Synthetic Hospital: Open Fully Synthetic EHR Benchmark With 1,268 Patients and Verifiable Ground Truth
Tim_Dettmers · x · 2026-09-25
SparkCPark's team released Synthetic Hospital: an open, fully synthetic longitudinal EHR benchmark with 1,268 patients and 5,602 encounters, zero PHI, shareable publicly without privacy risk.
Data quality is high enough that physicians cannot reliably distinguish its charts from real ones. Every record is fully verified with ground truth — making it ideal for RL training, as Tim Dettmers highlighted when sharing it.
For medical AI researchers, it tackles the two chronic pain points of real EHR data: privacy restrictions and missing labels. Paper, code, and blog post are all available.
Related event: Open-Source Synthetic Hospital Dataset Fooled Doctors(2 posts)→
More from Research
- Zeroing Under 0.05% of Weights Matches Full Fine-Tuning, Study Finds — agentic-learning-ai-lab · 2026-11-11
- TorchCurves: Differentiable Learnable Parametric Curves for PyTorch — Alex Shtoff · 2026-11-11
- AI-assisted proof shows no voting rule beats Borda on strategyproofness — conitzer · 2026-11-11
- Markowitz mean-variance portfolio theory applied to mixing MCMC samplers — michaelchchoi · 2026-11-11
- AI agents run 24/7 for a month to rewrite Météo-France's AROME weather model — IgorCarron · 2026-11-11
- NYU's Steertuning Lets a Robot Learn a Chore From 5 Minutes of Human Demo — CyberRobooo · 2026-11-11