Identifying multi-resolution clusters of diseases in ten million patients with multimorbidity in primary care in England

Identifying clusters of co-occurring diseases can aid understanding of shared aetiology, management of co-morbidities, and the discovery of new disease associations. Here, we use data from a population of over ten million people with multimorbidity registered to primary care in England to identify disease clusters through a two-stage process. First, we extract data-driven representations of 212 diseases from patient records employing i) co-occurrence-based methods and ii) sequence-based natural language processing methods. Second, we apply multiscale graph-based clustering to identify clusters based on disease similarity at multiple resolutions, which outperforms k-means and hierarchical clustering in explaining known disease associations. We find that diseases display an almost-hierarchical structure across resolutions from closely to more loosely similar co-occurrence patterns and identify interpretable clusters corresponding to both established and novel patterns. Our method provides a tool for clustering diseases at different levels of resolution from co-occurrence patterns in high-dimensional electronic healthcare record data.