A comparison of the quality of integrated case formulations produced by UK psychiatric trainees and an artificial intelligence-assisted application

Abstract

AIMS AND METHOD: This study aimed to evaluate an artificial intelligence-assisted tool for psychiatric case formulation compared with human trainees. Twenty trainees and an artificial intelligence system produced formulations for three simulated psychiatric cases. Formulations were scored using the integrated case formulation scale (ICFS), assessing content, integration and total quality. Time taken was recorded, and assessor predictions of formulation origin were analysed. RESULTS: < 0.001), driven by superior content scores, while integration scores were comparable. The assessor identified artificial intelligence-generated formulations with 71.4% sensitivity, but overall accuracy of who produced the formulations was only 58.3%. CLINICAL IMPLICATIONS: Artificial intelligence shows promise as a time-saving adjunct in psychiatric training and practice, but requires improvements in generating detailed content. Optimising teaching methods for trainees and refining artificial intelligence systems can enhance the integration of artificial intelligence into clinical workflows.

Ähnliche Arbeiten