Computational Phenomenology of Borderline Personality Disorder: A Comparative Evaluation of LLM-Simulated Expert Personas and Human Clinical Experts
Researchers from Poland conducted a study on whether large language models (LLMs) can assist in the analysis of clinical life-story interviews with patients diagnosed with Borderline Personality Disorder. The team used three LLMs - OpenAI's GPT, Google's Gemini, and Anthropic's Claude - to evaluate their capacity for qualitative clinical analysis. The results showed that while there was a variable overlap between human and AI interpretations, the models were able to identify
Researchers from Poland conducted a study on whether large language models (LLMs) can assist in the analysis of clinical life-story interviews with patients diagnosed with Borderline Personality Disorder. The team used three LLMs - OpenAI's GPT, Google's Gemini, and Anthropic's Claude - to evaluate their capacity for qualitative clinical analysis. The results showed that while there was a variable overlap between human and AI interpretations, the models were able to identify themes that humans had missed. In some cases, the performance of one LLM, Gemini 2.5 Pro, was comparable to that of human experts.
---
Why it matters: This study matters because it explores the potential for AI to assist in clinical analysis, which could help reduce the workload and improve accuracy of mental health professionals. The results also highlight the need for further research into the limitations and biases of LLMs in this context.
Source: https://arxiv.org/abs/2508.19008
This article was originally published at: https://arxiv.org/abs/2508.19008