AI predicts seizures 1 hour before onset with 99.6% accuracy, study finds

A deep learning model can predict epileptic seizures up to an hour in advance with near-perfect accuracy, according to a recent study from researchers at the University of Louisiana at Lafayette.

As described in the report, published in the October 2019 issue of IEEE Transactions on Biomedical Circuits and Systems, the novel artificial intelligence system collects, analyzes and classifies brain activity using EEG tests. The resulting predictive models can then be used to identify future seizures before they occur.

In the study, researchers used long-term EEG data from 22 patients at Boston Children's Hospital; the sample size was limited by the complex model's need to be trained on each individual patient, according to the researchers. As a result, the model achieved near-perfect accuracy — a rate of 99.6 percent — and only 0.004 false alarms per hour.

Eventually, the AI model would ideally be embedded into computer chips that could be implanted in epilepsy patients' smartphones or wearable devices and synchronized with headgear embedded with sensors.

More articles on AI:
Novartis launches program with Biofourmis for AI monitoring of heart failure patients
AI to have major effects on better-paid, higher-educated workers: 5 things to know
AI not likely to replace physicians, experts say

© Copyright ASC COMMUNICATIONS 2019. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Top 40 Articles from the Past 6 Months