Physician-scientist · Clinical AI evaluation · Global health
Suvrankar Datta
Founder and Group Lead, CRASH Lab · Simons Ashoka Early Career Fellow
Clinical AI should know when to answer—and when to hand over.
I am a radiologist and physician-scientist studying when clinical AI can be trusted, when it should defer and whether it works across Indian health systems. My work spans radiology benchmarks, multilingual tools, decision support and LMIC validation.
Email · LinkedIn · Google Scholar · CRASH Lab
Selected Work
- Radiology's Last Exam — co-first author, RadLE 1.0; co-lead, RadLE 2.0.
- PrAImaan — clinical research and technical teams supervision under PI Prof. Anurag Agrawal.
- V2DD — clinical evaluation team supervision under PI Prof. Anurag Agrawal.
- Equitable radiology foundation models — Principal Investigator.
- PREDICT-AI — Principal Investigator.
Clinical & Global Health
Public-sector and rural clinical practice shape how I evaluate medical AI across languages, populations, sites and constrained workflows.
Clinical grounding and health-system work →What counts as evidence
Capability, reliability, safety, deployment fit and impact—from retrospective evaluation through workflow testing, prospective validation and post-deployment monitoring.
Selected Publications
- MedVersa — NEJM AI, 2026.
- Radiology's Last Exam — RadLE 1.0 preprint, 2025.
- Fine-Grained Rib Fracture Diagnosis — MICCAI, 2025.
Background
- Simons Ashoka Early Career Fellow.
- Founder and Group Lead, CRASH Lab.
- Former Senior Resident, AIIMS New Delhi.
- MD, AIIMS New Delhi; MBBS, JIPMER.