Arrhythmia Classification Using Wearable Sensor: Machine Learning Approach

Abstract

Heart rate variability holds significant importance in assessing cardiac health.  Monitoring heart rate offers essential information regarding cardiovascular well-being, physical fitness, stress responses, and various health conditions. Additionally, categorizing heart rate patterns into normal, tachycardia, or bradycardia can support the early detection of arrhythmias. Utilizing real-time data for analysis yields more accurate and practical observations compared to relying solely on previously collected datasets. In this paper propose an interface for capturing the real heart rate data (bmp) from heart rate sensor & build machine learning model for analysis of data.SVM, Random Forest, KNN and logistic regression classifier used for classification of different types of arrhythmias. Logistic regression and random forest provides high accuracy as compared to SVM and KNN with accuracy rate 99.00% and 99.20%.Random Forest excels due to its ensemble learning capability and ability to handle non-linearity, resulting in superior performance. Similarly, Logistic Regression, being well-suited for both binary and multiclass classification, outperforms SVM and KNN.