Design of Wearable EKG with a predictive algorithm based on ML for Telemonitoring CHD in Peru

Abstract

IM is defined as a disease that reduces blood flow in coronary arteries, generating cell death in different regions of the brain. In 20201, in Peru IM registered 22,728 deaths, while according to PAHO in 2019 it was registered that those with IM had a total of 47.9 AVD. Then, the aim of this project was to develope a predictive and classifying algorithm based on ML for presumed abnormal segments of ECG signals, which is integrated into a portable dECG for telemonitoring patients with IM. For the methodology, international and national regulatory design requirements were initially defined for the design of the algorithm, the dECG and a mobile app which main function is to connect software and hardware and to be the interface for the user and medical staff. Normal sinus signals and left and right bundle branch block were obtained from a database as test diseases for the training process of the model. The final model achieved an accuracy of 97.2\% and a loss of 0.10 during training. Subsequently, tests were carried out to evaluate its performance with an ECG signal obtained in a laboratory environment with a male participant of 21 years old in healthy conditions. Finally, an accuracy of 95.51\% was obtained, classifying it as "normal sinus rhythm" in almost its entirety, which was according to what is was expected.

Tasks

Reproductions