Home Magazines Editors-in-Chief FAQs Contact Us

Machine learning models applied to predicate post activity oxygen saturation levels


Machine learning is a rapidly growing field with widespread application in various industries, including healthcare. In recent years, significant advancements in machine learning have enhanced our understanding of data analytics and prediction models in analyzing physiological data, such as blood oxygen levels, heart rate, and blood pressure. This paper focuses on physical activity in adolescence during the pandemic period. The study showed that going outside for a short walk can increase blood oxygen levels. Furthermore, the consumption of certain foods can also raise oxygen saturation levels. Maintaining a high blood oxygen level during exercise can improve athletic performance and reduce injury risk. However, during exercise, the blood oxygen level tends to decrease as the heart rate increases to supply more oxygen to the body. When the blood oxygen level drops too low during exercise, it can lead to fatigue, shortness of breath, and even fainting. By tracking heartbeats, blood oxygen levels, and other physiological parameters during exercise with the YAMAY Smart Watch wearable device, individuals can gain insights into their body’s response to physical activity and adjust their exercise routine accordingly. After collecting data using wearable devices and mobile apps, machine learning algorithms can be trained to predict changes in physiological parameters of post-activity, providing valuable insights into the effectiveness of different exercises and identifying potential health risks. Our study used supervised machine learning classification algorithms to predict the expected data with the target data. We used K-fold cross-validation techniques to split the dataset into training, validation, and test sets for the supervised machine learning classification algorithms. After testing each of the machine learning models (K-Nearest Neighbor (KNN), Naïve Bayes, and Random Forest), it was found that the Random Forest had the best prediction accuracy of 98.75%. On the other hand, KNN had a poor prediction accuracy, lower than 41.1%. Therefore, the Random Forest model can accurately predict the effect of change on the oxygen saturation level during exercise


machine learning classification algorithms, healthcare, prediction models, physiological parameters, physical activity in adolescents, oxygen saturation level, heartbeats