Sleep Staging Using Photoplethysmography Signal and kNN Nearest Neighbor Algorithm

Abstract

Diagnosis of obstructive sleep apnoea disease is carried out through devices that use Polysomnography method. In the polysomnography method, electroencephalography, electrooculography, electromyography, oral-nasal air flow, torako-abdominal movements, oxygen saturation, electrocardiography, and body position measurements are used as standard parameters. After all these parameters are studied carefully, processes of sleep staging and respiration scoring are realized. Due to the fact that the requirements of polysomnography is too much, that it requires expert personnel, and that they are not suitable to use for houses their natural measurement environments, a new system asking for less requirements is needed. In the diagnosis of the disease, using systems with Photoplethysmography signals are projected to meet the requirements of these systems. Photoplethysmography signals can be measured from any point on skin electrooptically through a non-invasive method. This study visit photoplethysmography sleep staging phase signal of the kNN classification method performed using nearby landscapes en aimed neighborhood algorithm. In the classification method, the nearest neighbor total k = 1,2,5,10,20 value were repeated work inside. In the results obtained, sleep and wakefulness can be determined with an accuracy rate of 89.46%. The sensitivity value concerning classification was found to be 0.9205, and the specificity value was found to be 0.8719. These results show that photoplethysmography signals can be an alternative for sleep staging.