Medical agents improve themselves safely by learning from failures
MedRSI: Recursive Self-Improvement for Medical Agents via Clinically Aligned Self-Evolution
Artificial Intelligence
Summary
Medical software tools usually have fixed abilities decided before they are used. This paper shows how medical agents can teach themselves new skills by learning from their mistakes, but with safety checks inspired by how doctors work. The authors created MedRSI, a system that focuses on fixing important errors first and carefully tests new abilities before making them permanent. MedRSI was tested on eye and heart disease tasks and got better over time, even finding solutions not planned by its creators.
What this means in practice
- •For clinical software developers: Build medical diagnostic systems that autonomously expand their skills based on clinical error impact and evolve safely over time.
- •For healthcare ai operations teams: Deploy medical agents that continuously adapt by adding validated new tools only after proven benefit, improving diagnostic accuracy.
Authors
Junde Wu, Jiayuan Zhu, Minghao Hu, Fenglin Liu, Jiazhen Pan
Abstract
Medical agents increasingly combine general reasoning models with specialized clinical tools, yet their capabilities remain largely fixed by what clinicians and engineers design before deployment. Recursive self-improvement (RSI) offers a different paradigm in which agents learn from their own failures and autonomously expand their capabilities, but directly applying RSI to medicine introduces fundamental safety challenges. We introduce MedRSI, the first recursive self-improvement framework for medicine, which continuously transforms diagnostic failures into new clinical capabilities through tool composition and task-specific model training. Inspired by clinical practice, MedRSI introduces two mechanisms for clinically aligned self-evolution. Clinical-cost-aware failure prioritization directs improvement toward errors according to their potential clinical consequences rather than frequency alone. Fast discovery with slow registration separates rapid capability invention from conservative adoption, allowing new tools to enter the persistent agent only after demonstrating sustained benefit across subsequent patient cohorts. Across public glaucoma and heart disease benchmarks and two private clinical tasks, MedRSI progressively develops segmentation, measurement, prediction, multimodal reasoning, and generative capabilities, surpasses manually engineered medical agents, and autonomously discovers solutions to clinical problems not anticipated by its original designers. Our results show that medical agents need not remain constrained by capabilities specified before deployment: with clinically grounded mechanisms governing what to improve and what to retain, they can continuously construct, validate, and accumulate new capabilities from diagnostic experience. Code is available at https://github.com/ImprintLab/MedRSI.