Part III — Rethinking Healthcare Data

Buy on Amazon

Here the vocabulary changes. Most so-called “reuse” of clinical data is actually repurposing—an active transformation from one context to another. That shift matters. Repurposing requires explicit assessment of original context, designed transformations, validation against the new purpose, and end-to-end provenance.

The part maps common scenarios: clinical-to-research (narrative to variables, cohort definition, de-identification); admin-to-quality (reinterpreting billing codes, aligning timestamps to clinical workflow); individual-to-population (aggregation, geocoding, enrichment with social data); and clinical-to-AI (labeling, bias checks, representativeness). Each demands different quality thresholds and exposes hidden biases if context is missing.

It then argues that data is a dynamic asset. Its value appreciates through use: relationships are formed, interpretations layered, quality refined, and metadata enriched. Treating data as “at rest” underestimates both risk and opportunity. Governance must therefore be evolutionary—tracking lineage, propagating corrections upstream, and reassessing value and quality as purposes change.

The takeaway is pragmatic: replace “copy once, use everywhere” with “transform deliberately, prove fitness.” Build shared transformation components, document them like code, measure semantic completeness (not just format compliance), and publish provenance with outputs. That’s how organisations move safely from siloed records to a learning system.