Artificial Intelligence in cardiology

According to the Ministry of Health of the Russian Federation, about 250,000 people in Russia die from cardiac arrest during a year. We have prepared an overview of AI-based solutions that may reduce this number.

Forecasts by retina

The development is by Verily. Their neural network has analyzed 284,335 images made by means of ophthalmoscopy, and accompanying data: gender, age, pressure, body mass, glycated hemoglobin and whether a person smokes or not. The neural network has learnt to forecast biological indicators of the experiment participants and identified the risk of cardiovascular diseases with 70% accuracy.

Initial processing of ambulance calls

The Corti algorithm analyzes speech and breath of the speaker and identifies on this basis heart attack during a phone call with 93.1% accuracy in 48 seconds (while human operators show 72.9% accuracy in 79 seconds). In October 2020, Moscow Emergency Medical Care Station A.S. Puchkov of The Healthcare Department of Moscow became the first to use a similar technology from Skolkovo.

Treatment efficiency prediction

In 2018, information appeared in the Anesthesiology journal about an algorithm that can predict hypotension, the abnormal lowering of the blood pressure, during an operation. The creation of the neural network required data from 1,334 patients with total volume of 545,959 minutes. The algorithm warned of hypotension in 10 and 15 minutes in 84% of cases and 5 minutes in 87% of cases.