LLM-Guided Evolution for Medical Decision Pipelines
About
Adapting large language models (LLMs) to clinical workflows often requires costly fine-tuning or manual prompt and pipeline engineering. We study LLM-guided MAP-Elites evolution as an inference-time alternative for discovering medical decision strategies and provide an implementation repository at https://github.com/univanxx/llm_guided_evo_medical. We formulate urgency triage, interactive consultation, and medical image classification as evolutionary searches over executable artifacts optimized by task-specific fitness functions. Across all three settings, evolution improves over manually designed baselines under practical constraints. In triage, evolved programs increase Semigran accuracy from $77.3\%$ to $87.1\%$ and emergency recall from $0.60$ to $0.97$, while improving safety-weighted held-out MIMIC-ESI performance. In interactive consultation, evolved policies improve the accuracy--cost frontier across Llama-3, Qwen-3.5, and Gemma-4 and transfer to held-out iCRAFTMD. In PneumoniaMNIST, prompt-only evolution improves frozen MedGemma VLMs while preserving strict JSON outputs. Qualitative analysis shows that the gains come from interpretable program-level mechanisms, calibrated triage boundaries, targeted evidence acquisition, selective commitment, and finding-oriented visual decision rules, rather than superficial prompt rewording alone.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Clinical triage evaluation | MEDIQ | Accuracy89.5 | 20 | |
| Pneumonia Classification | PneumoniaMNIST v1 (test) | Accuracy84.46 | 16 | |
| Emergency Severity Index Prediction | MIMIC-ESI (test) | Exact Accuracy62 | 11 | |
| Medical Question Answering | MediQ (test) | Accuracy75.6 | 10 | |
| Medical Triage | Semigran (test) | Accuracy87.1 | 10 | |
| Urgency Triage | Semigran (test) | Accuracy87.1 | 6 | |
| Medical Triage Prediction | Levine vignettes 2023 (held-out) | Accuracy (%)75.1 | 3 |