AI predicts risk of death from heart disease more accurately than experts

Scientists have designed a model using Artificial Intelligence that can predict risk of death in patients with coronary heart disease (CHD) better than expert-constructed models. According to a new study published in PLOS One, scientists from the Francis Crick Institute, working with University College London Hospitals NHS Foundation Trust and the Farr Institute of Health Informatics Research, developed the AI model using the data of 80,000 patients, available for researchers through UCL's CALIBER platform, which links four sources of electronic health data in England. The model that the AI one was compared to made predictions based on 27 variables chosen by medical experts, while the Crick team got their AI algorithms to train themselves, look for patterns and select the most relevant variables from a set of 600. Both machine learning and AI are picking up steam in healthcare, with hospitals testing or deploying the tech for a range of use cases from treating patients with pancreatic cancer to reducing surgical site infections while experts are saying the next generation of clinical decision support tools will include AI in workflow to improve diagnostics, imaging, radiology and pathology, among other functions. Consultancy McKinsey said last month that hospitals need a solid digital base comprising a modern infrastructure with cloud, mobile and web capabilities in place before starting down the road to AI and machine learning.