Assess the feasibility of measuring FSHD progression using Emerald, a contactless wireless sensor and machine learning platform for deep phenotyping of movement disorders.

FSHD is a rare, slowly progressive myopathy characterized by weakness initially in facial, shoulder, and upper limb muscles followed by abdominal and paraspinal weakness, and finally lower extremity weakness. FSHD affects between 16,000-38,000 people in the US, and its symptoms often progress over decades, creating significant challenges in tracking disease progression using in-clinic scales and functional tests.

Emerald is an emerging, contactless radio-wave based home monitoring system that measures gait speed, time in bed, sleep stages, and vital signs.  The device infers patient physiological signals by analyzing surrounding wireless signals using machine learning. Emerald has been validated in other neuromuscular populations including Parkinson’s Disease and is validated for the first time in FSHD.

12 FSHD patients were observed in their homes 24/7 for 3 months, using two Emerald devices per home, one each in the bedroom and living room. Patient disease severity ranged from mild to severe (Clinical severity score, CSS 1-4). Devices collected over 40,000 movement episodes, which were processed to extract gait speed and time to exit the bed. These metrics were correlated with CSS and Timed-Up-and-Go (TUG) measures collected at baseline.

Preliminary analysis shows a strong relationship between in-home gait speed and CSS. Initial investigation suggests evidence of a relationship between the time to exit bed and CSS with significant precision and accuracy relative to existing clinical metrics for FSHD.

Our results suggest that key FSHD progression traits can be sensitively tracked using continuous patient monitoring in the home setting. Emerald’s ability to sit in the background, and measure continuously in a touchless manner enables long duration studies with low overhead, which is valuable in a small patient population that requires extended observation.