Low-Cost Accelerometry-Based Posture Monitoring System for Stroke Survivors
Sonia Arteaga, Jessica Chevalier, Andrew Coile, Andrew William Hill, Serdar Sali, Sangheeta Sudhakhrisnan, Sri H. Kurniawan · 2008 · Proceedings of the 10th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '08) · doi:10.1145/1414471.1414519
Summary
This paper from the University of California Santa Cruz reports a low-cost wearable posture monitoring system for stroke survivors, costing approximately $100 for the full prototype. Stroke patients are believed to benefit from maintaining good posture during early rehabilitation, but they often spend long periods in inappropriate positions — slouching, letting the affected arm free-fall — without detection, as nurses rarely have time for deliberate posture adjustment. The system consists of ten three-axial accelerometer monitoring devices placed on each knee, each shoulder, the top of each wrist, the back of each hand (without beeper/vibrator), the back between neck and waistline, and the back of the head. Each device includes redundant warning modes: a beeper for auditory alerts, a red/green LED for visual feedback (green when calibrated correctly, red when inappropriate posture detected), and a vibrating motor — ensuring the wearer can perceive warnings even with sensory impairment in one modality. All devices connect to a waist-pack main board containing batteries, a microprocessor, interface ICs, real-time clock, and non-volatile EEPROM for logging posture violations.
Key findings
The system was calibrated per patient by having the wearer go through daily motions (sitting, standing, walking) while accelerometer values were recorded as acceptable ranges in a lookup table. The companion software provides therapists with a GUI featuring: a play function that replays logged violations on an outline human body image (flashing red at the violation location), the ability to save/import logs to patient records, and detailed statistical reports including violations per day, percentage of posture mode per day (e.g., 10% standing, 35% sitting, 20% walking, 35% lying down), violations per posture type, and violations per body location. Proof-of-concept testing with two healthy older adults wearing the system for 6 hours showed the device could detect bad postures above threshold. Participants were satisfied and noted the device helped them self-correct, though they found the beeper annoying when detecting postures that had already been corrected — leading to the sound being muted in favor of vibration, which some participants could not always feel. The system also addresses stroke-specific postures such as pusher syndrome, where patients with left or right brain damage actively push away from the non-hemiparetic side.
Relevance
This work demonstrates how low-cost sensor technology can support autonomous rehabilitation monitoring outside clinical settings, addressing a real gap in stroke aftercare. The redundant multi-modal warning approach (sound, light, vibration) is an excellent example of accessible design that accounts for the sensory impairments common in stroke survivors. For accessibility practitioners and rehabilitation technology developers, the system illustrates several important principles: calibration to individual abilities, therapist-friendly data visualization for remote monitoring, and the importance of user comfort and tolerance in wearable device design. The finding that beeper alerts were perceived as annoying highlights the tension between effective notification and user acceptance in health monitoring devices — a challenge that remains relevant as wearable health technology becomes mainstream.
Tags: stroke rehabilitation · wearable technology · accelerometer · posture monitoring · physical therapy · assistive technology · sensors · health technology