JMIR Publications

ABSTRACT

Background:

Severe work fatigue impacts personal health in the long run. Prior research indicates a research opportunity of using walking parameters to indicate mental status. However, an effective and lightweight approach for inferring work fatigue remains unexplored.

Objective:

The research goals of this study are 1) to explore the ability to use walking rhythms to indicate work fatigue in the lab settings, and 2) to design and propose a work fatigue assessment mobile framework leveraging multiple measurements.

Methods:

We conducted an in-lab experiment using the smartphone to collect 3-axis accelerometer data. All participants took part in a two-day walking test before and after work time on the flat ground and the stairs. For each trial, the participants first filled a set of digital scales, conducted a Reaction Time Test (RTT) to assess cognitive fatigue level, and then took a two-phase (natural pace and fast pace) 6-minute walk test (6MWT). Overall, 216 computing instances were collected. We then grouped participants into two groups and compared their walking and reaction time performance using the Generalized Linear Model (GLM). Moreover, we collected activity diary data and conducted user interviews to interpret participants’ walking performance.

Results:

All participants (n=26) were grouped into two groups, Fatigue Sensitive Group (FSG): n1=11, 42% and Fatigue Non-Sensitive Group (FNSG): n2=15, 58% based on mental sub-scores from three entry surveys (FS-14, 3D-WFI, and FSAS). Stair climb test shows that the speed change from natural to fast pace in the non-fatigue group was 37.5% higher than that in the fatigue group (P= .03). In comparing the two groups’ walking performance, the fatigue group covered fewer steps than the non-fatigue group and had a higher step duration variability in fast pace flat ground walking (P<.001). Group membership, time, and walking pace all have an effect on walking performance. Besides, the correlation between average cadence (normal and fast pace walk) and real-time WFI mental subscore is 0.46 (P<.001).

Conclusions:

Mobile-based walking measurement is effective to infer work-related fatigue among young people. With selective scales and activity diary, the walking assessment-centered mobile application could differentiate the fatigue sensitive and fatigue non-sensitive population. This study implies future research opportunities for utilizing walking performance to track people’s work fatigue status.