Artificial intelligence is quickly learning how to do all of the tasks humans have traditionally been required to take, oftentimes with stunningly better accuracy and efficiency. AI is finding a beneficial home in the medical industry in every facet from diagnosis, to personalized treatment, to faster drug discovery.
Now, artificial intelligence can help predict and protect one of our most vital organs–our heart.
In an article released earlier today on Science Magazine, Matthew Hutson described a recently published study by Weng, Reps, Kai, Garibaldi, and Kureshi that outlines the beneficial effects of machine learning for heart attack prediction.
The artificial intelligence trained itself on how to analyze and predict risk factors for heart attacks and other cardiovascular events from a large sample of UK medical records from 378,256 patients.
The guidelines that doctors use to predict heart attacks which include expected checklist items such as cholesterol and blood pressure proved less effective than the artificial intelligence which incorporated extra data points as varied as arthritis.
Of all the machine learning algorithms the artificial intelligence used to train itself, neural networks proved most effective and raised few false alarms than traditional, human prediction methods.
Weng et al. are not the only ones with evidence for the effectiveness of AI to predict heart attacks; last year, Nasrabadi and Haddadnia published this study on the same concept.
Of course, more accurate predictions can lead to early preventive measures which may lower the likelihood of a heart attack.
We may see more doctors adopt machine learning to assist in their diagnoses and predictions, just as we have seen drug discovery companies adopt artificial intelligence.
Not only can artificial intelligence predict heart attacks, but studies in January 2017 also point to the fact that machine learning can be used to predict heart failure as well.
In an article in RSNA’s Radiology journal, researchers describe an artificial intelligence that uses 30k data points of a patient’s heart to create a 3D scan which was then combined with eight years of prior patient medical records.
The patients in the study all had been diagnosed with pulmonary hypertension by a human doctor.
The combination of collected data and virtual heart model allowed the AI to make a prediction on the patient’s death with an accuracy of 80% for deaths within the next year, although the predictions spanned five years into the future.
We can expect to see increasing artificial intelligence input in most industries, but especially the health industry. Alongside big data and increasingly complex and autonomous learning algorithms, AI is quickly growing in its capabilities especially in terms of prediction.
Add these disruptive new technologies to the sci-fi films-turned-reality (like this real-life Tricorder called DxtER and these mice whose aging was reversed), and the future of healthcare looks brighter every day.