LinguaLeap AI Swallowing Assessment: Clinical Validation Progress

LinguaLeap AI Swallowing Assessment: Clinical Validation Progress

LinguaLeap is an AI-powered swallowing assessment device developed by Carewells Limited with support from Hong Kong Science and Technology Parks Corporation (HKSTP). It uses voice biomarkers and acoustic analysis from a standard microphone to provide objective dysphagia severity scoring without requiring videofluoroscopic swallow study (VFSS) or fibreoptic endoscopic evaluation of swallowing (FEES) equipment.

The Clinical Problem LinguaLeap Solves

The gold standard for dysphagia assessment is VFSS — a radiological procedure requiring specialist equipment, radiation exposure, and trained radiologists and speech-language therapists (SLTs). FEES, the endoscopic alternative, is less invasive but still requires specialist clinical staff and controlled clinical settings.

The result is that formal dysphagia assessment is unavailable or inaccessible for the vast majority of at-risk elderly people in Hong Kong care homes. Many residents with dysphagia are never formally assessed — their dietary modifications, if any, are based on observation by care staff rather than clinical measurement.

LinguaLeap's approach is fundamentally different: using acoustic features of a swallowing attempt (wet vocal quality, cough patterns, latency) captured by a standard smartphone microphone, our AI model provides a probability score for clinically significant dysphagia and a recommended IDDSI level. The assessment takes under 3 minutes and requires no specialist equipment.

Sensitivity and Specificity Targets

LinguaLeap's current validation study is comparing our AI model outputs against FEES gold-standard assessments in a matched cohort of 80 participants at a Hong Kong hospital partner site. Our targets are sensitivity ≥0.85 and specificity ≥0.80 for detecting aspiration risk (IDDSI Level ≤4 requirement).

Interim results from the first 40 participants show sensitivity of 0.89 and specificity of 0.77, which is encouraging but requires the full cohort to confirm statistical significance. Final results are expected Q3 2026.

HKSTP Support and TechUp Y06

LinguaLeap was incubated at Hong Kong Science Park and received TechUp Y06 programme funding in 2025. HKSTP has provided laboratory access, clinical liaison support, and mentorship through its MedTech incubation stream. The TechUp Y06 grant covered early-stage prototype development and the first phase of the clinical validation study.

Regulatory Pathway: SaMD and MDCO Hong Kong

LinguaLeap is classified as a Software as a Medical Device (SaMD) under the Hong Kong Medical Device Coordination Office (MDCO) framework. We are pursuing Class II registration under the MDCO risk-based classification system, which requires submission of clinical evidence, software validation documentation, and quality management system certification (ISO 13485).

The MDCO registration process is expected to take 12–18 months following final clinical validation data submission, which we plan to complete in Q3 2026. This timeline supports our Q1 2027 commercial launch target.

Q1 2027 Launch Plan

Commercial launch is planned for Q1 2027 in Hong Kong, with initial deployment into SeniorDeli's existing network of 200+ care home partners. The launch product will be a tablet-based application used by SLTs and trained care staff to conduct LinguaLeap assessments and receive IDDSI level recommendations integrated with care plans.

A second integration track involves feeding LinguaLeap assessment results directly into the AI 照護食 Adoption Assistant — creating an end-to-end dysphagia care pathway from assessment to texture-modified meal preparation.

For clinical partnership enquiries or to participate in the validation study, contact clinical@seniordeli.com.