An MIT team developed a machine learning model that can estimate the risk of death from heart conditions.
They have used this system called RiskCardio to establish risk patterns in patients who have suffered Coronary Syndrome. And for this, the algorithm only needs 15 minutes of reading an ECG to determine the degree of risk of the patient. That is, it does not take into account patient information or any kind of additional tests.
They trained this AI to create a risk score from the electrical activity of the heart, following the process detailed in their report. From previous studies of patients and their histories, the model created could establish patterns.
Through these patterns different levels of risk were created. In this way, he could determine whether a patient would suffer a cardiovascular death within a month, within 60 days or within a year. At the moment, the scope of this AI is limited and they expect in the future to add new datasets to apply this dynamic taking into account different genders and ages.
While it is only the beginning, the potential of this system is more than promising. It would help professionals to quickly diagnose their patients without going through long studies that steal time to receive the appropriate treatment.