A cross-cultural EMA study including digital, affective, and interpersonal markers to enhance the prediction of treatment response and dropout for patients with depression

点击翻译按钮获取中文摘要

The aim of this project is the prospective prediction of therapy success and discontinuation in patients with a depressive disorder based on pre-treatment intensive longitudinal data (Ecological Momentary Assessments; EMA). A sample of N = 300 patients receiving cognitive-behavioral therapy (CBT) at one of three clinical sites (University of Trier, UT; Justus Liebig University Gießen, JLU; and University of Buenos Aires, UBA) will be included in this study. In a development phase, a training sample (n = 150; 50%) will be used to develop a prediction model, which will be prospectively tested in a validation phase (n = 150; 50%). Before treatment start, positive and negative affect and interpersonal distress will be measured six times a day during a two-week diagnostic phase, using electronic diaries via smartphones. At the same time, data for digital phenotyping (i.e., heart rate variability, sleep duration and physical activity) will be passively measured using a fitness tracker. To assess therapy success, patients will complete the Patient Health Questionnaire-9 (PHQ-9), while discontinuation will be assessed by therapists’ reports. Based on this information, predictive models (EMA-enriched predictive models; EPM) for therapy success and discontinuation will be developed, and prospectively evaluated. The primary research question is whether complex measurements using electronic diaries and digital fitness trackers can improve predictive models based on one-time assessments (i.e., cross-sectional models; CSM) in terms of their accuracy. The CSM are based on the one-time collection of variables before the beginning of treatment. In addition, the individual site effects (UT, JLU and UBA) regarding the EPM’s accuracy will be investigated to gather evidence on the cross-cultural validity of the algorithms. This project will be the first to perform clinical individual predictions, using quantitative idiographic methods to establish patients’ affective, interpersonal, and digital phenotyping dynamics. Additionally, this study will be the first in prospectively testing the algorithms’ and investigate their cross-cultural generalizability. The results of this project will have important clinical implications for the personalization of psychotherapeutic treatments. In addition, the findings of the project will promote cross-cultural validity and replication of results enabling long-term cooperation between research groups with heterogeneous cultural backgrounds.