Electrocardiogram based evaluation of children with sleep related upper airway obstruction

Sleep disordered breathing became an important health issue. It remains under diagnosed due ihe complexity and cost of the standard overnight polysomnography. The ECG signal contains importunt information concerning sleep and autonomic function. In the present study we reassessed the behavior of several HRV parameters during sleep stages. In addition we developed an algorithm, able to discriminate between different sleep stages. This classification algorithm was based on phase space of HRV parameters that have proved to be significantly different between diferent sleep stages in children with sleep related breathing disorder and in normal children. The algorithm was then validated on 21 children using the "leave-one-out" technique. The sensitivity of the classijkation was 59%-69% for different sleep stages, and its specificity between 5/%-66%. The overall accuracy was 66%, less than inter-expert agreement and better than automatic PSG analysis.