Letting researchers define a disease once and run it on any health database
A knowledge-driven framework turns a clinical definition of a phenotype into queries that run automatically on different data sources, tested on five Scottish health datasets.
The question
To find, say, people with type 2 diabetes in routine health data, a researcher must turn a clinical definition into queries that a particular database can run. That takes specialist skill, copes badly with messy, varied data, and is slow and error-prone. It is also hard to share or reproduce in a different setting, which weakens the credibility of data-driven studies.
What we did
We proposed a framework that separates what a phenotype means from the data it is computed on. Researchers state the meaning in familiar terms, using a formal knowledge representation of ontologies and rules, and the system generates the right queries and runs them on the data sources. It can also map definitions between standard medical vocabularies.
What we found
We report preliminary results from deploying the framework on five Scottish health datasets.
Why it matters
A phenotype defined once can be shared and re-run elsewhere, which makes studies easier to check and repeat.