New AI-based model predicts risk of death in heart disease patients. Source: Francis Crick Institute
Researchers from the Francis Crick Institute have teamed up with the Farr Institute of Health Informatics Research and University College London Hospitals NHS Foundation Trust to create an accurate AI model that can predict the risk of death in patients who have been diagnosed with heart disease. This AI model has proven to be more accurate than doctors and other models that were developed based on expert data.
"It won't be long before doctors are routinely using these sorts of tools in the clinic to make better diagnoses and prognoses, which can help them decide the best ways to care for their patients," said Crick scientist Andrew Steele, first author of the paper.
"Doctors already use computer-based tools to work out whether a patient is at risk of heart disease, and machine-learning will allow more accurate models to be developed for a wider range of conditions."
The AI model was designed based on electronic data that is available on the Caliber program, which is home to data from over 80,000 patients. The team programmed the AI model to teach itself to predict the risk of death. The model searched for patterns and relevant variables based on a dataset of 600 patients.
After testing, the model proved to be more accurate than experts and past models that were developed based on expert data. The model even found new variables that the researchers had never consider.
"Along with factors like age and whether or not a patient smoked, our models pulled out a home visit from their GP as a good predictor of patient mortality," says Steele. "Home visits are not something a cardiologist might say is important in the biology of heart disease, but perhaps a good indication that the patient is too unwell to make it to the doctor themselves, and a useful variable to help the model make accurate predictions. Machine learning is hugely powerful tool in medicine and has the ability to revolutionize how we deliver care to patients over the next few years.”
The paper on the new AI model was published in PLOS One.
