The largest OSCE-domain improvement from multi-agent scaffolding was in communication performance¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r2 · q3 · i2–i3
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q3 i2 The multi-agent group scored higher than the control group on the OSCE communication domain (3.53 vs 2.64 on a 1–5 scale), a large effect. → Yang 2026 q3 i3 Regression associated multi-agent assignment with a 0.90-point increase in the OSCE communication score, while history-taking completeness did not reach significance. → Yang 2026 (2)

Evidence¶

Yang 2026¶

Yang, L., Liu, H., Li, S., Jia, R., Xiao, Y., Chen, G., & Lu, L. (2026). Evaluating Scaffolding-Oriented Multi-Agent Large Language Model System for Clinical Interview Training. https://arxiv.org/abs/2609.10939 doi:10.48550/arxiv.2609.10939

q3 · i2 · causal · r2

OSCE domain-score analysis of the randomized exam comparison found the most pronounced between-group difference in communication: means of 3.53 (MA) vs 2.64 (CT), "statistically significant(P = 4.14 × 10−4)", with a large effect size g = −0.79 (sign reflects coding direction).

"On the 1–5 OSCE rating scale, the mean communication score was higher in the multi-agent group than in the control group(3.53 vs. 2.64). This difference was statistically significant(P = 4.14 × 10−4) and corresponded to a large effect size(g = −0.79)"

Yang 2026 (2)¶

Yang, L., Liu, H., Li, S., Jia, R., Xiao, Y., Chen, G., & Lu, L. (2026). Evaluating Scaffolding-Oriented Multi-Agent Large Language Model System for Clinical Interview Training. https://arxiv.org/abs/2609.10939. doi:10.48550/arxiv.2609.10939

q3 · i3 · causal · r2

Regression analysis of the same randomized exam data yielded a 0.90-point communication increase (β = 0.90; 95% CI, 0.41 to 1.39; P = 4.50 × 10−4). History-taking completeness showed a weaker positive association that "did not reach conventional statistical significance(β = 0.34; 95% CI,− 0.05 to 0.73; P = 0.091)"; clinical reasoning differences were described as small.

"Assignment to the multi-agent group was associated with a 0.90-point increase in the OSCE communication score(β = 0.90; 95% CI, 0.41 to 1.39; P = 4.50 × 10−4)"

Discussion¶

Related Claims¶

• Multi-agent AI standardized patient training improved final OSCE-aligned examination scores compared with a structured non-LLM control condition — a broader claim this one bears on
• The largest checklist-item gain from multi-agent scaffolding was in observable empathic expression, with additional gains in selected history-taking items — related
• Multi-agent scaffolding showed no significant difference from the structured control in binary diagnostic accuracy or worksheet completion — related
• Controlled GenAI virtual-patient and communication studies reported short-term skill gains, but positive findings were not universal — a broader claim this one bears on
• Scaffolding level provides no consistent additional predictive information about subsequent exam performance beyond prior achievement and observed dialogue behavior — related
• MPathic-VR-trained students achieve significantly higher composite scores than computer-based learning students on a subsequent advanced communication OSCE, several days later — related