Pharyna — MSTFA Industries

~40%

Of neurological dysphagia patients aspirate silently

300K

US aspiration-pneumonia hospitalizations / yr

2 s

Classification window, on-device

Swallow classes distinguished

The Problem

Silent aspiration is invisible to patient, caregiver, and bedside exam.

When material passes below the vocal folds without triggering a cough reflex, nothing outwardly changes — until pneumonia develops. Today the only reliable detection is an in-clinic videofluoroscopic swallow study.

The consequences concentrate in neurological disease: aspiration pneumonia is the leading cause of death in ALS and carries roughly 35% mortality in stroke-related aspiration. Pharyna is designed to watch continuously, where the swallow study cannot.

Instrument

Specification.

Form factorFlexible adhesive patch, 60 × 40 mm, anterior neck, worn during waking hours — no patient interaction required

Sensor 1Surface EMG — submental and infrahyoid muscle activation during the swallow

Sensor 2Contact accelerometer / acoustic — the vibroacoustic swallow signature

Sensor 3Skin temperature — local inflammation and early infection flag

ElectronicsESP32-S3 with BLE, flexible PCB, hydrogel electrodes, 8+ hour battery, IPX4

InferenceOn-device classification over 2-second windows: normal swallow / silent aspiration / overt aspiration / non-swallow. No cloud dependency for alerting.

Pipeline

The sensing approach is established in literature.

Surface EMG (Park et al., 2020)

91.3% sensitivity and 88.7% specificity for swallow detection in stroke patients.

Neck accelerometry (Dudik et al., 2015)

89.4% accuracy for thin-liquid aspiration detection from dual-axis accelerometry.

Multimodal fusion (Lee et al., 2021)

sEMG + acoustic fusion outperforms single-sensor approaches by 8–14%.

Deep learning (Khalifa et al., 2023)

92% aspiration-detection accuracy validated against videofluoroscopy.

Development Status

Where the program stands.

StagePhase 1 clinical validation — the earliest-stage program in the portfolio.

Clinical inputInitial expert interview with a senior dysphagia specialist complete; advisory outreach in speech-language pathology in progress.

NextSensor-fusion prototype on the shared platform; classifier training against VFSS-annotated swallowing corpora.

PathwayFDA Class II 510(k), estimated 12–18 months with clinical data; predicate devices identified.

All systems → Discuss the program →