← All data
Health data
We’re on a mission to create the world’s largest consented, longitudinal health dataset, tracking how treatments work over time to power pre‑training, post‑training and reinforcement learning.
10M+Patients with longitudinal clinical records
5Linked health modalities
MultiplePartner hospital networks
2Countries, India and the US
The dataset
Longitudinal, multimodal clinical records.
Clinical records for 10M+ patients from India’s leading tertiary hospital networks.
Live records
From clinical systems, not claims or registries.
Longitudinal
Following each patient across visits, admissions and years.
Multimodal
Five linked health modalities.
EMR / EHROMOP CDM
ImagingDICOM
Pathology
Biobank
Omics
Model training
Scoped project by project.
A single tranche or the full dataset, by cohort, therapeutic area or data type.
Pre-training
Large-scale clinical records across therapeutic areas.
Post-training
Targeted patient cohorts for specific conditions or specialties.
Reinforcement learning
Clinically grounded tasks built from longitudinal patient journeys.
Trust
Consent and privacy come first.
- Consented
- Clinical records are collected with patient consent, through the partner hospitals’ own clinical systems.
- De-identified at source
- In several instances, data lakes run inside the hospital under formal agreements, so identifiable data never leaves the institution.
- Live and longitudinal
- New encounters flow in as care happens, not as periodic one-off extracts.
Next step
Request the health data brief.
Briefs for each dataset are available under NDA.