This algorithm is almost as good as a doctor

This algorithm is almost as good as a doctor

This neural network matches the ability of cardiologists to detect a heart attack

In recent years, the group of diseases of the circulatory system remains the leading cause of death (with a rate of 267.6 deaths per 100,000 inhabitants) in Spain.

When a blockage occurs in the circulatory system and blood is prevented from reaching the heart, this problem can lead to myocardial infarction or cardiac arrest. A quick intervention or quickly tackle the first signs of cardiovascular problems are key to successfully overcome this important health problem.

Artificial intelligence and technology are allied to patients and researchers Nils Strodthoff and Claas Strodthoff. These doctors have developed a neural network capable of detecting the signs of myocardial infarction.

A project that, according to the researchers, "is capable of equalizing the performance of human cardiologists".

To collect information from their neural network, the Strodthoff researchers have monitored and monitored the electrical signals of the 12 electrocardiograph cables. These data include time intervals and the predefined or automatically detected voltage values ​​of each beat.

However, for these data to be valid, it is necessary to create a database and clean up that information and extract only what is strictly necessary.

A complex and slow task for neural networks capable of identifying patterns and transferring them to the field of medicine. The Strodthoff network has about 148 electrocardiogram (ECG) records of patients with myocardial infarction and 52 healthy controls.

"The proposed architecture overcomes the most current avant-garde approaches in this data set and achieves a similar level of performance as that of human cardiologists," the researchers point out.