AI-Powered Holter for Affordable and Accurate Arrhythmia Detection

Cardiac arrhythmias pose significant health risks, and current detection systems often suffer from high costs and limited accessibility, particularly in resource-constrained settings. This research aimed to develop a portable, cost-effective Holter monitoring device for accurate arrhythmia detection using machine learning. By combining an inexpensive ESP32 microcontroller with an AD8232 ECG sensor, a data acquisition system was built. Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Multilayer Perceptron (MLP) models were trained and evaluated for arrhythmia classification. The SVM model achieved the highest accuracy (78.53%) using a linear kernel and features selected by a random forest algorithm. While KNN and MLP also showed promise, the results emphasized the importance of hyperparameter tuning and feature selection. This research demonstrated the feasibility of creating an affordable and intelligent Holter device capable of effective arrhythmia detection, potentially increasing access to cardiac monitoring and enabling early diagnosis in resource-limited environments.