Prospective randomized controlled trial to develop and evaluate personalized predictions of early treatment strategies in outpatient psychotherapy for depression

• Applicants: Anne-Katharina Deisenhofer, Ph.D.; Professor Dr. Wolfgang Lutz
• Subject Area: Personality Psychology, Clinical and Medical Psychology, Methodology
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 525286173

To improve psychotherapy outcome, statistical prediction and decision models have been introduced in recent years. However, the generalizability of recent findings remains limited, as data were mainly evaluated retrospectively and, in most studies, predictions were tested in the same data in which they were trained. Furthermore, most prediction models have not been applied in a real-world clinical setting. The proposed project aims to develop a personalized digital strategy recommendation (StratRec) tool, which suggests one of two alternative treatment strategies, and test it prospectively in new cases in a real-world setting. Data from 955 patients who were already treated in a routine care outpatient center will be used to develop the StratRec tool. Treatment strategies (either problem-solving or motivation-oriented) will be defined based on ratings of a wide range of applied therapeutic interventions in video-recorded sessions. Recommendations will be developed in a training sample (n = 668) and tested in an independent holdout sample (n = 287) via machine learning (ML) algorithms. The effectiveness of these personalized recommendations will be prospectively evaluated in an RCT investigating 440 patients with a depressive disorder. More precisely, the RCT will compare the treatment outcome as well as dropout rates of patients randomly receiving either personalized treatment based on the StratRec tool (IG; n = 220), or treatment as usual (with therapists having no access to the StratRec tool, CG; n = 220). Moreover, the implementation of the recommended strategy is assumed to mediate the association between groups (IG vs. CG) and outcome/dropout. To investigate the implementation of the StratRec tool, patients from the IG and the CG will be compared regarding their likelihood to receive the recommended treatment strategy. Furthermore, the effect of providing the recommended versus the non-recommended strategy on outcome and dropout will be tested. A software to provide therapists with the StratRec tool will be developed and translated into English. Finally, this software will be provided free of charge using a GNU General Public License and made available to interested users via GitHub at the end of the project.

• DFG Programme: Research Grants