Estimation of Oxygen Saturation from PPG Signal using Smartphone Recording

In order to determine human oxygen saturation (SpO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> ) levels, this research provides an innovative approach that uses photoplethysmography (PPG) signals acquired from mobile video. Conventional methods of measuring SpO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> frequently require specialist equipment, making them less accessible and useful. The suggested method entails taking PPG signals out of RGB channels and smartphone camera frames, then pre-processing and de-noising the signals to improve their quality. The preprocessed signals are then used to extract signal properties, including peak, mean, standard deviation, AC value, and RMS value. Following that, these features are fed into a number of machine learning models, such as logistic regression, decision tree regression, random forest regression, support vector regression (SVR), and K-neighbors regression. This strategy makes use of smartphones' sophisticated image processing capabilities and integrated cameras to take advantage of their ubiquitous use. A medically approved pulse oximeter is used to monitor blood oxygen saturation levels as a baseline, providing a comparison reference point. When the five regression models are applied to a collected dataset, the Logistic Regression Model exhibits the lowest Mean Absolute Error (MAE), of about 0.845. The suggested approach, which provides a cheap and portable substitute for SpO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> monitoring, shows promise for broad adoption in healthcare settings, especially in settings with limited resources.