[2511.16625] MedBayes-Lite: Bayesian Uncertainty Quantification for Safe Clinical Decision Support

Computer Science > Artificial Intelligence arXiv:2511.16625 (cs) [Submitted on 20 Nov 2025 (v1), last revised 30 Mar 2026 (this version, v2)] Title:MedBayes-Lite: Bayesian Uncertainty Quantification for Safe Clinical Decision Support Authors:Elias Hossain, Md Mehedi Hasan Nipu, Maleeha Sheikh, Rajib Rana, Subash Neupane, Niloofar Yousefi View a PDF of the paper titled MedBayes-Lite: Bayesian Uncertainty Quantification for Safe Clinical Decision Support, by Elias Hossain and 5 other authors View PDF HTML (experimental) Abstract:We propose MedBayes-Lite, a lightweight Bayesian enhancement for transformer-based clinical language models that improves reliability through uncertainty-aware prediction. The framework operates without retraining, architectural modification, or additional trainable parameters, and integrates three components: Bayesian Embedding Calibration via Monte Carlo dropout, Uncertainty-Weighted Attention for reliability-aware token aggregation, and Confidence-Guided Decision Shaping for abstention under uncertainty. Across MedQA, PubMedQA, and MIMIC-III, MedBayes-Lite improves calibration and trustworthiness, reducing overconfidence by 32--48\%. In simulated clinical settings, it further supports safer decision-making by flagging uncertain predictions for human review, particularly under distribution shift. For closed API models, the framework remains applicable through sampling-based predictive uncertainty and confidence-guided abstention, while full embeddi...