The aim of this study is to produce a prototype system within the first 12 months of the PAMBAYESIAN project that demonstrates the kind of functionality we are hoping to deliver for the real medical condition cases studies.

Hence we have developed a prototype Bayesian Network (BN) model for monitoring cardiac incidents. The model uses a combination of historical data about heart disease, background information about a patient and real-time information from the patient (via an Apple watch or similar device).

The model runs on a mobile phone that communicates with the Apple watch. After receiving inputs the Bayesian inference algorithm updates the model and determines whether an alert needs to be sent to the Apple watch. While this communication may happen many times per day, at rare intervals (typically once per week) the phone will upload the full patient data. This will enable appropriate modifications to be made to the model to ensure it is better calibrated for the particular patient.

The prototype is being tested by volunteers in the project team with known heart problems.