Biosensors Poster Presentation

P0019 - A Comparison of Digital Gait Technologies in a Population of People with MS –When the Same Isn’t Always Exactly the Same (ID 1944)

  • M. Gudesblatt
  • O. Kaczmarek
  • J. Srinivasan
  • C. Burke
  • A. Khokhar
  • M. Zarif
  • B. Bumstead
  • D. Golan
  • M. Buhse
  • E. Ofori
  • M. Gudesblatt
Presentation Number
Presentation Topic



Multiple Sclerosis (MS), a disease characterized both by relapses and progression, commonly impacts ambulation. Impaired ambulation results in loss of independence. Current approaches to document disease impact/progression on ambulation including EDSS/T25FW are insufficiently sensitive to quantify subtle but critical change. Patient Reported Outcomes (PRO) for gait relate to several critical elements of the gait cycle beyond velocity. Earlier recognition of critical change might improve disease modifying therapy choice and timing of change. Objective multi-dimensional analytics have included digital devices of varying types utilizing different technologies (e.g. foot pressure, accelerometer, 3D video capture). Comparing different technologies in people with MS (PwMS) along a spectrum of disability would be important to optimal technology choice. Simultaneous comparisons of similar outcome measures of gait components would enhance technology choice.


To compare and contrast quantified outcome measures of the gait cycle as measured by the use of three different validated and digital ambulatory devices.


PwMS performed one pass (20 feet) while ambulating at a preferred walking speed along the Zeno™ walkway (ZW, ProtoKinetics), while wearing Opal sensors (OS, APDM), and captured using VSTBalance (VB, VirtuSense) simultaneously. Relevant gait parameters (GP) captured: velocity, stride length, total double support, and cadence. Univariate regression modeling and T-tests were used for statistical analysis for each GP.


9 PwMS (69% female, average age =53.1±11.8 years). Regression modeling showed the following relationships: velocity: ZWvsVB (r2 =0.93, p=1.2E-25), ZWvsOS (r2 =0.99, p=6.9E-40), VBvsOS (r2 =0.96, p=1.8E-25) Stride Length: ZWvsVB (r2 =0.32, p=7.8E-25), ZWvsOS (r2 =0.9, p=1.02E-16), VBvsOS (r2 =0.30, p=1.3E-4). Total double support %: ZWvsVB (r2 =0.22, p=1.6E-3), ZWvsOS (r2 =0.87, p=3.2E-20), VBvsOS (r2 =0.27, p=3.9E-4). Cadence: ZWvsVB (r2 =0.18, p=5.3E-3), ZWvsOS (r2 =0.92, p=1.2E-23), VBvsOS (r2 =0.19, p=4.2E-3). T-tests showed the following relationships: velocity: ZWvsVB (p=0.47), ZWvsOS (p=0.21), VBvsOS (p=0.63). Stride Length: ZWvsVB (p=7.25E-6), ZWvsOS (p=0.08), VBvsOS (p=0.001). Total Double Support %: ZWvsVB (p=0.91), ZWvsOS (p=0.01), VBvsOS (p=0.02). Cadence: ZWvsVB (p=5.6E-5), ZWvsOS (p=0.92), VBvsOS (p=6.3E-5).


Gait velocity had the strongest concordant relationship between all three technologies. Despite this concordance, there was still ~10% variability of this important measure. Other elements of the gait cycle had sub-optimal cross-device relationships. There was considerable discordance with stride length and cadence (ZWvsVB and OSvsVB), and double support (ZWvsOS and VBvsOS). Inconsistent relationships demonstrate the need to carefully select digital gait outcome measurement devices for PwMS.