Predicting hypoxic hypoxia using machine learning and wearable sensors

The capability of detecting symptoms hypoxia (i.e., reduced oxygen) and other cognitive impairments in-flight with wearable sensors machine learning based algorithms will benefit the aviation community by saving lives preventing mishaps. In this study, knowledge discovery processes were implemented to build classification models predict from wearable, dry-EEG sensor data collected 85 participants in a two-phase study. Over 35-minute period while wearing flight masks which regulated their oxygen intake, would alternate between 2-minute test on CogScreen Hypoxia Edition 3-minute simulated flying task X-Plane 11, concentration reducing every 5 min following task. decrease each an increase altitude. Features extracted EEG waveforms transformed using principal component analysis reduce dimensionality data. Naïve Bayes, decision tree, random forest, neural network utilized classify brain wave as either normal or hypoxic. sensitivity ranged 0.83 1.00 specificity 0.91 1.00. This study makes step forward developing real-time, detection system.