View asset profile
Large-scale medical NLP training corpus: 50 million de-identified electronic medical records covering chief complaint, history of present illness, past history, admission notes, discharge summaries, and related sections.
Data source and delivery: authorized data or project customization. Data from Grade A tertiary hospitals can be customized by project; the actual scope, authorization conditions, quality requirements, and delivery method are subject to compliance review and written agreement.
This information describes data governance, medical annotation, research, and model-development scenarios. It does not constitute a diagnosis, treatment, efficacy, risk-prediction, or medical-device performance claim.