RISK CLASSIFICATION FOR SUDDEN CARDIAC DEATH IN PATIENTS WITH HYPERTROPHIC CARDIOMYOPATHY BASED ON MACHINE LEARNING ALGORITHMS

Abstract

Hypertrophic cardiomyopathy is one of the most prominent cardiovascular diseases, with almost 1 in 500 people suffering from it. It is of great importance for this disease to be detected in a timely manner, so that patients can be provided with an adequate therapy. This is also important for monitoring the future development of the disease so that those patients under a high risk of sudden cardiac death can be provided with lifesaving implantable cardioverter-defibrillators. Regression models were created for the purpose of this paper using the random forest regression algorithm to monitor the future states of patients based on their previously known parameters. Regression models were built by maximizing R2 score for important patient parameters. The training of classification models was done using the random forest and extreme gradient boosted trees algorithms for the purposes of risk prediction. The classification models achieved 96% and 99% F1 score over the high-risk class respectively and 99% prediction accuracy overall.