Radar-Based Automatic Detection of Sleep Apnea Using Support Vector Machine

Early diagnosis of sleep-apnea-related breathing problems helps to avoid the increased risk they can cause. In this study, we performed simultaneous radar measurements and polysomnography on patients with sleep apnea. A support vector machine algorithm was applied to the radar data to automatically detect sleep apnea events. Support vector machine parameters were optimized using the relationship between the radar and polysomnography data. The support vector machine was found to be effective in noncontact detection of central/mixed sleep apnea events using radar data. The proposed approach achieved an accuracy of 79.5%, a recall of 71.2%, and a precision of 71.2%.