A Respiration Quality Analysis and Prediction Using AI and ML Algorithms

Physiological data from 550 students showed that high anxiety is linked to irregular, shallow breathing and EEG changes—increased beta and decreased alpha power. A Kriya yoga intervention (regulated pranayama and calming postures) for a subset of anxious students reduced anxiety, increased alpha activity, decreased high-beta activity, and improved respiratory signals. A Random Forest classifier distinguished high-anxiety from post-intervention states with 84% accuracy. The authors propose that combining yoga, physiological monitoring, and machine learning offers a nonpharmacological approach to mental health and emotional management.

Study at a glance

Abstract

The study looks at the relationship between physiological indicators like respiration and electroencephalogram (EEG) activity and emotional states, particularly anxiety. Physiological data were gathered from a total of 550 students, comprising respiration signals and multi-channel EEG recordings. Upon conducting the initial analysis, it was discovered that people with high anxiety levels have the tendency to show abnormal respiratory patterns; for instance, they can breathe irregularly and shallowly along with certain EEG changes that are the most common such as the increase of beta band power and the decrease of alpha band power. These were the physiological signals that the participants experienced to be with high anxiety level. A structured intervention involving Kriya yogic practices, such as regulated pranayama and specific yogic postures known for their calming and regulating effects on the autonomic nervous system, was conducted under the supervision of an expert for a subset of students who had elevated anxiety levels. Data collected after the intervention revealed a considerable drop in anxiety levels, accompanied by a rise in alpha band activity and a fall in highbeta activity. Confirming its relaxing impact, the yoga intervention also enhanced respiratory signals. After training a Random Forest classifier to identify emotional states, it was able to differentiate between high-anxiety and postintervention states with an accuracy of 84%. According to the study, the effectiveness of mind-body therapies can be increased by combining yoga practices with physiological monitoring and machine learning classification to offer a nonpharmacological approach to mental health and emotional management.