AI predicts risk of death by analyzing heart scans


Researchers at Geisinger Health System, in the United States, have developed an artificial intelligence (AI) system that can be a great ally in the treatment of serious heart problems, as it can predict the risks of death of a person by analyzing images of his echocardiogram. The novelty was presented in a study published in Nature Biomedical Engineering, on the last day 8.

According to the health institution, the research team used specialized computer hardware to train the algorithm in 812,000 videos of ultrasound examinations of the heart (echocardiogram), performed by more than 34,000 Geisinger patients over the past 10 years.

Next, the researchers used the technology to compare the tests with diagnoses and prognoses made previously by cardiologists, to verify the effectiveness of the technological model. The results showed that AI is 13% more accurate in predicting mortality in one year than doctors.

“We are thrilled to discover that machine learning can take advantage of unstructured data sets, such as medical images and videos, to improve a wide variety of clinical forecasting models,” said the assistant professor at the Department of Translational Data Science and Informatics at Geisinger Chris Haggerty, who is co-author of the study.

The importance of technology

Although far from being 100% accurate, the AI ​​developed by Geisinger can be essential in helping the work of cardiologists, due to its ability to analyze large volumes of data in a short time.

According to the institution, each echocardiogram generates around 3,000 images of the heart, which cannot always be fully interpreted by specialists, amid other types of existing data for evaluation.

But with the use of the tool to manage such information, there is an improvement in the ability to provide faster and more accurate diagnoses, ensuring a longer time for the treatment of each patient.