Data governance
The rules, roles and accountability that make health data trustworthy — who owns it, what 'good' means, and how quality and access are kept honest over time. Not a document; an operating discipline.
In one line
Data governance is the framework of ownership, standards, roles and accountability that keeps an organisation's data accurate, consistent, secure and usable. In health, it is what lets you trust a number enough to act on it — for a patient, a report, or an AI model.
What it actually decides
Who is the steward for each data domain; the definitions ("what counts as an active patient?"); data-quality rules and how they're measured; master data management so one patient is one identity (the Master Patient Index); and the access and use policy — who may see what, for which purpose, tied to de-identification and consent.
Why it is the foundation for analytics and AI
Every dashboard, quality measure and AI model inherits the quality of its inputs. "Garbage in, garbage out" is a governance failure, not a modelling one. Trustworthy documentation and consistent codes are what make downstream analytics defensible.
Watch for — governance as a binder on a shelf
The classic failure is a governance policy nobody operates: no named stewards, no quality metrics, no review cadence. Real governance is lightweight but live — clear owners, a few measured quality rules, and a standing forum to resolve definition disputes before they corrupt the data.