Each year, more than 17 million people worldwide die from the effects of cardiovascular disease, including heart attacks, blocked arteries, and strokes. Although doctors have numerous tools that help them predict a patient’s health, sometimes, it’s just not enough. Luckily, a team of scientists at the University of Nottingham, U.K., has developed an algorithm that surpasses medical experts when it comes to predicting heart attacks.
In the study, the researchers took ACC/AHA guidelines and compared them to four machine-learning algorithms: random forest, logistic regression, gradient boosting, and neural networks. All of these algorithms began to train themselves using existing medical data on cardiovascular disease and started to look for patterns associated with cardiovascular events.
As it turned out, all four algorithms performed significantly better than the ACC/AHA guidelines, but the most successful one was the neural network algorithm. This one was correct 7.6% more than the ACC/AHA method and had 1.6% fewer false positives. In practical terms, this means that in a sample of about 83,000 patient records, 355 additional lives could have been saved.
If implemented (which is highly likely), this method could save thousands of lives a year. Considering it will probably become even more accurate in time, this and similar algorithms could indeed change the face of medicine.
Source:
Digital Trends (http://www.digitaltrends.com/health-fitness/ai-algorithm-heart-attack/)
The Wonderful World of Algorithms 