Beyond prescribed activities: examining passive postural sway and gait data in patients with multiple sclerosis

Symptoms of multiple sclerosis (MS) are highly variable and include impaired senses, instability, and fatigue, making persons with MS (PwMS) ill-suited for the traditional six-month office visit paradigm. Instead, PwMS are well suited for remote monitoring to capture their true impairment. The objective of this work is to investigate the value of free-living data compared to prescribed walking tasks. Wearable sensor data were utilized from six-weeks of data from 25 PwMS. Participants completed a daily 1-minute walk, 30-second standing task, and patient reported outcomes of balance confidence (ABC), fatigue (MFIS) and walking impairment (MSWS). We compared gait and sway, as well as correlated them to patient reported outcomes (PRO). Lastly, we used a regression to determine the variance accounted for (VAF) in each PRO by different data sources. Temporal gait features were moderately correlated (r=0.6 - 0.81) between passive and prescribed walking, however, no postural sway features were correlated with each other. Passive data was found to have greater clinical relevance in our sample of PwMS compared to prescribe tasks for both gait and sway analyses. Passive sway features were found to be moderately related to ABC, MFIS, and MSWS (r=0.42-0.74, VAF=0.42-0.7), while prescribe sway was only correlated to MFIS (r = 0.41, VAF = 0.44). Both passive and prescribed measures of gait were related to ABC and MSWS; stronger relationships were found in the passive data (r = 0.42-0.78, VAF = 0.64-0.78). Additionally, we found the performance increased for passive monitoring with a shorter monitoring duration - highlighting the need to properly match the monitoring and analysis duration to the population. Overall, our findings highlight the importance of including passive free-living analysis in future studies.