JMIR Publications

ABSTRACT

Background:

Medical students’ personality traits may influence their clinical diagnostic skills, but the specific pathways remain unclear, which hinders the design of targeted educational interventions.

Objective:

This study aimed to identify the key bridge nodes connecting personality traits and clinical diagnostic performance using psychological network analysis, and to validate the identified target through an AI-driven educational intervention.

Methods:

We collected Big Five personality traits (BFI-2) and seven multidimensional clinical diagnostic scores from 316 third-year preclinical medical students. A regularized Gaussian graphical model was estimated, and bridge expected influence (bridge EI) was calculated to identify nodes connecting the personality and performance subsystems. Based on the network findings, we developed an AI-driven virtual standardized patient system and tested it in a separate cohort of 202 students.

Results:

The personality–performance network showed weak cross-system connections overall. Clinical History Taking (CHT) emerged as the only positive bridge node (bridge EI=0.24), whereas Openness to Experience (OPE) showed a negative bridge effect (bridge EI=–0.12). Neuroticism and Conscientiousness had the strongest negative coupling (weight=–0.37). In the intervention cohort, AI-driven history-taking training significantly improved CHT scores (Δ=+3.8, p < 0.001) and showed spillover gains in related skills, while an unrelated skill remained unchanged.

Conclusions:

CHT is the key channel through which personality traits affect clinical diagnostic performance. A subsequent AI-driven intervention provided preliminary causal evidence supporting CHT as a valid intervention target. These findings support the development of AI-driven digital tools for personalized, scalable clinical training. Clinical Trial: The study protocol was approved by the Ethics Committee of Tangdu Hospital (Approval No. K-HG-202604-15)