Identification of Swallowing Events from sEMG Signals Obtained from Healthy Adults

Abstract

Surface electromyography (sEMG) is being used with increasing frequency to identify the occurrence of swallowing, to describe swallow physiology, and to treat impaired swallowing function in dysphagic patients. Despite this increased utilization, limited information is available regarding the validity and reliability of investigators and clinicians to interpret sEMG data in reference to swallowing. This study examines the validity and interjudge reliability of swallow identification using sEMG records obtained from healthy adults. Validity and reliability estimates were compared between experienced and naïve judges in the identification of swallows from graphic sEMG records. Multiple validity estimates were high, indicating a strong degree of accuracy in identification of swallows versus nonswallow movements from sEMG traces. Experienced judges were more accurate than naïve judges (classification accuracy: experienced = 90% vs. naïve = 81%; p = 0.006, kappa: experienced = 0.89 vs. naïve 0.62; p = 0.008). Judges in both groups were more likely to classify swallows as nonswallow movements (false negatives) than to classify nonswallow movements as swallows (false positives). Interjudge reliability estimates indicated a high degree of agreement among judges in the identification of swallows versus nonswallow movements from the sEMG signal, with higher agreement among experienced judges (average kappa coefficient: experienced = 0.75 vs. naïve = 0.51). These results suggest that the sEMG graphic record is a valid and reliable tool for identifying normal swallows and that experience with this technique results in better identification and interjudge agreement.