Quantifying movement reserve in multiple sclerosis via diurnal activity quantiles

Conventional clinical assessments in multiple sclerosis (MS), such as the Expanded Disability Status Scale (EDSS), often miss subtle functional changes. While accelerometry provides an objective measure of real-world motor activity, most daily summaries focus on average values, neglecting both peak performance and its variability throughout the day. This diurnal peak performance variability may reflect a person's capacity to sustain high-effort activity despite fatigue, a phenomenon we term observable movement reserve.

To evaluate whether upper diurnal activity quantiles derived from accelerometry data quantify observable movement reserve, and to examine their association with EDSS-measured disability in MS.

In a cohort of 248 adults with MS (mean age 54.8 years, 71% female; EDSS range 0-6.5), continuous wrist accelerometry was collected over two weeks. We used novel scalar-on-function regression (SOFR) to compare several diurnal activity characteristics: mean, variability, and 50th-100th percentiles. SOFR models adjusted for age, sex, and BMI, and their cross-validated $R^2$ were used to assess the strength of the association between EDSS and each diurnal activity curve.

The upper diurnal activity quantiles, particularly the 95th to 100th percentiles, demonstrated the strongest association with EDSS, outperforming both diurnal mean and variability-based diurnal activity curves (cross-validated $R^2$ increased from 0.12 for the diurnal mean to 0.22 for the diurnal 100th percentile). The largest contribution to predictive power came from the late afternoon and evening hours, highlighting the importance of time-of-day in assessing disability.

Diurnal peak activity provides a sensitive, time-of-day-specific measure of observable movement reserve that closely aligns with EDSS-measured disability. This reserve fluctuates across the day, likely reflecting circadian patterns of energy and fatigue. By capturing both the timing and intensity of peak activity, the proposed metrics offer a clinically meaningful tool for monitoring functional change over time and may enhance the ability to track disease progression in MS.