AgentRx (Patient Summary, Zero-Shot) - In-Hospital Mortality: leaderboard
Clinical risk prediction by vision-language models on MIMIC-IV ICU admissions (AgentRx, Al Jorf and Shamout, 2026): from a patient summary of the history and demographics recorded before ICU admission, the model gives the probability of death in hospital. Single agent with a zero-shot prompt; AUROC on the test split, from 0 to 1, higher is better.
Metric: AUROC (0-1; in-hospital mortality; MIMIC-IV ICU test split; input: a patient summary of the history and demographics recorded before ICU admission; single agent, zero-shot prompt, predicted probability). Source: arxiv.org. Saturation forecast: Estimated already saturated. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 VL 7B Instruct | 0.67 | #1842 |
| 2 | MedGemma-4B | 0.63 | #2460 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=agentrx-patient-summary-zero-shot-in-hospital-mortality · How It Works · Data refreshed daily, snapshot 2026-11-12.