Deep learning methods for early detection of myocardial ischemia and sudden cardiac death using digital ECGs, genomic and physical activity data
This project aims to assist in the identification of persons at risk of heart attacks and sudden unexpected deaths using ECG, genetic and activity data. ECG data is a non invasive means of assessing the electrical activity of the heart and is used everyday in clinical practice in the NHS, though its interpretation is limited to visual assessment by a trained physicians. Genetic data offers an alternate view point from which we might be able to predict poor health as does the physical activity data from wrist worn devices much like a fitbit. Novel algorithms can learn from large database and identify features that are otherwise hidden, to help identify those at risk of heart attacks and death. The UK biobank has a large resource of ECG, genetic and physical activity data which we believe hold these hidden features which we hope to find using complex computer algorithms. This would allow clinicians to intervene and reduce the risk of heart attacks and death of thousands of patients in the UK annually. We expect the project to take 3 years.