Researchers and computer scientists compared the risk guidelines for heart disease and stroke from the American College of Cardiology with 4 machine-learning algorithms to analyze the risk of a patient having each. The results showed that the artificially intelligent algorithms were significantly more accurate at predicting cardiovascular disease than the regular medical models.
“Current standard prediction models like the ACC are based on 8 risk factors including age, cholesterol level and blood pressure but are too simplistic to account for other factors like medications, multiple disease conditions, and other non-traditional biomarkers. These AI algorithms have the potential to help save more lives,” said Stephen Weng, a member of the University of Nottingham’s NIHR School for Primary Care Research, in a press release.
All 4 of the machine-learning algorithms were able to predict cardiovascular events with 1.7-3.6% accuracy. The highest accuracy of the neural networks algorithm had 7.6% better accuracy than regular medical models, according to the researchers.
“We were curious to find out how 4 modern machine learning algorithms would perform given the large data set of 378,256 patients from nearly 700 U.K. GP practices. Indeed, we think this is the first large-scale investigation of this type using routine clinical data,” said professor Jon Garibaldi and Jenna Reps, members of the Advanced Data Analysis Center in the School of Computer Science at the University of Nottingham.
Cardiovascular diseases affect the heart and blood vessels. Heart attacks and strokes are caused by blockages that prevent blood from flowing to the heart or brain. Nearly a third of all global deaths in 2015 were from some form of cardiovascular disease, according to the World Health Organization. Of those deaths, 6.7 million were from a stroke and 7.4 million were from coronary heart disease.
“Cardiovascular disease is the leading cause of illness and death worldwide,” Weng said. “Our study shows that artificial intelligence could significantly help in the fight against it by improving the number of patients accurately identified as being at high risk and allowing for early intervention by doctors to prevent serious events like cardiac arrest and stroke.”
The researchers plan to improve the predictive accuracy so it can be used with larger clinical datasets to predict the outcome of other diseases. They also suggest that artificial intelligence like the algorithms can help develop healthcare tools in the future that can deliver personalized medicine and individualized risk management.
[Want to stay more on top of MDO content? Subscribe to our weekly e-newsletter.]