Development of an Algorithm for Extracting Limb Movement Candidates Using Surface Electromyography

Periodic limb movement disorder (PLMD) is a sleep disorder characterized by repetitive involuntary limb movements (LMs) during sleep called periodic limb movements (PLMs). To adjust the volume of medication, it is important to measure the frequency of PLMs multiple nights because PLMD is likely to have day-to-day variation. However, due to the medical resource constraints, it is impractical for the multiple-night inspections by polysomnography (PSG), the gold standard for definitive diagnosis; thus, PLMs home monitoring is required. For accurate PLMs detection, it is necessary to measure the muscle activity itself by surface electromyography (sEMG), which is used in PSG. This study proposes a fourstep model for detecting PLMs from sEMG measured in daily life environments. As its first and second steps, we propose a preprocessing method and an algorithm for extracting LM candidates that does not depend on thresholds in the amplitude. In the evaluation using PSG-measured sEMG from 20 people diagnosed with PLMD in the clinical settings, our algorithm for extracting LM candidates achieved 96.7% sensitivity and 39.2% precision. Given that the final output sensitivity goal set by a physician is 85%, this performance is considered sufficient as the first and second steps of our model.Clinical relevance- This study will contribute to PLMs home monitoring using sEMG measured in daily life environments, regardless of the device used.