A Multi-Sensor, Multi-Movement Exploratory Study of Motion Tape Strain Data for Low Back Pain Classification.

Area of Science:

• Biomechanics
• Rehabilitation Engineering
• Medical Sensing

Background:

• Objective assessment of low back pain (LBP) is difficult due to subtle movement impairments.
• Existing sensing technologies inadequately capture these task-dependent impairments.
• Motion Tape (MT), a skin strain sensor, offers potential for localized biomechanical measurement.

Purpose of the Study:

• To investigate the discriminative utility of MT signals for distinguishing individuals with LBP from healthy controls.
• To analyze movement-specific classification using multi-sensor MT data.
• To identify informative movements, sensor placements, and features for LBP detection.

Main Methods:

• Collected data from 20 participants (LBP and healthy controls) performing 19 functional movements with six MT sensors.
• Evaluated movement-specific classification using a leave-pair-out protocol.
• Analyzed temporal dynamics and global strain statistics, and assessed pretrained time-series foundation models.

Main Results:

• Accurate classification was highly movement-selective, with forward flexion being most effective.
• Many movements showed classification performance near chance level.
• Temporal dynamics features improved classification in difficult cases, while signals from the lower lumbar spine were most informative.
• Pretrained foundation models demonstrated limited sensitivity to participant-level structure.

Conclusions:

• MT sensing can differentiate LBP from healthy individuals, but effectiveness is contingent on specific movements and sensor placement.
• Temporal dynamics and lower lumbar spine localization are crucial for informative MT signals in LBP assessment.
• Further validation is needed for broader clinical application of MT sensing in LBP.