Harnessing large language models for identification and treatment of obsessive-compulsive disorder

Obsessive-Compulsive Disorder (OCD) is a mental health condition characterized by recurrent intrusive thoughts and repetitive behaviors, causing significant distress and disruption to daily life. Early identification and intervention are crucial for improving outcomes. This research compared the performance of four AI models (ChatGPT-3.5, ChatGPT-4, Claude 3.5 Sonnet, and Gemini 1.5 Pro) to that of human mental health professionals in recognizing OCD, recommending evidence-based therapies, and assessing stigma attributions. Using six vignettes, the study analyzed 480 AI evaluations and 315 assessments by mental health professionals. The AI models demonstrated superior performance in OCD recognition (100 % vs. 87 % accuracy for humans) and in recommending evidence-based interventions (60–100 % vs. 61.9 % for humans). Additionally, the AI models showed lower stigma and danger estimations. These findings indicating that AI demonstrated higher accuracy in vignette-based recognition tasks highlight the potential for AI integration into mental health care, particularly for enhancing OCD diagnosis and treatment.