A national-scale digital atlas of recorded care patterns across more than 1000 diagnostic categories using real-world health data: a retrospective observational study

A research team created a digital atlas mapping healthcare patterns based on data from Estonia from 2012 to 2024. The study included 1080 diagnostic categories from a total of 1645 observed categories and analyzed data from 509,856 individuals. For each diagnosis, clinical events were summarized in the period 90 days before diagnosis, 30 days after diagnosis, and 365 days after diagnosis. Of 296,303 diagnosis-concept pairs, a language model classified 24 percent as directly related, 55 percent as indirectly related, and 21 percent as noisy. Ten experts rated the atlas highly for generating research questions and providing information that is difficult to obtain through conventional methods. The atlas represents recorded care patterns in the population and supports clinical orientation and planning of observational studies.