Work

Modelling pharyngolaryngeal effectiveness using multimodal non-invasive sensors: study protocol for the PhLEs clinical trial

Year: 2026

Type: article

Abstract: Background Oropharyngeal dysphagia (OD) is highly prevalent in ageing and neurological populations. OD is associated with serious complications including respiratory infections, malnutrition and increased mortality. Despite its impact, current gold-standard assessments such as videofluoroscopic swallowing study (VFSS) and fibreoptic endoscopic evaluation of swallowing (FEES) are not widely accessible and require specialised expertise. Moreover, conventional evaluation focuses primarily on bolus transport during eating and does not fully integrate the broader pharyngolaryngeal functions involved in airway protection such as coughing, breathing and phonation. These limitations contribute to delays in diagnosis and insufficient clinical management. Objectives The primary objective is to model a pharyngolaryngeal effectiveness (PhLEs) multidomain composite index based on signals collected from non-invasive sensors. Secondary analyses will explore the ability of this indicator to predict severity and clinically relevant complications during follow-up. Methods and analysis This multicentre, observational, prospective and longitudinal study aims to operationalise the concept of PhLEs, which integrates four key functions of the pharyngolarynx: swallowing, breathing, airway protection and phonation. A non-invasive multisensor device incorporating a cervical microphone, accelerometer, surface electromyography, nasal thermistor and pulse oximeter will record multimodal signals during standardised clinical tasks. Reference assessments include VFSS or FEES, cough reflex testing with citric acid, phonation analysis and meal observation. A total of 520 adults will be recruited from four French university hospitals: 440 patients with suspected or confirmed OD and 80 healthy volunteers. Participants will undergo monthly follow-up for 6 months to identify the onset of OD-related complications such as pneumonia, weight loss, dehydration and psychological burden. Recorded signals will be annotated and processed using machine learning methods. These models will be developed to automate the detection of pharyngolaryngeal events, predict severity of swallowing impairment and estimate complication risk based on PhLEs. Ethics and dissemination Data, electronic case report form entries and raw signals will be centralised on a dedicated server for storage and analysis and managed by a certified healthcare data hosting provider. This study introduces a novel comprehensive approach to OD evaluation using objective multisensor data and artificial intelligence. The expected outcomes include earlier identification of swallowing disorders, improved severity stratification and enhanced prognostic evaluation, ultimately supporting timely clinical decision-making. If validated, this tool could expand access to swallowing assessments by offering a non-invasive, portable and automated solution that complements gold-standard techniques and enhances prevention strategies across common aetiologies of dysphagia. Trial registration number NCT06014710.

Source: BMJ Open

DOI: 10.1136/bmjopen-2025-114814