Exploring the Relationship Between Heart Rate Variability and Stress Levels via Machine Learning Algorithms
HA, Jaimie Yoon
Abstract – Heart rate variability (HRV) is the variation of intervals between consecutive heartbeats. In healthy individuals, HRV increases during more calming activities and decreases in stressful conditions. This research focused on exploring how Heart rate variability (HRV) data could measure stress levels. The technology to measure HRV has evolved to instantaneous tracking in the 2010s, but the electrocardiogram (ECG) device used today is based on designs from the 1980s. An ECG works by placing electrodes on the skin, which detects strength and frequency of electrical activity of the heart. Data for this research was collected by recording ECG measurements on 25 subjects in an office setting under three working conditions: no stress, time pressure, and email interruptions. Multiple machine learning (ML) algorithms and deep learning algorithms were implemented in detecting stress levels and comparing for accuracy. Results for the 4 ML algorithms ranged from 99.8% to 100% accuracy and of the four deep learning algorithms, the most accurate was 99.2%. A potential benefit of this research in the future is to harness accurate stress level prediction and alert users of common wearable devices, such as the Apple Watch or Galaxy Watch.