多施設実臨床記録に基づく精神科診断のための大規模言語モデル：比較研究

Psychiatric disorders are diagnostically challenging and often rely on subjective clinical judgment, particularly in resource-limited settings. Large language models (LLMs) have demonstrated potential in supporting psychiatric diagnosis; however, robust evidence from large-scale, real-world clinical data remains limited.

This study aimed to evaluate and compare the diagnostic performance of multiple large language models for psychiatric disorders using multicenter real-world electronic health records.

We retrospectively analyzed 9,923 inpatient electronic health records collected from six psychiatric centers across China, encompassing all ICD-10 psychiatric categories. Three LLMs-GPT-4.0, GPT-3.5, and GLM-4-Plus-were evaluated against physician-confirmed discharge diagnoses. Diagnostic performance was assessed using strict accuracy criteria and lenient classification metrics, with subgroup analyses conducted across diagnostic categories and age groups.

GPT-4.0 achieved the highest overall strict diagnostic accuracy (71.7%) and the highest weighted F1 score under lenient evaluation (0.881), particularly for high-prevalence disorders such as mood disorders and schizophrenia spectrum disorders. Diagnostic performance varied across age groups, with the highest accuracy observed in elderly patients (up to 79.5%) and lower accuracy in adolescents. Across centers, model performance remained stable, with no significant inter-center differences.

Large language models-especially GPT-4.0-demonstrate promising capability in supporting psychiatric diagnosis using real-world electronic health records. However, diagnostic performance varies by age group and disorder category. LLMs should be regarded as assistive tools rather than replacements for clinical judgment, and further validation is needed before routine clinical implementation.