Title: Overcoming Medical Overuse with AI Assistance: An Experimental Investigation

Authors: Wang Ziyi, Wei Lijia (Corresponding Author), Xue Lian (Corresponding Author)
Journal: Journal of Health Economics, 2025, Vol. 103, 103043

A groundbreaking study led by researchers from Wuhan University reveals how artificial intelligence (AI) can effectively address medical overuse. Through two innovative experiments involving 196 practicing physicians from a Wuhan hospital and 120 medical students, the team examined AI’s impact on treatment decisions under three incentive structures (fixed, increasing, and decreasing payment models).

Key Findings:

1. AI Adoption: AI assistance increased prescription adoption rates by 25.7-28.4 percentage points, with higher acceptance observed in physicians’ less familiar medical domains.

2. Reduced Overtreatment: AI guidance decreased unnecessary medical interventions by 10.9-25.7 percentage points.

3. Improved Accuracy: Diagnostic accuracy improved by 9.8-13.3 percentage points with AI support.

4. Key Drivers: The primary factors contributing to overtreatment were identified as:

• Competence gaps (41.2%)

• Financial incentives (28.7%)

• Defensive medicine practices (18.3%)

• Patient welfare considerations (11.8%)

Methodological Innovation:
The study employed a novel medical prescription task that simulated real-world diagnostic scenarios, measuring how different compensation models affect physicians’ reliance on AI recommendations.

Policy Implications:
This research provides crucial evidence for healthcare administrators considering AI integration, demonstrating how technology can simultaneously improve care quality while reducing unnecessary treatments. The findings are particularly relevant for China’s healthcare reform efforts aimed at optimizing resource allocation.

Journal of Health Economics is classified as an A-level journal in Wuhan University’s academic journal ranking system.

Full Article:
https://www.sciencedirect.com/science/article/pii/S0167629625000785