Modelling of Pharyngeal Laryngeal Effectiveness to Assess Swallowing Disorders

What this Trial Is About

Researchers are studying swallowing disorders by collecting signals from five non-invasive sensors (microphone, accelerometer, surface electromyography, nasal cannula, and oximeter) to identify indicators of swallowing function, airway protection, and voice quality. The study aims to develop automated methods using artificial intelligence to assess swallowing efficiency, diagnose disorders, and predict complications. This research involves 440 patients with swallowing disorders and 80 healthy volunteers across multiple centers in a prospective, multicenter, and longitudinal design. Participants with swallowing disorders undergo swallowing assessments using standard exams like videofluoroscopy or nasofibroscopy while wearing the sensors. They also perform cough and phonation tests and complete questionnaires. Healthy volunteers have a similar evaluation during a single visit. Patients are followed monthly by phone for six months to monitor complications, with a final assessment at six months either in person or by phone. Throughout the study, researchers collect and analyze sensor signals synchronized with swallowing and speech events. They use statistical and machine learning methods, including deep neural networks, to model pharyngeal-laryngeal effectiveness and predict swallowing disorder severity and complication risks. Key outcome measures include swallowing safety and efficacy, cough power, voice and velar efficiency, and quality of life assessments. The study duration spans approximately 26 months, including follow-up for patients and one visit for healthy volunteers.

Modelling of Pharyngeal Laryngeal Effectiveness