Actionable Analytics Linking Patient, Practitioner and Population Primary Care (360G-Wellcome-209593_Z_17_Z)

Suboptimal primary care can have substantial impacts on population health. Concurrently, burgeoning electronic health record (EHR) data are available for analysis that can inform practitioners and patients of actions they could take to improve care. However, current information systems do not: 1) efficiently generate relevant clinical actions from population-level EHR data; and 2) optimally communicate these potential actions to clinicians and patients. This Fellowship proposes to capitalise on my PhD outputs to address these evidence gaps by: 1) developing an approach to discover potential clinical actions from EHR data using machine learning and continuous evaluation to suggest and refine them in practice; 2) developing a user interface that continually optimises the presentation of clinical actions to health professionals using existing theory and randomised "A/B" testing; 3) testing these methodologies/outputs in two different geographical settings (in the UK and Canada); 4) testing the feasibility and acceptability of communicating clinical actions directly to patients using a smartphone application. Anticipated research outputs include a generic model for generating and communicating clinical actions to practitioners/patients from population-level EHR data, and software implemented in clinical practice.