Introduction
After a heart attack, patients are prescribed drugs to 'thin' the blood and to prevent further heart attacks. The benefits and hazards of such treatments are likely to differ from patient to patient and it is currently difficult to identify which individuals are at risk of either. This project aims to utilise advanced computational techniques to develop, train, and test a risk predictor that provides accurate estimates for individual patients.
One of six British Heart Foundation funded projects.
Explaining the science
In patients with myocardial infarction (heart attack), treatment with anti-platelet or anti-coagulant therapies may either prevent future cardiac events or precipitate bleeding, with the balance between benefit and harm varying widely between individuals and populations.
Current risk prediction models are very inaccurate and based on selected clinical trial populations that have substantial differences in their risk-benefit profile from real-world populations. This project aims to harness routine data from electronic health records using machine learning techniques.
The methodology involves addressing competing risks and the interaction between co-morbidities (other diseases or conditions a person might have) and outcomes, using longitudinal data to track risk over time, and using neural networks to bypass incomplete data sets.
Project aims
To use advanced computational techniques to develop, train, and test a risk predictor tool that provides accurate estimates for individual patients following a heart attack. The predictor will be trained with routinely collected electronic health records from patients with heart attacks, from a cohort of 54,000 patients from 10 hospitals in Scotland.
The predictor tool will be able to estimate the risk of major adverse cardiac events and bleeding for an individual patient. Such a tool could support doctors when prescribing drugs to ‘thin’ the blood following a heart attack; to identify patients where a short course of treatment is required to avoid unwanted severe side-effects, or those for whom prolonged treatment is truly in their best interests.
Applications
The findings of the study will influence practice across the United Kingdom, Europe and worldwide given the current uncertainty and lack of clinical evidence. For patients, this may provide us with the tools we need to guide the use of therapies following acute myocardial infarction; identifying patients where shorter durations of therapy or less potent anti-platelet drugs are required to avoid unwanted severe side-effects, or those where we can reassure them that prolonged therapy is truly in their best interests.
Organisers
Dr Nick Mills
University of EdinburghResearchers and collaborators
Professor David Newby
University of Edinburgh
Dr Catherine Stables
University of Edinburgh
