Development and clinical validation of an orientation-sensor-integrated assistive device for drinking task retraining in stroke rehabilitation.

Background: Stroke leads to functional impairment, with 40%-80% of survivors experiencing incomplete upper extremity recovery, which impacts daily activities such as drinking. Key challenges include improper glass orientation and frequent spillage. However, very few assistive devices objectively measure glass orientation during real-time drinking task training.

Purpose: To develop and test a device to identify the orientation of drinking glass during real-world task performance.

Study design: An instrument clinical validation study.

Methods: This study was conducted in three phases. Phase 1: A tilt sensor device was developed to monitor glass orientation. Phase 2: Reliability was assessed under both static and dynamic conditions. Static testing compared device readings with goniometric angle measurements. Dynamic testing involved 84 healthy adults matching predefined tilt angles. Participants included 54 males and 30 females. Phase 3: Validity was assessed using Kinovea-based video angle analysis. The study involved 76 stroke survivors (63 males, 13 females). Average chronicity was 29 months, with varied resting pronation angles.

Results: The stand-alone tilt sensor device attachable to base of drinking glasses demonstrated excellent reliability (static condition: Pearson's correlation coefficient r = 0.99; dynamic condition: intraclass correlation = 0.97). Concurrent validity testing showed a strong positive correlation between the device-measured tilt angles and those analyzed using Kinovea software, with a mean value of 0.86, and a 95% confidence interval ranging from 0.83-0.89.

Conclusions: The tilt sensor device demonstrates superior reliability and validity, with adaptability to various drinking glasses. Its real-time, corrective graded feedback supports motor learning and offers potential for monitoring recovery progress and guiding treatment planning in stroke populations.