Introduction. Since its introduction in the DSM-5 and the ICD-11, the construct of personality functioning has received increased research interest. Recent studies have shown that psychotherapy contributes to an improvement in personality functioning. However, it remains unclear which factors predict an improvement. Methods. We used machine learning to filter out those variables that are relevant or irrelevant for the prediction of the improvement of personality functioning from all variables collected at the beginning of a therapy. We examined a sample of 648 completed psychotherapies from the Heidelberg Institute for Psychotherapy. Results. Overall, we found 4 groups of variables that were predictive of improvement in Personality Functioning: The patient's ability to enter relationships, his internalized relationship patterns, symptom severity, and how psychiatric the patient's disorder is. In addition, individual demographic factors and the patient's childhood memories proved to be predictive of the improvement in personality functioning. In contrast, the specific disorder pattern proved to be hardly predictive. Discussion. Our results thus reflect the experience of many therapists that for therapy to be successful, the external reality and inner world of experience should be the focus of treatment rather than the specific disorder. At the same time, our study with its many results provides a basis for future research.

R, 4.31

The dataset only contains results and our RMarkdown script. Due to restrictions of the ethical commitee we are not even allowed to publish anonymised patient data.