SSSimon Schreynemackers

Simon Schreynemackers

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Simon Schreynemackers has a total of 25 co-authors. Here are the top co-authors by number of shared publications:

  • Hanna Reich

    Depression Research Centre of the German Depression Foundation, Department of Psychiatry, Psychosomatic Medicine and Psychotherapy, University Hospital, Goethe University, Frankfurt am Main, Germany

    9 shared publications
  • Ulrich Hegerl

    Depression Research Centre of German Foundation for Depression and Suicide Prevention, Leipzig, Germany, 49 034122387420

    9 shared publications
  • Dominik Heider

    Institute of Medical Informatics, University of Münster, Münster, Germany

    4 shared publications
  • Johannes Leimhofer

    Research Centre of the German Foundation for Depression and Suicide Prevention, Department of Psychiatry, Psychosomatic Medicine and Psychotherapy, University Hospital, Goethe University, Heinrich-Hoffmann-Straße 10, 60528, Frankfurt am Main, Germany

    4 shared publications
  • Milica Petrovic

    Affiliation Research Center of the German Foundation for Depression and Suicide Prevention, Department of Psychiatry, Psychosomatics and Psychotherapy, University Clinic, Goethe University, Heinrich-Hoffmann-Str 10, Frankfurt, Germany

    4 shared publications
  • Sascha Ludwig

    Institute for Applied Informatics, University Leipzig, Goerdelerring 9, 04109, Leipzig, Germany

    4 shared publications

Latest Publications

  • Open Access

    Individualized and Shared Learning for Predicting Daily Depressive Symptom Severity from Multimodal Smartphone and Wearable Data: A Longitudinal Evaluation of Information Sharing

    by Johannes Leimhofer, Simon Schreynemackers, Hanna Reich, Sascha Ludwig, Andreas Dominik, Dominik Heider, Ulrich Hegerl, Angela Carell

    Background and Objective: Longitudinal smartphone and wearable monitoring provides increasingly rich personal histories for predicting depressive symptom severity, while data from other individuals may offer additional information. This raises an open question: when does cross-person information add predictive value beyond an individual’s own accumulating history?Methods: Eleven participants provided up to 1 year of longitudinal data, including daily symptom assessments and 118 smartphone and wearable predictors. Elastic net (EN) and random forest (RF) models were evaluated in consecutive 30-day windows after a 90-day tuning period using expanding training histories and strict temporal cutoffs. Individualized, centrally pooled, and federated learning were compared with a participant-specific historical-mean baseline using mean absolute error and R². Controlled empirical analyses varied personal history and contributor numbers, while a complementary simulation independently varied participant count, personal history, and between-participant heterogeneity.Results: Individualized learning had lower median mean absolute error than pooled and federated learning for EN (2.147 vs 3.743 and 3.655) and RF (2.268 vs 2.748 and 3.930), with significant Holm-adjusted differences versus pooling for EN and RF and versus federated learning for RF. Individualized learning outperformed the historical-mean baseline for all 11 participants with EN and for 10 of 11 participants with RF, although median R² remained negative. A shared approach outperformed both individualized learning and the baseline in 28% and 33% of EN and RF evaluation windows. Pooling remained less accurate, with a larger disadvantage at shorter personal histories and a tendency toward greater disadvantage with more contributors. The simulation showed that either approach could be favored depending on participant count, personal history, and heterogeneity.Conclusions: Individualized learning showed an advantage in this small-participant, long-follow-up cohort, while cross-person information remained beneficial in some evaluations despite limited absolute predictive performance. Controlled analyses and the simulation showed that its value varies with personal history, contributor numbers, and between-participant heterogeneity, helping characterize when individualized or shared learning may be useful.

  • Open Access

    Individual-Level Modeling of Depressive Symptom Severity Using Smartphone and Wearable Data: Time-Aware 1-Year Study With Feature-Group Contributions

    by Simon Schreynemackers, Johannes Leimhofer, Milica Petrovic, Hanna Reich, Sascha Ludwig, Andreas Dominik, Dominik Heider, Ulrich Hegerl

    Background: Smartphones and wearables can continuously capture behavioral and physiological data in everyday life. Such mobile-sensing data may help track depressive symptoms more closely than occasional retrospective questionnaires, but prior findings have been mixed. Inconsistent findings do not preclude the presence of predictive relationships in specific individuals or time periods. In addition, it remains unclear which broader sensor domains, rather than single features, contribute most to prediction at the individual level. Objective: The objective of this study is to evaluate whether smartphone and wearable data improve prospective prediction of daily depressive symptom severity beyond a person-specific baseline and, where improvement is observed, to identify contributing sensor feature groups. Methods: Data from the MONDY (secure and open platform for AI-based health care apps) n-of-1 study were analyzed, including adults with recurrent major depressive disorder who provided up to 12 months of daily self-reported depressive symptom ratings, together with continuous smartphone and smartwatch data. To capture individual-specific patterns, separate participant-specific (idiographic) linear and nonlinear models were developed for each individual. Performance was evaluated using repeated temporally separated training and test periods against a person-specific baseline model predicting the mean symptom score from the training period. Explainability analyses were restricted to models showing improvement over baseline, and summarized at the level of sensor feature groups. Results: Among 11 participants with sufficient data, linear models improved predictive performance over the person-specific baseline in 8 (73%) participants, and nonlinear models did so in a different set of 8 (73%) participants. Improvements over the person-specific baseline, expressed as reductions in mean absolute error (change in mean absolute error relative to baseline), ranged from 0.05 to 1.23 points for linear models and from 0.12 to 1.07 points for nonlinear models on the adapted Patient Health Questionnaire-2 scale. Predictive performance varied substantially between individuals, and although nonlinear models exhibited greater aggregate feature-group contributions, this did not translate into a consistent overall predictive advantage over linear models. Feature-group importance profiles were heterogeneous, and no single sensor domain dominated across all participants. Conclusions: Multimodal smartphone and wearable data can provide individual-specific predictive signals for daily depressive symptom severity in a subset of patients when evaluated under prospective, time-aware conditions. Both predictive performance and contributing sensor domains vary markedly across individuals, and no single modeling approach is uniformly superior. These findings suggest that previously reported associations in mobile-sensing research, often arising from heterogeneous study designs and feature representations, may not consistently translate into prospective prediction, and highlight the importance of evaluation frameworks that assess predictive signal at the temporal individual-participant level.

  • Open Access

    Clinical Roles and Implementation Requirements for Purpose-Bound Intraoperative Audio in Orthopedic Surgery: Qualitative Formative Study (Preprint)

    by Moritz Lackner, Luis Wenner, Simon Schreynemackers, Atrin Pashaei Farahani, Alexander Jahnke, Markus Rickert, Bernd Ishaque

    BACKGROUND Digital technologies increasingly use intraoperative data to support documentation, workflow analysis, and clinical decision support. However, the potential contribution of audio to perioperative care is often reduced to speech transcription or unrestricted operating room recording. Orthopedic surgery is a relevant setting because documentation, procedural workflow and implant-bone interaction may all generate useful audio information. OBJECTIVE This study examined the perceived clinical roles, system requirements, and implementation conditions of purpose-bound intraoperative audio across documentation, workflow reconstruction, and mechanical monitoring to inform future perioperative technology design and validation. METHODS We conducted a qualitative formative study integrating 2 exploratory interview datasets: 6 interviews on audio-assisted documentation and workflow reconstruction and 9 interviews on acoustic monitoring during cementless femoral stem implantation as an orthopedic use case. Deductive-inductive qualitative content analysis was used to map coded meaning units to 3 functional domains and cross-layer requirements. Intercoder agreement was assessed for both focus areas. RESULTS Fifteen interviews yielded n=248 coded assignments for our developed layer framework: Documentation 76 (30.6%); Workflow Reconstruction, 42 (16.9%); Mechanical Monitoring, 89 (35.9%); and Cross-layer requirements, 41 (16.5%). The most frequently coded requirement categories were Process and Event Recognition (42/248, 16.9%), Usability (30/248, 12.1%), and Structured Reporting (24/248, 9.7%). Participants favored structured and editable reports, robust recognition of events and terminology, assistive mechanical feedback, passive operation, and purpose-limited data processing. CONCLUSIONS Purpose-bound intraoperative audio may support documentary, procedural, and mechanical functions beyond simple recording or transcription. Each function requires a separate validation strategy, while usability, privacy, workflow neutrality, interpretability, and clinician oversight remain shared implementation conditions.

  • Open Access

    Modeling Temporal Associations Between Mobile Sensing Features and Depressive Symptom Severity in Adults with Depressive Disorders: A Proof-of-Concept Study. (Preprint)

    by Johannes Leimhofer, Simon Schreynemackers, Ulrich Hegerl, Hanna Reich, Andreas Dominik, Dominik Heider, Caroline Allenhof

    BACKGROUND Continuous smartphone sensors and app data offer opportunities to detect individual behavioral patterns related to depressive symptom severity. Prior analyses showed that behavior–depression relationships can shift substantially over time, underscoring the need for transparent methods that capture such temporal variability. Vector autoregressive (VAR) models provide an interpretable and parametrized framework for analyzing temporal dynamics, with VAR-derived metrics offering a rigorous basis for identifying significant relationships. They are also computationally efficient, enabling rapid estimation in sequential or rolling-window settings. In this study, we developed and evaluated a rolling static VAR (RSVAR) approach that captures temporal fluctuations using only past information. OBJECTIVE This study evaluated (1) in how many participants and how often RSVAR models identified statistically significant day-to-day directed temporal associations between depressive symptom severity and features derived from mobile sensing (MS), and (2) which features most frequently function as temporal predictors. METHODS One-year longitudinal data from 15 adults with major depressive disorder were analyzed, combining daily self-reported depressive symptom severity assessed using an adapted PHQ-2 score with smartphone- and smartwatch-derived behavioral and physiological features. Individual-level RSVAR models were applied to identify statistically significant day-to-day directed temporal associations between MS features and depressive symptom severity. RESULTS RSVAR models identified directed temporal associations between behavioral indicators and depressive symptom severity in 13 of 15 participants. Such associations were observed on 17% of days with stable VAR models, corresponding to approximately 4% of all monitored participant days. Higher pitch was most frequently associated with subsequent increases in symptom severity (75% of identified instances), whereas higher social media app usage showed a modest predominance of symptom severity increases (57%) over decreases. Higher night-time activity was more often associated with a higher subsequent symptom severity (77%). Indicators of autonomic cardiovascular regulation exhibited a heterogeneous pattern, with symptom severity decreases occurring slightly more often (56%) than increases. Higher physical activity was more often associated with subsequent increases in symptom severity (63%) than decreases. CONCLUSIONS In this proof-of-concept study, RSVAR identified individualized, temporally directed associations between selected MS features and depressive symptom severity in a subset of monitored days. These findings support further investigation of transparent time-series approaches for personalized mobile mental health monitoring. CLINICALTRIAL German Clinical Trials Register (DRKS), DRKS00032618. Registered 12 September 2023.

  • Open Access

    Individual-Level Modeling of Depressive Symptom Severity Using Smartphone and Wearable Data: A Time-Aware 1-Year Study with Feature-Group Contributions (Preprint)

    by Simon Schreynemackers, Johannes Leimhofer, Milica Petrovic, Hanna Reich, Sascha Ludwig, Andreas Dominik, Dominik Heider, Ulrich Hegerl

    BACKGROUND Smartphones and wearables can continuously capture behavioral and physiological data in everyday life. Such mobile-sensing data may help track depressive symptoms more closely than occasional retrospective questionnaires, but prior findings have been mixed. Inconsistent findings do not preclude the presence of predictive relationships in specific individuals or time periods. In addition, it remains unclear which broader sensor domains, rather than single features, contribute most to prediction at the individual level. OBJECTIVE The objective of the current study is to evaluate whether smartphone and wearable data improve prospective prediction of daily depressive symptom severity beyond a person-specific baseline and, where improvement is observed, to identify contributing sensor feature groups. METHODS Among 11 participants with sufficient data, 8 (73%) showed improved performance over baseline with linear models and 9 (82%) with nonlinear models. Among models showing improvement, fold-averaged mean absolute error ranged from 0.91 to 2.31 points on the adapted PHQ-2 scale for linear models and from 0.54 to 2.54 points for nonlinear models. Improvements over the person-specific baseline ranged from ΔMAE = 0.04 to 1.19 points for linear models and from ΔMAE = 0.11 to 1.09 points for nonlinear models. Predictive performance varied substantially between individuals, and no consistent overall advantage of nonlinear over linear models was observed. Feature-group importance profiles were heterogeneous, and no single sensor domain dominated across all participants. RESULTS Among 11 participants with sufficient data, 8 (73%) showed improved performance over baseline with linear models and 9 (82%) with nonlinear models. Predictive performance varied substantially between individuals, and no consistent overall advantage of nonlinear over linear models was observed. Feature-group importance profiles were heterogeneous, and no single sensor domain dominated across participants. CONCLUSIONS Multimodal smartphone and wearable data can provide individual-specific predictive signals for daily depressive symptom severity in a subset of patients when evaluated under prospective, time-aware conditions. Both predictive performance and contributing sensor domains vary markedly across individuals, and no single modeling approach is uniformly superior. These findings suggest that previously reported associations in mobile sensing research, often arising from heterogeneous study designs and feature representations, may not consistently translate into prospective prediction, and highlight the importance of evaluation frameworks that assess predictive signal at the temporal individual-participant level. CLINICALTRIAL German Clinical Trials Register (DRKS), DRKS00032618. Registered 12 September 2023.


  • Open Access

    Use of Mobile Sensing Data for Longitudinal Monitoring and Prediction of Depression Severity: Systematic Review

    by Rebeka Amin, Simon Schreynemackers, Hannah Oppenheimer, Milica Petrovic, Ulrich Hegerl, Hanna Reich

    Background: Depression is highly recurrent and heterogeneous. The unobtrusive, continuous collection of mobile sensing data via smartphones and wearable devices offers a promising approach to monitor and predict individual depression trajectories, distinguish illness states, and anticipate changes in symptom severity. Objective: This systematic review evaluates whether objective data from wearable devices and smartphones can (1) monitor and distinguish different states of depression, (2) predict changes in symptom severity, and (3) identify clinically relevant objective features for tracking and forecasting depression within diagnosed individuals. Methods: We searched PubMed and Web of Science databases for English-language studies (published 2012-2022) that used smartphone or wearable device data, included participants aged ≥14 years with a depression diagnosis, and collected continuous data for at least 12 weeks. Results: Out of 12,997 peer-reviewed articles, 9 original studies met the inclusion criteria, with sample sizes ranging from 45 to 2200 and durations of 12-52 weeks. Of the 9 studies, 3 used smartphone data, 1 used wearable device data, and 5 used both data types. Commonly collected variables were step count, distance moved, smartphone usage, call logs, sleep, heart rate, light exposure, and speech patterns. One study (11%) successfully differentiated between depressive states (worsening, relapse, or recovery). Six studies (67%) showed that mobile sensing data could predict depressive episodes or symptom severity. Four studies reported the predictive accuracy for depression using mobile sensing data from smartphones and wearable devices, ranging from 81% to 91%. Higher accuracy was achieved with personalized models or multimodal data. Conclusions: Real-time passive monitoring via wearable devices and smartphones holds promise for personalized self-management, but key gaps remain, such as a lack of longitudinal and long-term studies with data collection for 1 year or longer, studies with confirmatory parameters on an individual level, and studies with a strong correlation between parameters in individual patients to support clinical decision-making. Improvements in reporting standards are highly recommended to provide better-informed insights for clinicians. Throughout this process, there is a clear need to address various other issues, such as limited types of collected data, reliability, user adherence, and privacy concerns. Trial Registration: PROSPERO CRD42022355696; https://www.crd.york.ac.uk/PROSPERO/view/CRD42022355696


  • Open Access

    Links between self-monitoring data collected through smartphones and smartwatches and the individual disease trajectories of adult patients with depressive disorders: Study protocol of a one-year observational trial

    by Hanna Reich, Simon Schreynemackers, Rebeka Amin, Sascha Ludwig, Jil Zippelius, Johannes Leimhofer, Tobias Dunker, Elisabeth Schriewer, Angela Carell, Yvonne Weber, et al.


  • Open Access

    Use of Random Forest to Predict Adherence in an Online Intervention for Depression Using Baseline and Early Usage Data: Model Development and Validation on Retrospective Routine Care Log Data

    by Franziska Wenger, Caroline Allenhof, Simon Schreynemackers, Ulrich Hegerl, Hanna Reich

    Background: Online interventions, such as the iFightDepression (iFD) tool, are increasingly recognized as effective alternatives to traditional face-to-face psychotherapy or pharmacotherapy for treating depression. However, particularly when used outside of study settings, low adherence rates and the resulting diminished benefits of the intervention can limit their effectiveness. Understanding the factors that predict adherence would allow for early, tailored interventions for individuals at risk of nonadherence, thereby enhancing user engagement and optimizing therapeutic outcomes. Objective: This study aims to develop and evaluate a random forest model that predicts adherence to the iFD tool to identify users at risk of noncompletion. The model was based on characteristics collected during baseline and the first week of the intervention in patients with depression. Methods: Log data from 4187 adult patients who registered for the iFD tool between October 1, 2016, and May 5, 2022, and provided informed consent were statistically analyzed. The resulting data set was divided into training (2932/4187, 70%) and test (1255/4187, 30%) sets using a randomly stratified split. The training data set was utilized to train a random forest model aimed at predicting each user’s adherence at baseline, based on the hypothesized predictors: age, self-reported gender, expectations of the intervention, current or previous depression treatments, confirmed diagnosis of depression, baseline 9-item Patient Health Questionnaire (PHQ-9) score, accompanying guide profession, and usage behavior within the first week. After training, the random forest model was evaluated on the test data set to assess its predictive performance. The importance of each variable in predicting adherence was analyzed using mean decrease accuracy, mean decrease Gini, and Shapley Additive Explanations values. Results: Of the 4187 patients evaluated, 1019 (24.34%) were classified as adherent based on our predefined definition. An initial random forest model that relied solely on sociodemographic and clinical predictors collected at baseline did not yield a statistically significant adherence prediction. However, after incorporating each patient’s usage behavior during the first week, we achieved a significant prediction of adherence (P<.001). Within this prediction, the model achieved an accuracy of 0.82 (95% CI 0.79-0.84), an F1-score of 0.53, an area under the curve of 0.83, and a specificity of 0.94 for predicting nonadherent users. The key predictors of adherence included logs, word count on the first workshop’s worksheet, and time spent on the tool, all measured during the first week. Conclusions: Our results highlight that early engagement, particularly usage behavior during the first week of the online intervention, is a far greater predictor of adherence than any sociodemographic or clinical factors. Therefore, analyzing usage behavior within the first week and identifying nonadherers through the algorithm could be beneficial for tailoring interventions aimed at improving user adherence. This could include follow-up calls or face-to-face discussions, optimizing resource utilization in the process.


  • Open Access

    Use of Mobile Sensing Data for Longitudinal Monitoring and Prediction of Depression Severity: Systematic Review (Preprint)

    by Rebeka Amin, Simon Schreynemackers, Hannah Oppenheimer, Milica Petrovic, Ulrich Hegerl, Hanna Reich

    BACKGROUND Depression is highly recurrent and heterogeneous. The unobtrusive, continuous collection of mobile sensing data via smartphones and wearable devices offers a promising approach to monitor and predict individual depression trajectories, distinguish illness states, and anticipate changes in symptom severity. OBJECTIVE This systematic review evaluates whether objective data from wearable devices and smartphones can (1) monitor and distinguish different states of depression, (2) predict changes in symptom severity, and (3) identify clinically relevant objective features for tracking and forecasting depression within diagnosed individuals. METHODS We searched PubMed and Web of Science databases for English-language studies (published 2012-2022) that used smartphone or wearable device data, included participants aged ≥14 years with a depression diagnosis, and collected continuous data for at least 12 weeks. RESULTS Out of 12,997 peer-reviewed articles, 9 original studies met the inclusion criteria, with sample sizes ranging from 45 to 2200 and durations of 12-52 weeks. Of the 9 studies, 3 used smartphone data, 1 used wearable device data, and 5 used both data types. Commonly collected variables were step count, distance moved, smartphone usage, call logs, sleep, heart rate, light exposure, and speech patterns. One study (11%) successfully differentiated between depressive states (worsening, relapse, or recovery). Six studies (67%) showed that mobile sensing data could predict depressive episodes or symptom severity. Four studies reported the predictive accuracy for depression using mobile sensing data from smartphones and wearable devices, ranging from 81% to 91%. Higher accuracy was achieved with personalized models or multimodal data. CONCLUSIONS Real-time passive monitoring via wearable devices and smartphones holds promise for personalized self-management, but key gaps remain, such as a lack of longitudinal and long-term studies with data collection for 1 year or longer, studies with confirmatory parameters on an individual level, and studies with a strong correlation between parameters in individual patients to support clinical decision-making. Improvements in reporting standards are highly recommended to provide better-informed insights for clinicians. Throughout this process, there is a clear need to address various other issues, such as limited types of collected data, reliability, user adherence, and privacy concerns. CLINICALTRIAL PROSPERO CRD42022355696; https://www.crd.york.ac.uk/PROSPERO/view/CRD42022355696


  • Open Access

    Use of Random Forest to Predict Adherence in an Online Intervention for Depression Using Baseline and Early Usage Data: Model Development and Validation on Retrospective Routine Care Log Data (Preprint)

    by Franziska Wenger, Caroline Allenhof, Simon Schreynemackers, Ulrich Hegerl, Hanna Reich

    BACKGROUND Online interventions, such as the iFightDepression (iFD) tool, are increasingly recognized as effective alternatives to traditional face-to-face psychotherapy or pharmacotherapy for treating depression. However, particularly when used outside of study settings, low adherence rates and the resulting diminished benefits of the intervention can limit their effectiveness. Understanding the factors that predict adherence would allow for early, tailored interventions for individuals at risk of nonadherence, thereby enhancing user engagement and optimizing therapeutic outcomes. OBJECTIVE This study aims to develop and evaluate a random forest model that predicts adherence to the iFD tool to identify users at risk of noncompletion. The model was based on characteristics collected during baseline and the first week of the intervention in patients with depression. METHODS Log data from 4187 adult patients who registered for the iFD tool between October 1, 2016, and May 5, 2022, and provided informed consent were statistically analyzed. The resulting data set was divided into training (2932/4187, 70%) and test (1255/4187, 30%) sets using a randomly stratified split. The training data set was utilized to train a random forest model aimed at predicting each user’s adherence at baseline, based on the hypothesized predictors: age, self-reported gender, expectations of the intervention, current or previous depression treatments, confirmed diagnosis of depression, baseline 9-item Patient Health Questionnaire (PHQ-9) score, accompanying guide profession, and usage behavior within the first week. After training, the random forest model was evaluated on the test data set to assess its predictive performance. The importance of each variable in predicting adherence was analyzed using mean decrease accuracy, mean decrease Gini, and Shapley Additive Explanations values. RESULTS Of the 4187 patients evaluated, 1019 (24.34%) were classified as adherent based on our predefined definition. An initial random forest model that relied solely on sociodemographic and clinical predictors collected at baseline did not yield a statistically significant adherence prediction. However, after incorporating each patient’s usage behavior during the first week, we achieved a significant prediction of adherence (P<.001). Within this prediction, the model achieved an accuracy of 0.82 (95% CI 0.79-0.84), an F1-score of 0.53, an area under the curve of 0.83, and a specificity of 0.94 for predicting nonadherent users. The key predictors of adherence included logs, word count on the first workshop’s worksheet, and time spent on the tool, all measured during the first week. CONCLUSIONS Our results highlight that early engagement, particularly usage behavior during the first week of the online intervention, is a far greater predictor of adherence than any sociodemographic or clinical factors. Therefore, analyzing usage behavior within the first week and identifying nonadherers through the algorithm could be beneficial for tailoring interventions aimed at improving user adherence. This could include follow-up calls or face-to-face discussions, optimizing resource utilization in the process.