Validation and comparative study of the Motus system for accurately identifying movement behaviours using different...

Area of Science:

• Biomedical Engineering
• Public Health
• Wearable Technology

Background:

• Accurate movement behavior data is crucial for public health policy and interventions.
• Existing accelerometer systems like ActiPASS face challenges with slow data processing and administrative burden.
• The Motus system, utilizing SENSmotionPlus accelerometers and cloud storage, offers a scalable alternative.

Purpose of the Study:

• To validate the performance of the Motus system in identifying diverse movement behaviors.
• To assess the impact of different sampling frequencies (25 Hz vs. 12.5 Hz) on Motus's accuracy.
• To compare the performance of Motus against the established ActiPASS system.

Main Methods:

• Participants wore thigh-mounted SENSmotionPlus (25 Hz and 12.5 Hz) and AxivityX3 (25 Hz) accelerometers during laboratory and 48-hour free-living sessions.
• Laboratory sessions included video-recorded structured and semi-structured movements (sedentary, walking, running, etc.).
• Data were analyzed using Motus, with free-living data also processed by ActiPASS; F1-score, balanced accuracy, and Bland-Altman plots were used for evaluation.

Main Results:

• Motus achieved high performance (F1-score and balanced accuracy of 0.94) in the laboratory setting at both 25 Hz and 12.5 Hz.
• Comparing Motus 12.5 Hz vs. 25 Hz showed minimal bias (mean bias F1-scores ±0.01) in the laboratory.
• In free-living, Motus 25 Hz showed excellent agreement (±1 min difference) with ActiPASS; Motus 12.5 Hz also showed good agreement (±1 min) for sedentary, stair climbing, and cycling, with minor differences for other activities.

Conclusions:

• The Motus system demonstrates high feasibility and accuracy in classifying movement behaviors across different sampling frequencies.
• Reducing the sampling frequency of SENSmotionPlus accelerometers from 25 Hz to 12.5 Hz maintains high performance.
• These findings suggest that Motus can be utilized with lower sampling frequencies, potentially enhancing battery life, data transfer efficiency, and storage management for large-scale public health studies.