Kun-Hsing Yu

Associate Professor

Harvard University
Country flagUnited States
ApplyKite TwinLearns as you apply

Make your next move.

Prepare your application

Review drafts before you send them.

Explore related searches

Articles (2)

Generative artificial intelligence (AI) is rapidly populating medical records with synthetic or partially AI-generated content, creating a feedback loop where future models are increasingly at risk of training on uncurated AI-generated data. However, the clinical consequences of this AI-generated data contamination remain unexplored. Here, we show that in the absence of mandatory human verification, this self-referential cycle drives a rapid erosion of pathological variability and diagnostic reliability of medical data at population scale. By analysing more than 800,000 synthetic data points across clinical text generation, vision–language reporting, and medical image synthesis, we find that models progressively converge toward generic phenotypes regardless of the model architectures. Specifically, rare but critical findings, including pneumothorax and effusions, vanish from the synthetic content generated by AI models, while demographic representations skew heavily toward middle-aged male phenotypes. Crucially, this degradation is masked by false diagnostic confidence. Models continue to issue reassuring reports while failing to detect life-threatening pathology, with false reassurance rates tripling to 40%. Blinded physician evaluation confirms that this decoupling of confidence and accuracy renders AI-generated documentation clinically useless after just two generations. We systematically evaluate three mitigation strategies that can be easily integrated into existing clinical workflows, finding that while synthetic volume scaling fails to prevent collapse, mixing real data with quality-aware filtering effectively preserves diversity. Ultimately, our results suggest that without policy-mandated human oversight, the deployment of generative AI threatens to degrade the very healthcare data ecosystems it relies upon.

Year:

2026

This study compared machine-learning models for predicting recurrence-free survival (RFS), disease-specific survival (DSS), and overall survival (OS) using clinicopathologic data from 1,621 stage I/II primary cutaneous melanoma patients. Our time-to-event models achieved concordance indices of 0.829 for RFS, 0.812 for DSS, and 0.778 for OS. Tumor thickness and mitotic rate were the most important predictors for RFS. Charlson comorbidity score and insurance type were critical for DSS and OS.

Year:

2025

Collaborators (4)

Ching-Yu Cheng

Professor

National University of Singapore (NUS)

SINGAPORE
View details

Qingyu Chen

Assistant Professor

National Institutes of Health

UNITED STATES
View details

Dean Ho

-

SINGAPORE
View details

Guihong Wan

Assistant Professor

University of Chicago

UNITED STATES
View details