales, N., Åkesson, M., Hadlaczky, G., Vogel, F., & Bjureberg, J. (in preparation). Recurring feedback loops in adolescents’ perceived causal networks of mental health problems: Feasibility, reliability, and clinical utility . Manuscript in preparation. von Klipstein, L., Stadel, M., Bos, F. M., Bringmann, L. F., Riese, H., & Servaas, M. N. (2025). Opening the contextual black box: A case for idiographic experience sampling of context for clinical applications. Quality of Life Research , 34 (3), 595-604. Kroenke, K., Spitzer, R. L., & Williams, J. B. W. (2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16 (9), 606–613. Kuyken, W., Padesky, C. A., & Dudley, R. (2008). The science and practice of case conceptualization. Behavioural and Cognitive Psychotherapy, 36 (6), 757–768. Lai, L., Li, Y., Zhao, Z., & Ren, Z. (2025). Efficacy of a process-based, Mobile-delivered personalized CBT for anxiety disorders: Study protocol for a randomized controlled trial. Internet interventions , 39 , 100805. https://doi.org/10.1016/j.invent.2025.100805 Lavefjord, A., Sundström, F. T. A., Hammar, A., Preihs, L., van de Leur, J. C., Forslund, S., ... & McCracken, L. M. (2026). Testing the network centrality hypothesis within process-based acceptance and commitment therapy–A single case experiment utilizing perceived causal networks. Journal of Contextual Behavioral Science , 100997. Levinson, C. A., Hunt, R. A., Keshishian, A. C., Brown, M. L., Vanzhula, I., Christian, C., ... & Williams, B. M. (2021). Using individual networks to identify treatment targets for eating disorder treatment: A proof-of-concept study and initial data. Journal of Eating Disorders , 9 (1), 147. Mansueto, A. C., Wiers, R. W., van Weert, J., Schouten, B. C., & Epskamp, S. (2023). Investigating the feasibility of idiographic network models. Psychological methods , 28 (5), 1052. Ong, C. W., Sheehan, K., Mann, A. J., & Fox, E. (2025). Examining the effects of process-based therapy: A multiple baseline study. Journal of Contextual Behavioral Science, 35 , 100875. Posit Team. (2023). RStudio: Integrated development environment for R . Posit Software.https://www.posit.co R Core Team. (2024). R: A language and environment for statistical computing (Version 4.3.3). R Foundation for Statistical Computing. Reichert, J., & Vogel, F. (2024). PECAN2 (R package). https://github.com/JR-psych/PECAN2 Reichert, J., Klimov, M., Kachel, A., et al. (2025, May 12). Perceived causal networks in patients with fibromyalgia and depression: Construction of a structured interview and testing reliability (Version 1) [Preprint]. Research Square. https://doi.org/10.21203/rs.3.rs-6352353/v1 Rubel, J. A., Fisher, A. J., Husen, K., & Lutz, W. (2018). Translating person-specific network models into personalized treatments: Development and demonstration of the dynamic assessment treatment algorithm for individual networks (DATA-IN). Psychotherapy and psychosomatics , 87 (4), 249-251. Scheffer, M., Bockting, C. L., Borsboom, D., Cools, R., Delecroix, C., Hartmann, J. A., … Nelson, B. (2024). A dynamical systems view of psychiatric disorders—Theory: A review. JAMA Psychiatry, 81 (6), 618–623. Siepe, B. S., Sander, C., Schultze, M., Kliem, A., Ludwig, S., Hegerl, U., & Reich, H. (2024). Time-varying network models for the temporal dynamics of depressive symptomatology in patients with depressive disorders. JMIR Mental Health, 11 , e50136. Spitzer, R. L., Kroenke, K., Williams, J. B. W., & Löwe, B. (2006). A brief measure for assessing generalized anxiety disorder: The GAD-7. Archives of Internal Medicine, 166 (10), 1092–1097. Tauseef, H. A., Coppersmith, D. D., Reid-Russell, A. J., Nagpal, A., Ross, J., Nock, M. K., & Eisenlohr-Moul, T. (2024). A call to integrate menstrual cycle influences into just-in-time adaptive interventions for suicide prevention. Frontiers in Psychiatry, 15 , 1434499. van der Tuin, S., Hoekstra, R. H., Booij, S. H., Oldehinkel, A. J., Wardenaar, K. J., van den Berg, D., … Wigman, J. T. (2023). Relating stability of individual dynamical networks to change in psychopathology. PLoS ONE, 18 (11), e0293200. Vaidyanathan, R., Xie, Y., Allaire, J., Cheng, J., Sievert, C., & Russell, K. (2023). htmlwidgets: HTML widgets for R (R package version 1.6.4).https://CRAN.R-project.org/package=htmlwidgets Vogel, F., Blanken, T. F., Burger, J., Reichert, J., Scholten, S., & Klintwall, L. (2025). How perceived causal networks can complement case conceptualization, diagnostic classification, and data-based networks. Journal of Psychopathology and Clinical Science . Wickham, H. (2016). ggplot2: Elegant graphics for data analysis . Springer. Wickham, H. (2023). stringr: Simple, consistent wrappers for common string operations (R package version 1.5.1).https://CRAN.R-project.org/package=stringr Wickham, H., & Bryan, J. (2025). readxl: Read Excel files (R package version 1.4.5).https://readxl.tidyverse.org Wickham, H., François, R., Henry, L., Müller, K., & Vaughan, D. (2023). dplyr: A grammar of data manipulation (R package version 1.1.4).https://dplyr.tidyverse.org Wickham, H., Vaughan, D., & Girlich, M. (2024). tidyr: Tidy messy data (R package version 1.3.1).https://tidyr.tidyverse.org Additional Declarations No competing interests reported. Supplementary Files AppendixA.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 19 May, 2026 Reviewers agreed at journal 30 Apr, 2026 Reviewers invited by journal 28 Apr, 2026 Editor assigned by journal 25 Apr, 2026 Submission checks completed at journal 25 Apr, 2026 First submitted to journal 23 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9504631","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":634083243,"identity":"a0d23715-aae7-453d-beec-b9fe84adfe6e","order_by":0,"name":"Julian Reichert","email":"","orcid":"","institution":"University Medical Center of the Johannes Gutenberg University Mainz","correspondingAuthor":false,"prefix":"","firstName":"Julian","middleName":"","lastName":"Reichert","suffix":""},{"id":634083246,"identity":"61631395-b70a-46f3-a6e7-e511f313b4a2","order_by":1,"name":"Tessa F. Blanken","email":"","orcid":"","institution":"University of Amsterdam","correspondingAuthor":false,"prefix":"","firstName":"Tessa","middleName":"F.","lastName":"Blanken","suffix":""},{"id":634083247,"identity":"5b97a17b-f75a-46e6-8958-0973daed54f5","order_by":2,"name":"Therese Anderbro","email":"","orcid":"","institution":"Stockholm University","correspondingAuthor":false,"prefix":"","firstName":"Therese","middleName":"","lastName":"Anderbro","suffix":""},{"id":634083249,"identity":"6d7deec3-6b8b-4aab-80b6-05c1be2ed7b5","order_by":3,"name":"Ämma Tangen","email":"","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":false,"prefix":"","firstName":"Ämma","middleName":"","lastName":"Tangen","suffix":""},{"id":634083250,"identity":"cb1b1efd-e2b2-496d-a5c4-fd12faab1525","order_by":4,"name":"Rikard Fjällström","email":"","orcid":"","institution":"Linnaeus University","correspondingAuthor":false,"prefix":"","firstName":"Rikard","middleName":"","lastName":"Fjällström","suffix":""},{"id":634083251,"identity":"fee9a11d-6e79-47d9-8b3c-d76f8635dd78","order_by":5,"name":"Klara Lychou","email":"","orcid":"","institution":"Region Stockolm","correspondingAuthor":false,"prefix":"","firstName":"Klara","middleName":"","lastName":"Lychou","suffix":""},{"id":634083252,"identity":"3cdb8d0e-5546-44dd-a942-ff5884502413","order_by":6,"name":"Gustav Almelöv","email":"","orcid":"","institution":"Region Stockolm","correspondingAuthor":false,"prefix":"","firstName":"Gustav","middleName":"","lastName":"Almelöv","suffix":""},{"id":634083253,"identity":"dbeb6337-38c9-40ec-81f3-cf30614add6c","order_by":7,"name":"Nitya Jayaram-Lindström","email":"","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":false,"prefix":"","firstName":"Nitya","middleName":"","lastName":"Jayaram-Lindström","suffix":""},{"id":634083254,"identity":"272d0924-c0de-4e62-b0b9-d8c960aac0c3","order_by":8,"name":"Julian Burger","email":"","orcid":"","institution":"Stony Brook University","correspondingAuthor":false,"prefix":"","firstName":"Julian","middleName":"","lastName":"Burger","suffix":""},{"id":634083255,"identity":"5162f975-d249-4e82-8ec9-8c070e46ad36","order_by":9,"name":"Lars Klintwall","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9klEQVRIiWNgGAWjYBACPjB5wIIHTD8osAGSjI0H8Glhg2iR4AGzEgzSQFoaiNLCANVyGMLHq4X97OMPDGckZMzleww/JBict1vbfhhoS41NNE4tPOlmEgw3JHgs23iMJRIMbidvO5MI1HIsLbcBp8PSgC76IMFjcIx3A1iL2QGgFsaGw7i18D9j/gDVsvlHgsG5ZLPzDwlokUhjADsMqGUb0JYDdmY3CNki8YxNIuEMSEv+N4sEg+QEsxtAWxLw+IWfP435w4djNvYGh48l3/hQYWdvdj794YMPNTY4tYBBAhI7sQFdhCCwJ0XxKBgFo2AUjAwAABuRXCL6BK6EAAAAAElFTkSuQmCC","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":true,"prefix":"","firstName":"Lars","middleName":"","lastName":"Klintwall","suffix":""}],"badges":[],"createdAt":"2026-04-23 09:09:50","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9504631/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9504631/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108944254,"identity":"a057be34-1986-4172-8a9f-afbe84954c01","added_by":"auto","created_at":"2026-05-11 05:58:03","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":139094,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eAggregated networks for patients 103 (left) and 107 (right). Edge weight corresponds to the proportion of days the relation was perceived. Most reported edge by patient 103 (“negative emotions” causing “stuck worrying”) was reported 52 % of days. Most reported edge by patient 107 (“Flashbacks” causing “Compulsive behaviors”) was reported 46 % of days. Note that these network visualizations do not take node frequency into account (e.g. node sizes are all equal, edge weights are not adjusted by source or target node frequency).\u003c/em\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-9504631/v1/6b06ca0809a71d8b792e4da4.png"},{"id":108944286,"identity":"831db1be-05ae-4696-92f3-41fd8581dd65","added_by":"auto","created_at":"2026-05-11 05:58:12","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":54865,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eOut-degree centralities from the aggregated networks for patients 103 and 107\u003c/em\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-9504631/v1/e3e4b56ab2b1d97a6046306f.png"},{"id":108944294,"identity":"b8c95c72-633c-45cf-a48b-e7907ec714fe","added_by":"auto","created_at":"2026-05-11 05:58:17","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":118823,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eCausal saturation plots for patients 103 and 107. Violins indicate distribution of spearman correlations between two repeatedly drawn samples, with x-axis indicating the size of the drawn samples. Colored bar indicates average across all bootstrapped samples.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-9504631/v1/818cea3f7dd2936018e50016.png"},{"id":108944261,"identity":"42402e2a-ad37-43e8-aa17-3e16ea1196d5","added_by":"auto","created_at":"2026-05-11 05:58:04","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":75176,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eCausal drift plots for patients 103 and 107. A slope indicates that daily network structures tend to be more dissimilar the further away two days are two one another, i.e. the patient exhibits differing phases across the assessment period (evident in 103 and not 107).\u003c/em\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-9504631/v1/a743c6380d6fff48b4aa3881.png"},{"id":108944259,"identity":"93f2b438-a195-4c00-82c9-dbe102650213","added_by":"auto","created_at":"2026-05-11 05:58:04","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":212698,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eSeparate frames for days, patient “103”. Animated GIF in online supp.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-9504631/v1/5d5585ac25689c21d270dba7.png"},{"id":108977257,"identity":"918e1a00-00f9-483f-a2bd-bc7eb91cf11e","added_by":"auto","created_at":"2026-05-11 11:31:05","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":922882,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9504631/v1/bb81d732-5b1a-41fa-9713-f99016f61dd3.pdf"},{"id":108944242,"identity":"46102a8d-d41b-44a1-ad54-de41e08b451c","added_by":"auto","created_at":"2026-05-11 05:57:54","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":21339,"visible":true,"origin":"","legend":"","description":"","filename":"AppendixA.docx","url":"https://assets-eu.researchsquare.com/files/rs-9504631/v1/64aa37335171e6b9be533e9d.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Day-to-day variation in perceived causal networks: a feasibility study in two psychiatric settings","fulltext":[{"header":"Background","content":"\u003cp\u003eDiagnostic categories in psychiatry are broad constructs that encompass substantial heterogeneity across individuals (Allsopp et al., 2019). Precision psychiatry aims to identify clinically meaningful differences between patients and to use this information to tailor treatment to the specific problems and needs of the individual, with the goal of improving treatment outcomes (Fernandes et al., 2017).\u003c/p\u003e \u003cp\u003eAccording to the network approach, psychopathology can be understood as a stable state in a network of causally entangled variables of behavioral, emotional, somatic and contextual variables (Borsboom, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). The network approach is particularly compatible with the goals of precision psychiatry. Rather than assuming that symptoms reflect a single latent disorder that is similar across individuals, the network approach conceptualizes psychopathology as a system of directly interacting symptoms and other factors that may differ substantially between individuals (Robinaugh et al., 2020). It is assumed that the relevant variables, i.e. nodes in the network, should cover symptoms irrespective of current diagnostic categories, and that these networks are to some extent person-specific (Vogel et al, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Following this conceptualisation, the network of a particular person can be seen as shifting between different distinct states, between healthy and pathological.\u003c/p\u003e \u003cp\u003eAlthough originating from different intellectual backgrounds, the network approach is similar to case conceptualizations in psychotherapy. In case conceptualizations, therapists and patients try to figure out how the context, behaviors and feelings of the patient are related to one another, in order to change central processes to maximize downstream effects (Kuyken, Padesky, \u0026amp; Dudley; \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). In other words, the therapist and patient work together to capture both the day-to-day variability and which factors are consistently causing other problems. Such analysis is an important clinical skill for therapists, but typically relies on somewhat unsystematic data collection and qualitative analysis (Haynes et al, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe network approach might contribute with better methods for both data collection and analysis, i.e. for personalizing treatment (Hofmann, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Lavefjord et al, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). The network approach is primarily a conceptual model of psychopathology, but it has also given rise to a growing set of empirical methods for assessing, estimating, and visualizing individual networks (Scheffer et al, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Networks are visualized as \u0026ldquo;nodes\u0026rdquo; (variables) connected by \u0026ldquo;edges\u0026rdquo; that reflect statistical or perceived relations between variables (Borsboom, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Edges can be undirected (\u0026ldquo;these nodes tend to co-occur\u0026rdquo;), or directed (\u0026ldquo;variability in this node predicts or is perceived to cause variability in this other node\u0026rdquo;). A concept that follows from this is idea of differing importance or influence of nodes within a network: i.e. \u0026ldquo;centrality\u0026rdquo;. Different centrality metrics are used to summarize how much a given node is connected to other nodes, for example by quantifying how many and how strong connections it has to other parts of the network (Bringmann et al, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Such central nodes are often interpreted as potentially influential points for intervention, under the assumption that changes in central nodes may have downstream effects on other problems (Vogel et al, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Note however that this has also been critiqued (e.g. Bringmann et al, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe network approach does not rely on a specific method for data collection, or a particular analysis. Rather, the approach is a way of thinking about psychopathology as a system of interacting problems, where the structure of causal relations may vary across persons and over time. From a clinical perspective, this offers a formalized language for ideas that are already familiar in psychotherapy, and opening up possibilities to investigate them in a more rigorous manner.\u003c/p\u003e \u003cp\u003ePerson-specific networks are typically estimated using dense repeated measures over several weeks (Ecological momentary assessment; EMA), which can be analysed using e.g. VAR (Vector Autoregression; for an overview see Bringmann et al, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). In a network estimated from such data, the directed edges between nodes in the network represent how much variation in one node predicts variation in the other nodes at the next assessment point (typically 2\u0026ndash;3 hours later). This method has been used widely to map between-person heterogeneity and inform intervention foci, for a wide range of patient populations, e.g. depression (Siepe et al, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), eating disorders (Levinson et al, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) and post-traumatic stress (Bridges-Curry, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Some studies use idiosyncratic nodes (for an overview, see Andreoli et al, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eHowever, time-series analyses comes with risk of missing relations because processes operate on multiple timescales within the same network, which makes alignment with sampling densities difficult (Bringmann et al, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Further, such time-series analyses are necessarily very data-hungry, limiting their clinical feasibility (Mansueto et al, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2023\u003c/span\u003e: Hall et al, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Another technique, Perceived Causal Networks (PECAN; Klintwall et al, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Vogel et al, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), creates a person specific network by simply asking the patient which causal relations he or she perceives between the nodes. This has the benefit of being able to capture processes of varying speed, but is obviously reliant on patient self insight and memory. The PECAN has been used in clinical settings (Andreasson et al, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), and in both self-guided questionnaire formats (B\u0026aring;ngstad et al, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and as a structured interview (Kaariniemi et al, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Klintwall et al, in preparation; Reichert et al, 2025).\u003c/p\u003e \u003cp\u003eBoth these methods assume that the network structure of a particular patient is stable across time, or at least across the time period of investigation. Indeed, there are empirical papers that find that at least for a majority of patients, this is true (e.g., van der Tuin et al, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Other studies by contrast find that networks change quite quickly (e.g., Cusack et al, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Hulsman et al, 2024; Siepe et al, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Some of this variation might plausibly be considered \u0026ldquo;noise,\u0026rdquo; i.e., random daily events affecting mood and behaviors, but e.g. for women some variation is likely due to hormonal cycles (see for example Tauseef et al, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) or important contextual factors such as weekdays vs weekends.\u003c/p\u003e \u003cp\u003eA method that bridges EMA and PECAN is Longitudinal PECAN (L-PECAN). In this method, the patient is asked each day which symptoms were experienced that day, and how these symptoms were perceived to causally influence one another. This approach avoids both the timescale misalignment of EMA and decreases the recall bias of PECAN. Daily assessments can be aggregated into an averaged network to highlight the most frequently perceived causal relations. Crucially, this method moreover allows an analysis of how the perceived causal links fluctuate from day to day. This offers insight into both the variability and the stability of node interactions. By examining these two aspects together, we can identify nodes that are not only consistently central across time but also understand when and how their causal influence changes. This information may be useful for tailoring interventions to individual patients (Rubel et al, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Hall et al, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn the study by Burger et al. (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), participants recruited through social media completed four weeks of daily assessments in which they selected the problems they had experienced that day from a predefined list (covering common symptoms, behaviors, and emotions) and rated how these symptoms influenced one another (i.e., both their causes and effects). For each participant, a \u003cem\u003ecore set\u003c/em\u003e of nodes was defined as those reported on at least one third of the days, and an aggregated network was created to identify consistent causal links within this core. For each participant, the rate of \u003cem\u003ecausal saturation\u003c/em\u003e was computed, defined as the number of assessed days needed to aggregate a stable network (i.e., when collecting more assessment days no longer changed the aggregated network). \u003cem\u003eDistance plots\u003c/em\u003e were also visualized for each participant, showing whether days tended to be more alike the closer they were to each other in time. While this study demonstrated that individuals can indeed provide daily reports on perceived causal relations over extended periods, there were several limitations noted. The self-selected, non-clinical sample limited generalizability to clinical settings. The predefined node-list was experienced as limiting by respondents, and excluded contextual factors altogether (e.g., \u003cem\u003efinancial problems\u003c/em\u003e; an issue pointed out previously, e.g. von Klipstein et al, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Interestingly, only about half the sample reached \u003cem\u003ecausal saturation\u003c/em\u003e, that is, consistent causal links over days, raising the question whether a stable network is to be expected, and whether L-PECAN could be used to investigate this within-person variability further.\u003c/p\u003e \u003cp\u003eThe aim of the present study was to build on the previous L-PECAN study by improving on the method (e.g., individualized items), and evaluate the clinical feasibility with actual psychiatric patients of differing severity. Importantly, we also aim to make the within-person variability the focus, in order to investigate whether L-PECAN can be used to analyze such day-to-day fluctuations. We report how we determined our sample size, all data exclusions, all manipulations, and all measures in the study.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eDesign and Setting\u003c/h2\u003e \u003cp\u003eWe use a within-subject observational design in which patients provide repeated daily assessment of perceived causal relations between personalized nodes over a three-week period. Each patient serves as their own unit of analysis, allowing both analysis of day-to-day variability and aggregation of person-specific networks. Data were collected in two clinical settings aimed at adult patients, but differing in symptom severity. The first setting is a university psychotherapy training clinic, i.e., outside of the public health care system, offering cognitive-behavioral therapy for adults with mild to moderate psychological problems such as anxiety, depression, or stress. The second setting is a specialist psychiatric outpatient clinic, part of routine secondary care, treating patients with severe and often comorbid conditions, including affective and neurodevelopmental disorders. Due to differing organizational requirements, L-PECAN was implemented in Qualtrics for the university clinic setting, and in RedCAP for the specialist psychiatry setting. Data collections has correspondingly two separate ethical approvals from the national Ethical Review Authority (reg. nrs. 2022\u0026ndash;03141 and 2024\u0026ndash;04165 respectively). To ensure patients are impossible to identify, we report no data on gender or age. The university clinic part of the study was preregistered at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://aspredicted.org/CY9_7WN\u003c/span\u003e\u003cspan address=\"https://aspredicted.org/CY9_7WN\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. All data and code can be found at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://osf.io/hma3e/overview?view_only=2fa0f1f341434be98dca8774bc930950\u003c/span\u003e\u003cspan address=\"https://osf.io/hma3e/overview?view_only=2fa0f1f341434be98dca8774bc930950\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. No funding was received for the conduct of this study or the preparation of this manuscript.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eProcedure\u003c/h3\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eRecruitment\u003c/h2\u003e \u003cp\u003eRecruitment was done during a phase of 6 months at the two participating clinics. During this time-period, all patients filling inclusion criteria were approached with information about the study, which stressed that participation was both voluntary and that findings could be used to personalize their subsequent care. Interested patients were sent a brief psychoeducational video introducing the purpose of L-PECAN, explaining the purpose of the method and how to interpret the causal question in the daily assessments. The video illustrated concrete examples to ensure the patients understood the purpose and setup of the method before they consented to participation, thus hopefully reducing the dropouts observed in the Burger et al (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) study. Patients still interested were given written study information and give their verbal and written consent.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eTraining Week\u003c/h3\u003e\n\u003cp\u003ePatients completed one week of daily assessments using a generic list of 18 common problems (e.g., overthinking, sleep problems, financial issues, see appendix A). This predefined list was an updated version of the list used in Burger et al (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), expanded to include contextual factors. Each evening, patients received a text message at a timepoint decided beforehand (between 18:00 and 20:30). The text message contained a link to the L-PECAN survey. Participants could complete the assessment the following morning, but they were instructed that answering at the end of the day in question was to be preferred (the proportion of responses that were in fact made the next morning is seen in Table \u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). They reported which items they had experienced that day, and described the selected items in their own words (e.g., \u0026ldquo;You selected \u003cem\u003eoverthinking\u003c/em\u003e, what did you think about?\u0026rdquo;). Importantly, respondents picked which of the other items (whether experienced on that day or not) had contributed to the experienced problem (i.e., ingoing edges). Patients were also allowed to write causes other than the other items.\u003c/p\u003e\n\u003ch3\u003eCollaborative Item Individualization\u003c/h3\u003e\n\u003cp\u003eAt the end of the training week, each patient met with a clinician (authors K.L. or G.A. in the university clinic, or \u0026Auml;.T. or R.F. in the specialist clinic) to review their week-one data. The clinician had prepared a summary of the freetext answers (item specifications and causes) and quantitative data (i.e., item frequencies and out-degree centralities), and prepared a suggestion for a shortened item-list with not more than ten items. This was collaboratively edited so that each item was relevant and made sense to the patient. In practice, this procedure worked by both combining and splitting items to find a good level of abstraction, and add idiosyncratic information provided by the patient in free-text boxes. We were particularly careful to avoid content overlap by formulating items on a more abstract level (e.g., combining \u0026ldquo;pain\u0026rdquo; and \u0026ldquo;tired\u0026rdquo; into one item if these were hard to distinguish). Items were prioritized if they were perceived to be causing other items, or were experienced as important by the patient. This finalized list was used to create a personalized for the L-PECAN for the remainder of the study for each patient. Finally, the patient could choose to change the timepoint each evening the text-message with a link to the survey was sent out.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eL-PECAN phase\u003c/h2\u003e \u003cp\u003ePatients in the university clinic completed at least three weeks of daily assessments using the L-PECAN. Patients from the specialist clinic could respond for a longer period, depending on their planned sessions. In each assessment, the patient was first shown the individualized list of items and asked to select those that were experienced that day (note that for item \u0026ldquo;insomnia\u0026rdquo;, the respondent was asked to answer about the previous night; meaning that for responses delayed to the next morning, the patient was answering about a night two nights distant). Next, contributing causes for each selected item was asked about (e.g., \u0026ldquo;Today, what contributed to \u0026lt;staying home from work\u0026gt;?\u0026rdquo;) with the other individualized items as potential contributing causes. The patient could select as many contributing causes as he or she wished, including none. An option \u0026ldquo;other causes\u0026rdquo; was also available. As compared to the Burger et al (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) version, the present study asked only about the causes (i.e., edges going into the node), not their effects (i.e., edges going out from the node). Another difference was that items were available as causes even though that item had not been reported as experienced that day (i.e. all items were always available as causes). The latter was changed to allow causal relations for time-scales longer than that day (e.g., being \u003cem\u003etired\u003c/em\u003e due to \u003cem\u003edrinking alcohol\u003c/em\u003e the previous day). For each item, pictograms were created and used throughout to reduce cognitive load and make the repeated assessments less tedious. Note that responses were binary: an item was either selected as a cause or not (i.e., there was no grading of the causal strength or certainty of the perception). For the university sample, some extra questions were asked in the L-PECAN, described in appendix B.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eOther measures\u003c/h3\u003e\n\u003cp\u003eDepressive symptoms were assessed with the \u003cem\u003ePatient Health Questionnaire\u0026ndash;9\u003c/em\u003e (PHQ-9; Kroenke, Spitzer, \u0026amp; Williams, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2001\u003c/span\u003e). The PHQ-9 consists of nine items corresponding to the diagnostic criteria for major depressive disorder. Each item is rated on a 4-point scale from 0 (\u003cem\u003enot at all\u003c/em\u003e) to 3 (\u003cem\u003enearly every day\u003c/em\u003e), yielding a total score between 0 and 27. Higher scores indicate greater depressive symptom severity.\u003c/p\u003e \u003cp\u003eAnxiety symptoms were assessed with the \u003cem\u003eGeneralized Anxiety Disorder\u0026ndash;7\u003c/em\u003e (GAD-7; Spitzer, Kroenke, Williams, \u0026amp; L\u0026ouml;we, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2006\u003c/span\u003e). The GAD-7 contains seven items measuring the frequency of core anxiety symptoms during the past two weeks, rated from 0 (\u003cem\u003enot at all\u003c/em\u003e) to 3 (\u003cem\u003enearly every day\u003c/em\u003e). Total scores range from 0 to 21, with higher scores reflecting higher levels of anxiety.\u003c/p\u003e\n\u003ch3\u003eParticipants\u003c/h3\u003e\n\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eUniversity Psychotherapy Clinic\u003c/h2\u003e \u003cp\u003ePatients self-referred to the university training clinic by completing an online intake form, which included the PHQ-9 (Kroenke et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2001\u003c/span\u003e) and the GAD-7 (Spitzer et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2006\u003c/span\u003e) for symptom screening. The university clinic routinely excludes individuals with psychosis, acute suicidality, substance dependence, or other conditions judged unsuitable for student-provided therapy. All patients are required to have sufficient proficiency in \u0026lt;redacted language\u0026thinsp;\u0026gt;\u0026thinsp;for psychotherapy. Our pre-registered criteria (requiring patients to have PHQ9 and GAD7 scores both over 5) were not applied, as inclusion was slower than we had expected. Nine patients were approached with information about the study. Two declined participation after having been given written information about the study. Another two declined participation after having been given verbal information about the study and seen the inclusion video. Thus, five out of nine approached patients were included in this setting. Descriptives in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eSpecialist Psychiatric Clinic\u003c/h2\u003e \u003cp\u003ePatients were recruited by clinicians during the early stages of their psychiatric evaluation before intervention initiation. Twelve patients were approached with study information. Four declined participation after receiving detailed information, primarily because the daily ratings were perceived as demanding. Two were excluded due to unclear technical difficulties that prevented completion of any of the daily assessments (i.e., could not get the link to work). Thus, six patients initially consented participation in the study. Of these, three dropped out during data-collection for unclear reasons. Thus, three out of twelve patients were included from this setting. Descriptives in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDescriptives of included patients\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePatient\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePHQ-9\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGAD-7\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eGiven reason for seeking therapy\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e \u003cp\u003e\u003cem\u003eUniversity clinic\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003ePatient 101\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAnxiety, relationship issues\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003ePatient 102\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAnxiety, depression\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003ePatient 103\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003ePerformance anxiety, social anxiety\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003ePatient 104\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eObsessive thoughts and compulsive behaviors\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003ePatient 105\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAnxiety, depression\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e \u003cp\u003e\u003cem\u003eSpecialist psychiatric clinic\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003ePatient 106\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eDepression, relationship issues\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003ePatient 107\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAnxiety, anger\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003ePatient 108\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003emissing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTrouble focusing, restlessness\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eData Analysis\u003c/h2\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003eFeasibility and adherence metrics\u003c/h2\u003e \u003cp\u003eFor each participant, we calculated the percentage of completed daily assessments (i.e., percentage of days answered in between the first and last assessments), and the median time needed to complete the daily assessments. The typical time of completion (type response hour) was derived from time stamps of submitted surveys. Finally, we calculated the proportion of causal questions on which the \u0026ldquo;other causes\u0026rdquo; response option was used. All of the above are presented individually for each patient.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eSaturation analysis\u003c/h2\u003e \u003cp\u003eWith the aim of quantifying day-to-day variation and to what extent assessing more days would eventually saturate a stable aggregated network for each patient, this analysis was a slight adaptation to the same analysis in Burger et al (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). For each participant, we constructed one directed daily adjacency matrix (0/1) per assessment day, where each cell was coded as 1 if the perceived causal relation between the corresponding ordered pair of nodes was reported that day. To quantify how many assessment days were required for a participant\u0026rsquo;s network to stabilize, we repeatedly drew random sets of \u003cem\u003ea\u003c/em\u003e assessment days from the participant\u0026rsquo;s available data without replacement and randomly partitioned each set into two equal-sized subsets of days. For each subset, daily adjacency matrices were aggregated to form a weighted network reflecting the proportion of days on which each causal relation was reported. Similarity between the two aggregated networks was quantified using the Spearman correlation between their edge weights. This procedure was repeated 1,000 times for each value of \u003cem\u003ea\u003c/em\u003e, with \u003cem\u003ea\u003c/em\u003e ranging from four days to the participant\u0026rsquo;s total number of available days in steps of two. The resulting distributions of correlations were summarized across values of \u003cem\u003ea\u003c/em\u003e to obtain a saturation curve. In contrast to Burger et al. (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), these analyses were not restricted to a core set of nodes, because nodes could be indicated as causes even if they were not reported as experienced on a given day; accordingly, all personalized nodes were retained for each participant. Also, pilot studies using individualized nodes indicated that this approach ensured nodes with adequate day-to-day frequency, avoiding the problem seen in Burger et al (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), with most nodes having zero frequency.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eCausal drift analysis\u003c/h2\u003e \u003cp\u003eAs in Burger et al (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), we aimed to identify whether individual patients exhibited trends in their daily network structures. This was assessed by quantifying the extent to which similarity between daily perceived causal networks changed as a function of distance between assessment days. For each participant, all unique pairs of daily adjacency matrices were identified. Similarity between each pair was computed using the phi coefficient, reflecting the binary nature of the daily adjacency matrices (Burger et al \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e used spearman correlations). Temporal distance was defined as the absolute difference in assessment day indices. For each participant, day to day trend was summarized descriptively as the regression slope obtained by regressing network similarity on temporal distance, with more negative slopes indicating greater divergence of network structure with larger distance between assessed days. Slopes were not subjected to inferential statistical testing.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eNetwork visualization and centrality measures\u003c/h2\u003e \u003cp\u003ePersonalized causal networks were visualized using aggregated networks constructed across all available assessment days for each participant, with edge weights representing the proportion of days on which a directed causal relation was reported. To examine temporal variability in perceived causal structure, daily networks were also visualized for each participant, presented as a table with each cell representing the network for that day.\u003c/p\u003e \u003cp\u003eTo characterize the relative importance of individual nodes within aggregated networks, we computed outstrength centrality, defined as the sum of outgoing edge weights for each node. Outstrength values were standardized by the number of possible outgoing edges for each node to account for differences in number of nodes in the individual networks. Centrality measures were computed on the aggregated networks and interpreted descriptively as indicators of nodes with relatively strong perceived causal influence.\u003c/p\u003e \u003cp\u003eData were analyzed with R (version: 4.3.0, R Core Team 2023) in R-Studio (version: 2023.06.1, Posit team \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) using the following packages: PECAN2 (v0.1.0, Reichert \u0026amp; Vogel. 2024), visNetwork (v2.1.2, Almende \u0026amp; Thieurmel, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), ggplot2 (v3.5.1, Wickham \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), chromote (v0.5.1, Aden-Buie et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), htmlwidgets (v1.6.4, Vaidyanathan et al., \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), webshot (v0.5.5, Chang et al., 2023), jaccard (v0.1.0, Chung et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), stringr (v1.5.1, Wickham, \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), psych (v2.4.6.26, Revelle, 2024), car (v3.1-3, Fox et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2012\u003c/span\u003e), tidyr (v1.3.1, Wickham et al., \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2024\u003c/span\u003e)), janitor (v2.2.0, Firke, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), dplyr (v1.1.4, Wickham et al., \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), and readxl (v1.4.3, Wickham \u0026amp; Bryan, 2023).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eFeasibility and adherence\u003c/h2\u003e \u003cp\u003eAs can be seen in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, all five patients from the university clinic completed the 21-day L-PECAN protocol with good adherence. The number of completed assessments ranged from 15 to 21, corresponding to 71\u0026ndash;100%. Median response duration per day ranged from 2 to 6 minutes. The \u0026ldquo;other causes\u0026rdquo; response option was used on 11 to 28% of causal questions, indicating that most perceived causal relations were captured within the network. The three patients from the specialist psychiatric clinic showed lower levels of adherence, ranging from answering 60% to 83% of days (note that the length of the L-PECAN phase varied for these patients). Response durations were similar to the university sample: ranging from 1 to 8 minutes. Also similar was the use of the \u0026ldquo;other causes\u0026rdquo; response alternative, which varied from 7% to 28%. Proportion of responses that were made the next day varied substantially, from none to almost three quarters.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eFeasibility and adherence metrics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePatient ID\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e101\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e102\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e103\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e104\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e105\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e106\u0026thinsp;=\u0026thinsp;HS\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003e107\u0026thinsp;=\u0026thinsp;PJ\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003e108\u0026thinsp;=\u0026thinsp;DS\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDays responded\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e25\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResponse rate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e86%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e71%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e100%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e83%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e79%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e60%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResponse duration (median minutes)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e1.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e8.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResponse hour (type value)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9 pm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1 am\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8 pm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e10 pm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e8 pm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e9 pm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e12am\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e8 pm\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResponses next morning\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e19%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e3%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e72%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage \u0026ldquo;other causes\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e28%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e17%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e11%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e12%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e28%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e7%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e20%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eNetworks\u003c/h2\u003e \u003cp\u003eIndividualized items can be found in Appendix C. Two example aggregated networks, one from the university clinic sample, and one from the specialized psychiatric clinic, can be seen in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e (all aggregated patient networks can be found in Appendix D). Outdegree centralities for these two networks can be found in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eSaturation and drift\u003c/h2\u003e \u003cp\u003eAs expected, networks tended to saturate (i.e., stabilize) the more days were included in aggregations. As expected, how quickly and to what extent networks saturated varied from patient to patient. Two \u0026ldquo;saturation plots\u0026rdquo; for two patients are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Note that causal saturation was calculated using all nodes for each participant (i.e., not limited to the core set, as in Burger et al, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). The plots of day to day trends (causal drift) for these two patients can be found in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe day-by-day variability for patient 103 can be seen in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e. In this figure, note how initially only a few nodes are selected each day as seen in red. Starting day 11, more nodes are selected each day, and with them more perceived causal relations are reported. One interpretation is that the increase of experienced nodes is maintained by a direct loop between \u0026ldquo;stuck worrying\u0026rdquo; and \u0026ldquo;negative emotions\u0026rdquo;. After a week, on day 18, the network winds down.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eClinical utility\u003c/h2\u003e \u003cp\u003eAlthough clinical utility was not assessed systematically, anecdotal observations from patients and clinicians are worth noting. One patient in the university clinic sample commented: \u0026ldquo;The first week, the assessments made me feel worse. But that was probably because I had to accept the bad state I\u0026rsquo;m in. It made it more tangible how I actually felt. But after week one, it got better.\u0026rdquo; The same patient also noted that while the aggregated network itself was not particularly useful, the process of answering the questions was helpful: \u0026ldquo;Thinking about the causes did me good. It helped me understand that things are in fact connected. So when I saw the network, it was not a big surprise.\u0026rdquo; Similar reactions were reported by another patient in the university clinic, who described the network as largely consistent with her expectations and gave her no new insights. Therapists described the method as a potentially useful complement to clinical assessment, while also noting the trade-off between the burden of daily monitoring and help with identifying consistently central problems.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eMain findings\u003c/h2\u003e \u003cp\u003eWe examined the feasibility of using daily perceived causal ratings to construct idiographic symptom networks before the start of psychiatric treatment in two clinical settings: five patients in a university clinic, and three from a specialist psychiatric clinic. Whereas the original demonstration of the method (Burger et al, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) relied on a self-selected, online-recruited sample which showed substantial attrition (making it impossible to know how representative the final analyzed sample was), we tested the method in routine clinical contexts with a known recruitment process, and patients who knew their responses might be used for their own care planning. We also updated the method by adding a training week to ensure the respondents understood the purpose and setup of the method, and adding a data-informed method for formulating idiosyncratic nodes. Further, we simplified the assessments by asking only about edges going into the node (not edges going out from the node).\u003c/p\u003e \u003cp\u003eThe results suggest that the updated method is feasible in a university clinic with relatively high-functioning patients, but more challenging to implement in specialist psychiatric care. In the university clinic, roughly half of those approached about participation in the study agreed to participate, and none discontinued after enrolling. Compliance was high, with three of five patients completing all assessment days and the lowest completion rate being 71%. By contrast, recruitment in the specialist setting was more difficult: of the patients approached about participation, only a quarter completed the full protocol, and adherence among these completers was still lower than in the university sample (60\u0026ndash;79%). These findings suggest that, at least in its current form, daily causal self-ratings may not yet be realistic for patients with more severe or complex psychiatric presentations.\u003c/p\u003e \u003cp\u003eFor patients who did complete the protocol, the method functioned as intended. Daily assessments were short, between one and eight minutes on average. Patients largely stayed within their personalized item sets when assigning causes, suggesting that patients did view the constructed items as causally entangled (in line with the network approach to psychopathology).\u003c/p\u003e \u003cp\u003eThe within-person aggregated networks suggested that influence was unevenly distributed across problems: some nodes showed higher outgoing centrality (outstrength), indicating that they were perceived as affecting many other nodes on many days. These centrality patterns are descriptive and should be treated as hypothesis-generating rather than evidence of causal effects, but they may still be clinically useful for prioritizing potential treatment targets when planning interventions.\u003c/p\u003e \u003cp\u003eDay-to-day variability in perceived causal relations was large. This is illustrated by the example in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, and by the slow causal saturation observed for most participants. Even with many days of observations, no patient had an aggregated network that was representative of all days. In other words, while some idiosyncratic problems were consistently present for a patient, the perceived causal structure connecting them varied across time, and there was little evidence that networks for patients converged toward a stable network with many days of data. These findings suggest that perceived causal structures are variable across time within individuals. From a methodological perspective, this highlights the value of explicitly examining fast acting temporal change in networks. As showcased in the present study, the causal drift metric can be used detect trends or phase shifts over time, e.g. over a menstrual cycle or important contextual changes. L-PECAN may be particularly well suited to an idiographic approach in which variability in problem networks is treated as information rather than noise.\u003c/p\u003e \u003cp\u003eFinally, preliminary patient feedback in the university sample suggested that the procedure was acceptable but that the resulting network summary was not experienced as particularly informative. Patients engaged with the daily questions without difficulty, yet seeing the aggregated network did not seem to add insight beyond what they had already reflected on during the rating period.\u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eLimitations\u003c/h2\u003e \u003cp\u003eThis study was small and designed for feasibility rather than generalizability, so the results should be viewed as preliminary. We did not evaluate whether the networks actually informed differential diagnostics, case formulation or treatment decisions, and importantly we also do not know whether the perceived causal relations correspond to actual causal relations. The method therefore remains unvalidated.\u003c/p\u003e \u003cp\u003eThe added value of the longitudinal procedure, with repeated small assessments, over a simpler single-timepoint PECAN interview is unknown. The initial training week may be helpful for learning the task and providing data for individualizing items, but simpler procedures might be sufficient. Importantly, the problems with using the method in a specialist psychiatric clinic is a notable limitation, since patients with more complex presentations are exactly those where idiographic networks may be most useful.\u003c/p\u003e \u003cp\u003eFinally, the finding that perceived edges varied from day to day. This could reflect genuine fluctuation (i.e., true causal relations), but it could also arise from a tendency to report \u003cem\u003econtrasts\u003c/em\u003e rather than absolute causes: one can easily imagine a patient who is always sad due to rumination, but instead of consistently reporting that cause, the patient each day reports causes that differ from day to day (e.g., one day conflicts at work, another day financial issues). Without external validation, it is difficult to know how to interpret this variability. Another issue is to what extent the observed within-person variability in network structures can be explained by variability in the symptoms, not the edges (e.g. worry is infrequently reported, but when it is the perceived contributing cause might consistently be conflicts with spouse). This could be explicitly analysed.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003eFuture directions\u003c/h2\u003e \u003cp\u003eMost importantly, clinical utility needs to be evaluated. Future studies should ask patients and clinicians how findings from a person-specific L-PECAN can be integrated with current practices of case conceptualization, either as data to be discussed together (e.g. Hall et al, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) or used as part of an algorithm to choose treatment foci (e.g. Rubel et al, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Validity can be assessed after some treatment has taken place, i.e., \u0026ldquo;retrospective validity\u0026rdquo; (Andreasson et al, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Ultimately, the utility should be measured in terms of improved or faster treatment outcomes (for a planned such study, see Lai et al, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe substantial day-to-day variation in perceived edges should be explored more systematically. It may be useful to distinguish between stable influences and day-specific fluctuations, for example by prompting patients to indicate whether a reported edge reflects a general pattern or something unique about that day. Importantly, L-PECAN could be used to investigate cyclical patterns (e.g., symptoms linked to the menstrual cycle) or important contextual factors (e.g., different network structures on weekdays as opposed to weekends).\u003c/p\u003e \u003cp\u003eAnother possible extension would be to better capture the time-scales of different perceived causal relations. For every causal relation reported, the system could ask something like \u0026ldquo;Your said X contributed to Y today, how long after X did Y occur?\u0026rdquo; (phrased differently whether X and Y are discrete events or longer phenomena). Another possibility is that when a respondent indicates that X contributed to Y today, the patient could be asked when X occured. Possibly, the L-PECAN could use information about when X last occurred, so that the system could ask targeted follow-up questions. For example, if a respondent reports rumination today and selects a cause that was not experienced on the same day, the prompt could refer to the most recent occurrence of that symptom (e.g., \u0026ldquo;You reported symptom X four days ago. Is this what contributed to your rumination today?\u0026rdquo;). Such an approach could help distinguish short- from longer-term perceived influences while minimizing respondent burden, and may provide useful guidance for setting EMA sampling schedules aimed at estimating the same relations more objectively.\u003c/p\u003e \u003cp\u003eThe collaborative co-creation of nodes by patients and a clinician might seem atheoretical, but was obviously influenced by the predefined items used in the training week, and by the theoretical leanings of the researchers and clinicians who were all trained in CBT. This can be seen in for example in the university clinic sample, where all five patients had some variant of the three nodes \u0026ldquo;negative thoughts\u0026rdquo;, \u0026ldquo;physically inactive\u0026rdquo; and \u0026ldquo;tired\u0026rdquo; (i.e., a classic CBT triad of cognitions, behaviors and emotion/somatic). Since patients likely need guidance to formulate distinct nodes, it might make more sense to make the theoretical framework explicit. Indeed several network studies have used the framework of process based therapy (Hoffmann, 2025; Ong et al, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Another interesting avenue might be to combine L-PECAN with a diagnosis specific model. An example of this can be found in Hertz-Palmor et al (under review), a study in which patients diagnosed with PTSD completed daily assessments about perceived causal relations between nodes which were personalized from a predefined model of PTSD. Only causal relations \u0026ldquo;allowed\u0026rdquo; by the model were asked about, thus decreasing the burden on participants at the cost of limiting the network to a specific diagnosis (e.g. precluding psychiatric comorbidities). In other words, focusing the data-collection or the analysis using priors about the specific patient (Burger et al, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) or the patient population (Bellander et al, under review) might decrease participant burden.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eL-PECAN with personalized nodes is a structured method for assessing day-to-day variability in perceived causal networks. In this study, patients undergoing psychiatric evaluation were carefully involved in setting up their own data collection by first making sure they were on board with the purpose and idea of the study, e.g. by use of a psychoeducational video, and then trying out the method during an initial \u0026ldquo;training week\u0026rdquo;. Based on this, patients collaboratively formulated their personalized items. Evaluated under these relatively robust conditions, both some recurring causal relations and day-to-day variability were observed, pointing to the limitations of aggregated (i.e. assumed stable) person-level networks. Identifying both stability and fluctuation may be helpful when deciding on intervention targets and possibly timing.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eJR ran data analyses and edited manuscript. TB supervised and edited manuscript. TA supervised. \u0026Auml;T, RF , KL and GA collected data. NJL and JB edited manuscript. LK initiated project and wrote first draft of maniscript. All authors reviewed the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eFrida Koernig, Alexandra L\u0026ouml;nnroos, Olivia Wideroth, Jonas Holgersson and all participating patients and clinicians.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003ehttps://github.com/JR-psych/R_code_publications/tree/main/code%20and%20data%20lpecan\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAden-Buie, G., Chang, W., \u0026amp; Schloerke, B. (2025). \u003cem\u003echromote: Headless Chrome web browser interface\u003c/em\u003e (R package version 0.5.1).https://CRAN.R-project.org/package=chromote\u003c/li\u003e\n\u003cli\u003eAlmende B.V., \u0026amp; Thieurmel, B. (2022). \u003cem\u003evisNetwork: Network visualization using the \u0026ldquo;vis.js\u0026rdquo; library\u003c/em\u003e (R package version 2.1.2).https://CRAN.R-project.org/package=visNetwork\u003c/li\u003e\n\u003cli\u003eAndreasson, M., Schenstr\u0026ouml;m, J., Bjureberg, J., \u0026amp; Klintwall, L. (2023). Perceived causal networks: Clinical utility evaluated by therapists and patients. \u003cem\u003eJournal for Person-Oriented Research, 9\u003c/em\u003e(1), 29.\u003c/li\u003e\n\u003cli\u003eAndreoli, G., Rafanelli, C., Hofmann, S. G., \u0026amp; Casu, G. (2025). A Systematic Scoping Review of Fully Idiographic Network Analysis in Mental Health. \u003cem\u003eCognitive Therapy and Research\u003c/em\u003e, 1-23.\u003c/li\u003e\n\u003cli\u003eB\u0026aring;ngstad, A., Fellman, J., Rosendahl, C., Bellander, M., Cervin, M., Bjureberg, J., \u0026amp; Klintwall, L. (2022). Perceived causal symptom network of adolescent mental health issues. \u003cem\u003eJournal of Child \u0026amp; Adolescent Mental Health, 34\u003c/em\u003e(1\u0026ndash;3), 101\u0026ndash;114.\u003c/li\u003e\n\u003cli\u003eBellander, M., Toftg\u0026aring;rd, A., Wedberg, R., Deserno, M. K., \u0026amp; Klintwall, L. (under review). \u003cem\u003eIntegrating expert expectations and patient perceptions in case conceptualizations\u003c/em\u003e. Manuscript under review.\u003c/li\u003e\n\u003cli\u003eBorsboom, D. (2017). A network theory of mental disorders. \u003cem\u003eWorld psychiatry\u003c/em\u003e, \u003cem\u003e16\u003c/em\u003e(1), 5-13.\u003c/li\u003e\n\u003cli\u003eBridges-Curry, Z. An Idiographic Network Approach to Posttraumatic Stress Disorder: Symptom Heterogeneity and Implications for Treatment. Cogn Ther Res 49, 1045\u0026ndash;1055 (2025). https://doi.org/10.1007/s10608-025-10575-4\u003c/li\u003e\n\u003cli\u003eBringmann, L. F., Albers, C., Bockting, C., Borsboom, D., Ceulemans, E., Cramer, A., \u0026hellip; Wichers, M. (2022). Psychopathological networks: Theory, methods and practice. \u003cem\u003eBehaviour Research and Therapy, 149\u003c/em\u003e, 104011.\u003c/li\u003e\n\u003cli\u003eBringmann, L. F., Elmer, T., Epskamp, S., Krause, R. W., Schoch, D., Wichers, M., \u0026hellip; Snippe, E. (2019). What do centrality measures measure in psychological networks? \u003cem\u003eJournal of Abnormal Psychology, 128\u003c/em\u003e(8), 892\u0026ndash;903.\u003c/li\u003e\n\u003cli\u003eBurger, J., Andikkhash, V., J\u0026auml;ger, N., Anderbro, T., Blanken, T. F., \u0026amp; Klintwall, L. (2024). A novel approach for constructing personalized networks from longitudinal perceived causal relations. \u003cem\u003eBehaviour Research and Therapy, 173\u003c/em\u003e, 104456.\u003c/li\u003e\n\u003cli\u003eBurger, J., Epskamp, S., van der Veen, D. C., Dablander, F., Schoevers, R. A., Fried, E. I., \u0026amp; Riese, H. (2022). A clinical PREMISE for personalized models: Toward a formal integration of case formulations and statistical networks. \u003cem\u003eJournal of Psychopathology and Clinical Science, 131\u003c/em\u003e(8), 906\u0026ndash;918.\u003c/li\u003e\n\u003cli\u003eChang, W. (2023). \u003cem\u003ewebshot: Take screenshots of web pages\u003c/em\u003e (R package version 0.5.5).https://CRAN.R-project.org/package=webshot\u003c/li\u003e\n\u003cli\u003eChung, N. C., Miasojedow, B., Startek, M., \u0026amp; Gambin, A. (2018). \u003cem\u003ejaccard: Test similarity between binary data using Jaccard/Tanimoto coefficients\u003c/em\u003e (R package version 0.1.0).https://CRAN.R-project.org/package=jaccard\u003c/li\u003e\n\u003cli\u003eCusack, C. E., Sandoval-Araujo, L. E., Hern\u0026aacute;ndez, J. C., Pennesi, J. L., Lazarus, G., Levinson, C. A., \u0026amp; Fisher, A. J. (2025). What\u0026rsquo;s strength centrality got to do with it? Examining the stability of central symptoms across symptom ensembles and time in idiographic networks. \u003cem\u003eJournal of Psychopathology and Clinical Science\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eFirke, S. (2023). \u003cem\u003ejanitor: Simple tools for examining and cleaning dirty data\u003c/em\u003e (R package version 2.2.0).https://CRAN.R-project.org/package=janitor\u003c/li\u003e\n\u003cli\u003eFox, J., Weisberg, S., Adler, D., Bates, D., Baud-Bovy, G., Ellison, S., \u0026hellip; Heiberger, R. (2012). \u003cem\u003ecar: Companion to applied regression\u003c/em\u003e (R package). R Foundation for Statistical Computing.\u003c/li\u003e\n\u003cli\u003eHall, M., Lappenbusch, L. M., Wiegmann, E., \u0026amp; Rubel, J. A. (2025). To Use or Not to Use: Exploring Therapists\u0026apos; Experiences with Pre-Treatment EMA-Based Personalized Feedback in the TheraNet Project. \u003cem\u003eAdministration and policy in mental health\u003c/em\u003e, \u003cem\u003e52\u003c/em\u003e(1), 41\u0026ndash;58. https://doi.org/10.1007/s10488-023-01333-3\u003c/li\u003e\n\u003cli\u003eHaynes, S. N., Smith, G. T., \u0026amp; Hunsley, J. D. (2018). \u003cem\u003eScientific foundations of clinical assessment\u003c/em\u003e. Routledge.\u003c/li\u003e\n\u003cli\u003eHertz-Palmor, N., Brown, G., Lazarov, A., Porat, J., Bevan, A., \u0026amp; Dalgleish, T. (under review). \u003cem\u003eMomentary Structured Networks (MSN): A theory-guided approach to patient-elicited causal formulation\u003c/em\u003e. Manuscript under review.\u003c/li\u003e\n\u003cli\u003eHofmann, S. G. (2025). A network control theory of dynamic systems approach to personalize therapy. \u003cem\u003eBehavior Therapy, 56\u003c/em\u003e(1), 199\u0026ndash;212.\u003c/li\u003e\n\u003cli\u003eHulsmans, D. H., Oude Maatman, F. J., Otten, R., Poelen, E. A., \u0026amp; Lichtwarck-Aschoff, A. (2024). Idiographic personality networks: Stability, variability and when they become problematic. \u003cem\u003eJournal of Research in Personality, 109\u003c/em\u003e, 104468.\u003c/li\u003e\n\u003cli\u003eKaariniemi, E., Bosund, V., Reichert, J., Bjureberg, J., \u0026amp; Klintwall, L. (2025). Perceived causal networks created using structured interviews: Feasibility and reliability. \u003cem\u003eCognitive Behaviour Therapy\u003c/em\u003e, 1\u0026ndash;23.\u003c/li\u003e\n\u003cli\u003eKlintwall, L., Bellander, M., \u0026amp; Cervin, M. (2023). Perceived causal problem networks: Reliability, central problems, and clinical utility for depression. \u003cem\u003eAssessment, 30\u003c/em\u003e(1), 73\u0026ndash;83.\u003c/li\u003e\n\u003cli\u003eKlintwall, L., Bellander, M., Stenius, L., Kenabi, S., Ebrahimi, O., Morales, N., \u0026Aring;kesson, M., Hadlaczky, G., Vogel, F., \u0026amp; Bjureberg, J. (in preparation). \u003cem\u003eRecurring feedback loops in adolescents\u0026rsquo; perceived causal networks of mental health problems: Feasibility, reliability, and clinical utility\u003c/em\u003e. Manuscript in preparation.\u003c/li\u003e\n\u003cli\u003evon Klipstein, L., Stadel, M., Bos, F. M., Bringmann, L. F., Riese, H., \u0026amp; Servaas, M. N. (2025). Opening the contextual black box: A case for idiographic experience sampling of context for clinical applications. \u003cem\u003eQuality of Life Research\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(3), 595-604.\u003c/li\u003e\n\u003cli\u003eKroenke, K., Spitzer, R. L., \u0026amp; Williams, J. B. W. (2001). The PHQ-9: Validity of a brief depression severity measure. \u003cem\u003eJournal of General Internal Medicine, 16\u003c/em\u003e(9), 606\u0026ndash;613.\u003c/li\u003e\n\u003cli\u003eKuyken, W., Padesky, C. A., \u0026amp; Dudley, R. (2008). The science and practice of case conceptualization. \u003cem\u003eBehavioural and Cognitive Psychotherapy, 36\u003c/em\u003e(6), 757\u0026ndash;768.\u003c/li\u003e\n\u003cli\u003eLai, L., Li, Y., Zhao, Z., \u0026amp; Ren, Z. (2025). Efficacy of a process-based, Mobile-delivered personalized CBT for anxiety disorders: Study protocol for a randomized controlled trial. \u003cem\u003eInternet interventions\u003c/em\u003e, \u003cem\u003e39\u003c/em\u003e, 100805. https://doi.org/10.1016/j.invent.2025.100805\u003c/li\u003e\n\u003cli\u003eLavefjord, A., Sundstr\u0026ouml;m, F. T. A., Hammar, A., Preihs, L., van de Leur, J. C., Forslund, S., ... \u0026amp; McCracken, L. M. (2026). Testing the network centrality hypothesis within process-based acceptance and commitment therapy\u0026ndash;A single case experiment utilizing perceived causal networks. \u003cem\u003eJournal of Contextual Behavioral Science\u003c/em\u003e, 100997.\u003c/li\u003e\n\u003cli\u003eLevinson, C. A., Hunt, R. A., Keshishian, A. C., Brown, M. L., Vanzhula, I., Christian, C., ... \u0026amp; Williams, B. M. (2021). Using individual networks to identify treatment targets for eating disorder treatment: A proof-of-concept study and initial data. \u003cem\u003eJournal of Eating Disorders\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(1), 147.\u003c/li\u003e\n\u003cli\u003eMansueto, A. C., Wiers, R. W., van Weert, J., Schouten, B. C., \u0026amp; Epskamp, S. (2023). Investigating the feasibility of idiographic network models. \u003cem\u003ePsychological methods\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e(5), 1052.\u003c/li\u003e\n\u003cli\u003eOng, C. W., Sheehan, K., Mann, A. J., \u0026amp; Fox, E. (2025). Examining the effects of process-based therapy: A multiple baseline study. \u003cem\u003eJournal of Contextual Behavioral Science, 35\u003c/em\u003e, 100875.\u003c/li\u003e\n\u003cli\u003ePosit Team. (2023). \u003cem\u003eRStudio: Integrated development environment for R\u003c/em\u003e. Posit Software.https://www.posit.co\u003c/li\u003e\n\u003cli\u003eR Core Team. (2024). \u003cem\u003eR: A language and environment for statistical computing\u003c/em\u003e (Version 4.3.3). R Foundation for Statistical Computing.\u003c/li\u003e\n\u003cli\u003eReichert, J., \u0026amp; Vogel, F. (2024). \u003cem\u003ePECAN2\u003c/em\u003e (R package). https://github.com/JR-psych/PECAN2\u003c/li\u003e\n\u003cli\u003eReichert, J., Klimov, M., Kachel, A., et al. (2025, May 12). Perceived causal networks in patients with fibromyalgia and depression: Construction of a structured interview and testing reliability (Version 1) [Preprint]. Research Square. https://doi.org/10.21203/rs.3.rs-6352353/v1\u003c/li\u003e\n\u003cli\u003eRubel, J. A., Fisher, A. J., Husen, K., \u0026amp; Lutz, W. (2018). Translating person-specific network models into personalized treatments: Development and demonstration of the dynamic assessment treatment algorithm for individual networks (DATA-IN). \u003cem\u003ePsychotherapy and psychosomatics\u003c/em\u003e, \u003cem\u003e87\u003c/em\u003e(4), 249-251.\u003c/li\u003e\n\u003cli\u003eScheffer, M., Bockting, C. L., Borsboom, D., Cools, R., Delecroix, C., Hartmann, J. A., \u0026hellip; Nelson, B. (2024). A dynamical systems view of psychiatric disorders\u0026mdash;Theory: A review. \u003cem\u003eJAMA Psychiatry, 81\u003c/em\u003e(6), 618\u0026ndash;623.\u003c/li\u003e\n\u003cli\u003eSiepe, B. S., Sander, C., Schultze, M., Kliem, A., Ludwig, S., Hegerl, U., \u0026amp; Reich, H. (2024). Time-varying network models for the temporal dynamics of depressive symptomatology in patients with depressive disorders. \u003cem\u003eJMIR Mental Health, 11\u003c/em\u003e, e50136.\u003c/li\u003e\n\u003cli\u003eSpitzer, R. L., Kroenke, K., Williams, J. B. W., \u0026amp; L\u0026ouml;we, B. (2006). A brief measure for assessing generalized anxiety disorder: The GAD-7. \u003cem\u003eArchives of Internal Medicine, 166\u003c/em\u003e(10), 1092\u0026ndash;1097.\u003c/li\u003e\n\u003cli\u003eTauseef, H. A., Coppersmith, D. D., Reid-Russell, A. J., Nagpal, A., Ross, J., Nock, M. K., \u0026amp; Eisenlohr-Moul, T. (2024). A call to integrate menstrual cycle influences into just-in-time adaptive interventions for suicide prevention. \u003cem\u003eFrontiers in Psychiatry, 15\u003c/em\u003e, 1434499.\u003c/li\u003e\n\u003cli\u003evan der Tuin, S., Hoekstra, R. H., Booij, S. H., Oldehinkel, A. J., Wardenaar, K. J., van den Berg, D., \u0026hellip; Wigman, J. T. (2023). Relating stability of individual dynamical networks to change in psychopathology. \u003cem\u003ePLoS ONE, 18\u003c/em\u003e(11), e0293200.\u003c/li\u003e\n\u003cli\u003eVaidyanathan, R., Xie, Y., Allaire, J., Cheng, J., Sievert, C., \u0026amp; Russell, K. (2023). \u003cem\u003ehtmlwidgets: HTML widgets for R\u003c/em\u003e (R package version 1.6.4).https://CRAN.R-project.org/package=htmlwidgets\u003c/li\u003e\n\u003cli\u003eVogel, F., Blanken, T. F., Burger, J., Reichert, J., Scholten, S., \u0026amp; Klintwall, L. (2025). How perceived causal networks can complement case conceptualization, diagnostic classification, and data-based networks. \u003cem\u003eJournal of Psychopathology and Clinical Science\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eWickham, H. (2016). \u003cem\u003eggplot2: Elegant graphics for data analysis\u003c/em\u003e. Springer.\u003c/li\u003e\n\u003cli\u003eWickham, H. (2023). \u003cem\u003estringr: Simple, consistent wrappers for common string operations\u003c/em\u003e (R package version 1.5.1).https://CRAN.R-project.org/package=stringr\u003c/li\u003e\n\u003cli\u003eWickham, H., \u0026amp; Bryan, J. (2025). \u003cem\u003ereadxl: Read Excel files\u003c/em\u003e (R package version 1.4.5).https://readxl.tidyverse.org\u003c/li\u003e\n\u003cli\u003eWickham, H., Fran\u0026ccedil;ois, R., Henry, L., M\u0026uuml;ller, K., \u0026amp; Vaughan, D. (2023). \u003cem\u003edplyr: A grammar of data manipulation\u003c/em\u003e (R package version 1.1.4).https://dplyr.tidyverse.org\u003c/li\u003e\n\u003cli\u003eWickham, H., Vaughan, D., \u0026amp; Girlich, M. (2024). \u003cem\u003etidyr: Tidy messy data\u003c/em\u003e (R package version 1.3.1).https://tidyr.tidyverse.org\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"cognitive-therapy-and-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"cotr","sideBox":"Learn more about [Cognitive Therapy and Research](http://link.springer.com/journal/10608)","snPcode":"10608","submissionUrl":"https://www.editorialmanager.com/cotr/default.aspx","title":"Cognitive Therapy and Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"case conceptualization, psychotherapy, idiographic networks, within-person dynamics, perceived causal relations, symptom fluctuations","lastPublishedDoi":"10.21203/rs.3.rs-9504631/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9504631/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003ePurpose\u003c/h2\u003e \u003cp\u003eThe network approach conceptualizes psychopathology as systems of causally interacting problems of living, more or less unstable over time. Longitudinal Perceived Causal Networks (L-PECAN) is a self-report method designed to capture day-to-day variability in patients\u0026rsquo; perceived causal relations between problems. We examined the feasibility of an L-PECAN procedure in two clinical settings and explored within-person variability in structures.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003ePatients in two psychiatric clinics first completed a training week using a generic item list. Based on this, personalized items were formulated and used for at least three additional weeks. Feasibility was evaluated via recruitment rates and adherence. Within-person variability was examined using saturation and drift analyses.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eEight patients completed the protocol. As expected, daily assessments were completed quickly, and participants largely perceived their personalized nodes as being caused by the other nodes. Some causal relations were perceived across days, but overall network structures showed substantial day-to-day variability.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eAn L-PECAN procedure, with a training week and personalized nodes, appears feasible in higher-functioning psychiatric patients and offers a structured way to examine within-person variability in perceived causal problem networks. Such variability may be clinically informative when investigating cyclical (e.g. menstrual) or context-dependent patterns. Further research is needed to evaluate validity and clinical utility.\u003c/p\u003e","manuscriptTitle":"Day-to-day variation in perceived causal networks: a feasibility study in two psychiatric settings","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-11 05:55:19","doi":"10.21203/rs.3.rs-9504631/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-19T13:10:07+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"120166395096343712002488178867619077410","date":"2026-04-30T12:36:06+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-28T13:01:20+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-25T11:01:37+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-25T11:01:05+00:00","index":"","fulltext":""},{"type":"submitted","content":"Cognitive Therapy and Research","date":"2026-04-23T09:04:48+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"cognitive-therapy-and-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"cotr","sideBox":"Learn more about [Cognitive Therapy and Research](http://link.springer.com/journal/10608)","snPcode":"10608","submissionUrl":"https://www.editorialmanager.com/cotr/default.aspx","title":"Cognitive Therapy and Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"b26f487c-8fee-4a38-b4ec-3b6284fc3be0","owner":[],"postedDate":"May 11th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-19T13:10:07+00:00","index":19,"fulltext":""},{"type":"reviewerAgreed","content":"120166395096343712002488178867619077410","date":"2026-04-30T12:36:06+00:00","index":17,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-11T05:55:20+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-11 05:55:19","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9504631","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9504631","identity":"rs-9504631","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}