Objective: To present a systematic framework for identifying the most sensitive, reliable, and valid digital gait measures to be used as digital endpoint in clinical trials, demonstrated through an application to Spinocerebellar Ataxia (SCA).

Background: While various digital gait measures can capture disease-specific gait characteristics, their sensitivity, reliability, and validity as clinical trial endpoints must be established.

Method: This study included 192 individuals with SCA [SCA1 (n=28), SCA2 (71), SCA3 (40), SCA6 (53)], 39 premanifest SCA, and 100 age-matched healthy controls (HC). Subjects wore 6 inertial sensors (Opal, by APDM Precision Motion-Clario) on both feet, wrists, sternum, and lumbar region. Subjects performed a 2-minute walk at their natural pace across a 10 m walkway with 180-degree turns. 75 gait measures were derived. To investigate the reliability and validity, a Multiple Criteria Decision Analysis (MCDA) approach was used, with experts weighing the importance of : 1) Discriminating SCA vs HC, 2) Premanifest SCA vs HC, 3) Fallers vs Non-fallers, 4) Reliability, 5) Correlation with EQ-5D and 6) Correlation with the total SARA score. Area Under Curve (AUC) using logistic regression was used for criteria 1 to 3. Split-half reliability (ICC 2,1) was used for criterion 4. Pearson correlation coefficients were used for criteria 5 and 6. Fallers were defined as those with ≥1 fall in the past year.

Results: The top 2 gait measures identified by MCDA were variability of foot placement: Toe Out Angle (leg external rotation)  standard deviation (SD) and Toe Off Angle (foot plantar flexion) SD. AUC discriminating SCA from HC was 0.95 and 0.94; for premanifest SCA from HC, 0.69 and 0.60; and for Fallers from Non-fallers, 0.76 and 0.75, respectively. Split-half reliability of both of these measures was very good, with an intraclass coefficient (ICC) of 0.87 and 0.91, respectively. Both measures correlated strongly with EQ-5D (r=0.45, p<0.0001; r=0.60, p<0.0001) and SARA (r=0.60, p<0.0001; r=0.66, p<0.0001).

Conclusion: The MCDA methodological framework provides a transparent and systematic approach for prioritizing statistical evidence to select, among many, potential digital gait outcomes for clinical trials. The MCDA approach shows promise to identify the best digital gait measures for SCA that are sensitive, reliable, and valid, and hence, can be used as primary/secondary endpoints for clinical trials.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/a-methodological-framework-for-selecting-digital-gait-outcomes-in-clinical-trials/