"But, I don't take steps": Examining the Inaccessibility of Fitness Trackers for Wheelchair Athletes
Patrick Carrington, Kevin Chang, Helena Mentis, Amy Hurst · 2015 · ASSETS '15: Proceedings of the 17th International ACM SIGACCESS Conference on Computers & Accessibility · doi:10.1145/2700648.2809845
Summary
This paper investigates why wheelchair athletes show low adoption of wearable fitness trackers despite interest in tracking physical activity. The authors interviewed five wheelchair athletes (participating in hand cycling, rugby, basketball, and tennis) and three physical and occupational therapists, supplemented by an analysis of 131 fitness devices from the Vandrico wearables database. Despite the potential benefits—participation in adaptive sports improves quality of life outcomes for people with spinal cord injuries—none of the wheelchair users interviewed had ever used a wearable fitness device, though four of five expressed interest in tracking activity metrics like calories burned, distance, and heart rate. The fundamental problem is captured in the paper's title: fitness trackers are built around steps as the dominant unit of physical activity. "They track steps and [he does] not take steps," one participant explained. This assumption pervades the entire design of these devices and their ecosystems. The analysis of the Vandrico database revealed that 56% of fitness devices use accelerometers (commonly for step counting), 55% are wrist-worn, and the dominant metrics are steps, distance, and calories—all framed around ambulatory activity. Beyond step counting, other design assumptions create barriers. Certain sensors are physiologically incompatible: people with spinal cord injuries may not sweat below their injury level, making sweat-analysis sensors like the Electrozyme ineffective. Foot-worn or ankle-worn trackers (7 devices in the database) are obviously unusable. Device clasps like the Nike+ GPS Watch may be difficult for users with limited hand dexterity. Even heart rate—potentially useful for wheelchair users—is typically used to estimate "intensity" calibrated against walking or running baselines.
Key findings
The authors propose five thematic dimensions for evaluating wearable fitness device design: (1) Aesthetics and Device Visibility—participants preferred low-profile devices that blend with wheelchair and body, partly to avoid interference during games; (2) Sensing Hardware—existing accelerometer/gyroscope sensors could theoretically work, but need different analysis algorithms; (3) Analysis of Sensor Data—the critical gap where raw motion data is converted to "steps" rather than wheelchair-relevant metrics like push count or propulsion force; (4) Fit and Ergonomics—the wrist emerged as the preferred location for flexibility across activities, though devices designed for legs or ankles would miss upper-body activity; and (5) Presentation and Feedback—participants wanted charts and graphs for comparing data across sessions, not just raw numbers. Several concrete opportunities emerged. First, update recognition algorithms: wrist-worn accelerometer/gyroscope combinations can capture wheelchair pushing, but the analysis layer must be retrained to recognize this as valid activity rather than filtering it as noise. Second, instrument the wheelchair itself: adaptive sports use highly specialized custom equipment (rugby chairs have different designs for offensive vs. defensive players), and data loggers on wheels could capture propulsion, rotation, speed, and distance without requiring body-worn devices that might be prohibited during competition. Third, use inclusive language: presenting data as "activity" rather than "steps" would make the same underlying metrics accessible to wheelchair users. One therapist emphasized the importance of being able to share and compare data "with other people with the same level of ability, performing similar actions." Therapists expressed particular interest in using wearable tracking to verify patient compliance: "At the moment, we can suggest or prescribe that the person be active or perform different exercises but we really have no way of knowing whether they did it besides their word."
Relevance
This paper exemplifies how ableist assumptions become embedded in technology design without conscious intent. The fitness tracker industry did not deliberately exclude wheelchair users, but by centering "steps" as the universal activity metric, designers created products that are fundamentally incompatible with a large user population. The finding that only 5% of analyzed devices offer appropriate alternative outputs (audio, tactile) for users with sensory impairments compounds the exclusion. For accessibility practitioners, the paper offers a useful analytical framework: separating sensing hardware from analysis algorithms from data presentation reveals where exactly accessibility breaks down. The sensing hardware (accelerometers, gyroscopes, heart rate monitors) is largely adequate; the inaccessibility lies in how data is processed and presented. This suggests that software updates and API changes could significantly improve accessibility without requiring new hardware—a lower barrier to remediation than designing entirely new devices. The concept of instrumenting the wheelchair rather than the body is particularly innovative, leveraging the fact that adaptive sports already require specialized equipment. This approach sidesteps body placement issues, competition equipment restrictions, and the challenge of capturing upper-body activity with devices designed for legs. The broader lesson is that accessibility sometimes requires rethinking where technology lives, not just how it presents information.
Tags: wheelchair · adaptive sports · fitness tracking · wearable computing · Quantified Self · mobility impairment · spinal cord injury