2026 American Academy of Neurology Abstract Website

Karen Lynch¹, Karissa Gable², Matthew P. Mavor³, Mohammad H. Akhavanfar⁴, Kristen H.E. Beange⁵, Ryan B. Graham³, Jan C. Schuller⁶, Alex Seluzhytsky¹
¹Sanofi, Cambridge, MA, USA, ²Neuromuscular Division, Department of Neurology, Duke University Medical Center, Durham, NC 27710, USA, ³School of Human Kinetics, Faculty of Health Sciences, University of Ottawa, Ottawa K1N 6N5, Ontario, Canada, ⁴School of Human Kinetics, Faculty of Health Sciences, University of Ottawa, Ottawa K1N 6N5, Ontario, Canada; Celestra Health Systems, Ottawa K2K 0G7, Ontario, Canada, ⁵Celestra Health Systems, Ottawa K2K 0G7, Ontario, Canada, ⁶Enovalife, Cœur Défense A – 110 Esplanade du Général de Gaulle – 92931 Paris La Défense Cedex, France

Objective:

To develop a digital biomarker that passively assesses gait quality in people with Chronic Inflammatory Demyelinating polyneuropathy/polyradiculoneuropathy (pwCIDP) using instrumented shoe insoles.

Background:

Biomechanical gait assessments offer an objective alternative to clinical gait evaluations. However, they rely on non-transportable technology, requiring patients to travel and providing infrequent snapshots of walking ability. Using wearable technology, such as insoles, continuous monitoring in free-living conditions becomes possible, enabling the development of digital biomarkers that comprehensively describe gait quality.

Design/Methods:

Up to two-hundred mobile pwCIDP will perform 5- to 15-minute walks 3-times per week for
12-months under free-living conditions while wearing insoles (pressure, accelerometer, gyroscope) streaming to a smartphone app (Celestra Health Systems, Canada); to date, 107 participants have been recruited. First, insole data will undergo human activity recognition; classified ‘walk’ data are segmented into standardized lengths and analyzed for spatiotemporal and sensor waveform features. Participants will undergo remote Inflammatory Neuropathy Cause and Treatment and Inflammatory Rasch-built Overall Disability scale assessments at six- and three-month intervals, respectively. Meaningful gait features will be derived by correlating with assessment scores and t-tests to identify significant differences from previously collected controls. By training a support vector machine to classify controls vs pwCIDP, a gait composite index score (gCI; 0-100%) will be developed and benchmarked against the most informative features and assessment scores. Locally estimated scatterplot smoothing regressions will be used to identify meaningful trends in gCI scores over time.

Results:

Data collection is ongoing. Trend analysis for all participants with >6 months of walking will be presented by the meeting date; a preliminary gCI will be developed and benchmarked using all recruited participants.

Conclusions:

Establishing a gCI for pwCIDP will enable longitudinal monitoring of disease progression from free-living walking bouts, reshaping how neurologists may understand and evaluate disease progression (i.e., improvement, worsening, maintenance) and treatment response (i.e., exercise, pharmacological, assistive devices).