Calculating Breathing Rates from Remote PPG Signals Using Machine Learning Methods

Remote detection of health metrics, such as heart rates or breathing rates, has grown in interest. There is numerous amount of literature on calculating breathing rates using RGB cameras or thermal cameras but extracting the respiratory signal from NIR cameras is not straightforward. Thus, different ideas must be implemented. It is possible however to extract the PPG signal from an NIR camera, and from PPG signals breathing rates can be calculated. This research presents the first step in being able to predict the breathing rates from NIR cameras, which is to be able to predict the respiratory signal based on a remotely acquired PPG signal. Using machine learning, models were trained to predict respiratory signals from PPG signals. Results show that the models are predicting the respiratory signals with fair accuracy.