Dr. Gabriel Bonnin
Psychology

Psychotherapy is one of the most effective treatments for mental health problems, but its success depends on accurate diagnostic assessments. Current diagnostic tools often use standardized closed-ended scales that, while reliable, may fail to capture the complexity and individuality of mental states. In collaboration with Dr. Oscar Kjell at Lund University, this project leverages advances in artificial intelligence (AI) and natural language processing (NLP) to transform the way mental health is assessed. 
The project is based on a unique longitudinal dataset collected over 10 years at the Mental Health Research and Treatment Center (FBZ) at Ruhr University Bochum. It consists of more than 3,000 patient records, including written self-reports in which patients describe their mental health problems, functional impairment, and therapy goals in their own words. These texts are linked to key clinical outcomes such as diagnoses, symptom severity, functional impairment, and treatment success, providing a rich, ecologically valid resource for understanding patient progress in psychotherapy.

While most previous NLP-based mental health studies have focused on social media posts, limiting their clinical applicability, this project applies state-of-the-art large language models (LLMs) to real-world clinical data. By analyzing patients' open-ended responses, we aim to uncover patterns in how they articulate mental health, emotions, and treatment expectations. These insights will inform the development of AI-powered tools that offer more personalized and clinically relevant assessments, surpassing traditional methods in accuracy and depth. Ultimately, this research aims to support clinicians in making more informed treatment decisions, enhance personalized care, and contribute to the modernization of mental health assessment.