AUTHOR=Yao Lijun , Zhao Xudong , Xu Zhiwei , Chen Yang , Liu Liang , Feng Qiang , Chen Fazhan 

TITLE=Influencing Factors and Machine Learning-Based Prediction of Side Effects in Psychotherapy

JOURNAL=Frontiers in Psychiatry

VOLUME=Volume 11 - 2020

YEAR=2020

URL=https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2020.537442

DOI=10.3389/fpsyt.2020.537442

ISSN=1664-0640

ABSTRACT=Background: Side effects in psychotherapy was a common phenomenon, and the influencing factors were complex. Unfortunately, many psychotherapists or clinicians fail to identify and manage these side effects, because the predictors of side effects in psychotherapy were not clear.
Aims: To predict whether clients or patients would experience consulting side effects during psychotherapy and to analyze the related influencing factors. 
Methods: A self-compiled "Psychotherapy Side Effects Questionnaire (PSEQ)" was delivered online by a WeChat official account. Three hundred and seventy participants who had received psychotherapy were included in the cross-sectional analysis. Consulting outcomes were classified as with side effects and without side effects. Six machine learning-based algorithms were selected and trained by our dataset to build outcome prediction models. 
Results: Among the 370 participants, 115 participants reported having experienced side effects during psychotherapy, with the incidence being 31.1%. The most common side effect was “feel bad in psychotherapy” (24.6%, 91/370). “Psychotherapist’s mental activity” (χ2 = 13.163) and “the theoretical orientation of psychotherapy” (χ2 = 9.715) were the two most recognizable features used to distinguish whether clients had consulting side effects. A Random Forest-based machine learning model offers the best prediction performance of the outcomes of psychotherapy between participants with or without side effects, with an F1-score of 0.797 and an AUC value of 0.804. 
Conclusions: Side effects experienced by clients during psychotherapy were common. Therapists or clinicians need to pay more attention to the side effects during treatment. Our Random Forest-based machine learning classifier accurately predicted the outcome of a client in psychotherapy. The classifier's prediction could facilitate the optimization of counseling strategies, thereby improving the effectiveness of psychotherapy.