Using artificial intelligence to identify good candidates for heart surgery in the UK

The BHF-Turing Cardiovascular Data Science Awards (Second Call): Machine learning for risk prediction in adult cardiac surgery in United Kingdom (joint funding with The Alan Turing Institute)  

People who require open heart surgery must be informed about the risks related to the operation. Currently, heart surgeons in the UK calculate the risk using a mathematical model called EuroSCORE. This model is no longer considered to be accurate in the UK due to changes in patient characteristics and improvements in care. The EuroSCORE tends to overestimate the risk of death, therefore patients or surgeons may choose not to go ahead with surgery that has a good chance of a successful operation. Professor Umberto Benedetto, a heart surgeon at the University of Bristol, and Professor Chris Holmes, a statistician at the University of Oxford, will lead a team of physicians and data scientists to develop a new, highly accurate risk model for patients having heart surgery in the UK. The team will apply a type of artificial intelligence called machine learning to a large dataset of information which is routinely collected from all patients undergoing major heart surgery in the UK. Machine learning is increasingly being used in health research because it has the potential to overcome some of the limitations of traditional statistical methods. This research could lead to an improved risk model being implemented in all cardiac centres in the UK, helping surgeons identify patients who are likely to have successful heart surgery.