Related Experiment Videos
Validity and Reliability Evaluation of Four Commercial Activity Trackers' Step Counting Performance
Katrine Hauer Tophøj1, Matilde Grøndahl Petersen1, Casper Sæbye2
11 Department of Engineering, Aarhus University , Aarhus, Denmark .
Summary
Consumer activity trackers show accuracy for brisk walking speeds (≥4 km/h) but are unreliable for slower rehabilitation paces (≤2 km/h). These devices are not recommended for low-speed indoor walking in patient rehabilitation settings.
Area of Science:
- Rehabilitation Medicine
- Biomedical Engineering
- Wearable Technology
Background:
- Activity trackers are increasingly used in physical rehabilitation.
- Many consumer devices lack accuracy and precision at lower walking speeds relevant to clinical practice.
- Validity and reliability are critical for selecting appropriate rehabilitation tools.
Purpose of the Study:
- To assess the validity and reliability of four consumer activity trackers.
- To evaluate device performance at clinically relevant walking speeds.
- To determine suitability for patient groups undergoing rehabilitation.
Main Methods:
- Four trackers (Fitbit Surge, Fitbit Charge HR, Microsoft Band 2, A&D 101NFC) were tested.
- Participants walked at 2, 4, 4.5, and 5.5 km/h on a treadmill.
- Measures included Mean Average Percentage Error (MAPE) and Intraclass Correlation Coefficient (ICC).
Main Results:
- Microsoft Band 2, Fitbit Surge, and A&D 101NFC showed accuracy at brisk walking speeds (≥4 km/h).
- All trackers demonstrated significant inaccuracies at lower speeds (≤2 km/h), with A&D underestimating by 53%.
- Intraclass Correlation Coefficients ranged from moderate (0.52) to excellent (0.98).
Conclusions:
- Consumer activity trackers are accurate for brisk walking but not for slower speeds common in rehabilitation.
- The evaluated devices are not suitable for monitoring patients walking at low speeds or for indoor walking.
- Clinical selection requires careful consideration of device accuracy across relevant speed ranges.