PRISM Study - Zero-Shot Text - ICU: leaderboard
Two zero-shot text-only frontier references on MDS-ED, kept separate from ECG waveform, rendered-image and trained multimodal systems. AUROC is macro-averaged within 1,428 discharge diagnoses, six deterioration tasks, two ICU tasks and seven mortality tasks. Does not credit PRISM-trained performance to a general LLM.
Metric: AUROC (%). Source: arxiv.org. Saturation forecast: Around January 2027. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 70.7 | #150 |
| 2 | GPT-5 | 70.04 | #94 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #7 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=prism-study-zero-shot-text-icu · How It Works · Data refreshed daily, snapshot 2026-10-16.