Do Fitness Trackers Accurately Count Steps in People With ALS?
A progressive neuromuscular disease is a real stress test for wrist-based step counting. The data holds up better than a blanket accuracy claim might suggest, with real limits.
This piece covers consumer smartwatch step-count accuracy specifically in people with amyotrophic lateral sclerosis (ALS), compared against a research-grade accelerometer over months of concurrent wear. It does not cover other neuromuscular or movement conditions.
A prospective observational study followed 39 people with ALS wearing a consumer smartwatch (Fitbit Sense) and a research-grade accelerometer on opposite wrists over months, comparing daily step counts on more than 500 concurrent valid days. The smartwatch and accelerometer showed a strong statistical correlation, but the smartwatch consistently undercounted, by an average of 586 steps a day, with a wide range of individual-day disagreement. The two devices tracked a similar decline in step counts over six-plus months of monitoring, meaning the smartwatch captured the overall functional trend reasonably well even though its day-to-day counts ran systematically lower than the research-grade reference.
Why ALS is a real test case for step counting
Amyotrophic lateral sclerosis is a progressive neuromuscular disease that changes gait mechanics and overall movement over time, exactly the kind of condition that could plausibly break a wrist accelerometer's step-detection algorithm, which is built around detecting a fairly standard walking motion pattern. Objective, scalable measures of function are something ALS research and care genuinely need, and if a consumer smartwatch could substitute for a research-grade accelerometer, that would meaningfully lower the barrier to long-term remote monitoring.
That's the specific practical question this study set out to answer: not step accuracy in movement disorders broadly, but a direct, sustained, head-to-head comparison in an ALS population specifically, over a genuinely long tracking period rather than a short lab session.
1 study
- In 39 participants with ALS wearing a Fitbit Sense and an ActiGraph GT9X Link accelerometer concurrently on opposite wrists, daily step counts across 503 valid concurrent days showed strong correlation (r = 0.91) between devices, but the smartwatch's mean daily count ran systematically lower, with a mean bias of -586 steps and wide 95% limits of agreement (-4193 to 3021 steps). Among participants with at least six months of concurrent data, both devices captured a similar rate of decline over time.
What a strong correlation with wide limits of agreement actually means
These are two different statistics answering two different questions, and both matter here. The strong correlation (r = 0.91) means that on days when someone walked more, both devices tended to register more steps, and on lower-activity days, both registered fewer. That's the relationship a researcher needs to track overall trends and relative activity changes over time.
The wide limits of agreement (-4193 to 3021 steps) describe something different: how much the two devices could disagree on any single given day. That's a large range for someone trying to use the smartwatch number as a precise daily step count in isolation. The practical read is that the Fitbit tracked the shape of activity change reliably across this ALS population, useful for monitoring functional decline over months, while individual daily counts carried real, sometimes substantial, disagreement with the research-grade reference.
This study used one specific consumer device (Fitbit Sense) against one specific research-grade accelerometer (ActiGraph GT9X Link) in a single-center cohort of 39 participants. Results with other consumer wearables, or in ALS populations with different disease-stage distributions, are not established by this study alone.
How this differs from the site's general movement-disorders coverage
The existing movement-disorders accuracy piece covers a broad category of conditions with a general accuracy question. This study is narrower and, in a specific way, more useful: a sustained, months-long comparison in one well-defined disease population, tracking both single-day accuracy and long-term trend fidelity separately, rather than a single-session snapshot.
For someone with ALS or a caregiver considering a consumer smartwatch as a lower-burden alternative to research-grade monitoring equipment, the honest summary from this study is that trend tracking held up meaningfully well over months, while any single day's exact step count should be read with real skepticism given how wide the disagreement range was.
Common questions
How accurate is a Fitbit's step count in people with ALS?
A 2026 study found strong statistical correlation with a research-grade accelerometer (r = 0.91), but the Fitbit undercounted by an average of 586 steps per day, with a wide range of possible disagreement on any single day (95% limits of agreement from -4193 to 3021 steps).
Can a smartwatch track ALS disease progression over time?
In participants followed for six months or more, the smartwatch and research-grade accelerometer showed a similar rate of decline in step counts over time, suggesting the trend, not just single-day counts, was captured reasonably consistently.
Is this the same finding as general movement-disorder step-count studies?
This study is specific to ALS and involved months of concurrent monitoring rather than a single testing session, which is a different and more sustained comparison than most general movement-disorder accuracy research.
Should someone with ALS trust their smartwatch's daily step count?
The evidence suggests trend tracking over weeks to months held up reasonably well, while any single day's exact count carried real potential for disagreement with a research-grade reference, given the wide range found in this study.