Introduction

Peripheral neuropathy can arise from many different underlying medical conditions and can lead to serious functional limitations and significant long-term healthcare costs. Misdiagnoses of neuropathies are common due to similarities in patient-reported symptoms and pathologies. Instrumented assessments of motor control disruptions have shown promise as sensitive tools for the assessment of multi-system impairments arising from a wide range of medical conditions. Lumbar spinal stenosis (LSS) and diabetic peripheral neuropathy (DPN) often present similarly, so the ability to classify these impairments non-invasively and with high specificity at a low cost would be beneficial. Head-mounted, triaxial inertial measurement unit-based sensors offer a unique, non-invasive method to identify various features in physiological vibration acceleration (“phybrata”) signals [1-3] following sensory reweighting to classify various pathologies. This classification has not been examined in differing neuropathies to date. The aim of this observational pilot study is to investigate the application of phybrata sensing to differentially classify LSS from DPN and controls.

Methods

One male DPN patient (67 years, 170 cm, 75 kg), one female LSS patient (33 years, 75 kg, 165 cm), and one healthy control (CON) who presented without symptoms (age 42 years, 176 cm, 74 kg) participated in this pilot study. The phybrata sensor was attached to each participant’s right mastoid using a disposable medical adhesive. Participants were instructed to stand upright in a relaxed position with feet together and their arms at their sides during testing. Patients maintained this position twice for 20 s each test: once with eyes open (EO) and once with their eyes closed (EC). During EO, patients maintained their gaze on a marker placed at eye-level, 1.5 m in front of them. EO was always performed first. The filtered phybrata data were collected at 100 Hz. EO and EC phybrata powers, EC:EO power ratio, left-right asymmetry, time-resolved  power spectral density (PSD) distributions (0-50 Hz), and sensory reweighting profiles were compared between the three participants. Spectral shifts during each test were quantified using the mean frequency of a rolling 500 ms (250 ms overlap) analysis window.

Results

EO and EC phybrata powers were similar for LSS and the age-matched control but were both meaningfully elevated for DPN (Figure 1). The EC:EC ratio was higher in LSS compared to either DPN or CON. Left-right asymmetry was highest in the control during EO but was highest in DPN during EC. PSD distributions showed greater power at higher frequencies (12 to 15 Hz) for LSS, compared to either DPN or control. This behaviour did not change significantly for LSS during sensory reweighting.

Discussion

Changes to PSD frequencies reveal a shift to a more conservative postural control strategy triggered by LSS disruption of peripheral motor control. These preliminary data suggest that unique phybrata signatures for DPN and LSS may enable classification of different patient pathologies without the need for more costly and time-consuming diagnostic tests. Phybrata signals may also be used as a quantitative tool to longitudinally track patient responses to treatment and rehabilitation based on each patient’s unique phybrata signature.