Development and clinical application of a computer-aided real-time feedback system for detecting in-bed physical activities

Liang Hsuan Lu, Shang Lin Chiang, Shun Hwa Wei, Chueh Ho Lin, Wen Hsu Sung

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Background and objective Being bedridden long-term can cause deterioration in patients’ physiological function and performance, limiting daily activities and increasing the incidence of falls and other accidental injuries. Little research has been carried out in designing effective detecting systems to monitor the posture and status of bedridden patients and to provide accurate real-time feedback on posture. The purposes of this research were to develop a computer-aided system for real-time detection of physical activities in bed and to validate the system's validity and test–retest reliability in determining eight postures: motion leftward/rightward, turning over leftward/rightward, getting up leftward/rightward, and getting off the bed leftward/rightward. Methods The in-bed physical activity detecting system consists mainly of a clinical sickbed, signal amplifier, a data acquisition (DAQ) system, and operating software for computing and determining postural changes associated with four load cell sensing components. Thirty healthy subjects (15 males and 15 females, mean age = 27.8 ± 5.3 years) participated in the study. All subjects were asked to execute eight in-bed activities in a random order and to participate in an evaluation of the test–retest reliability of the results 14 days later. Spearman's rank correlation coefficient was used to compare the system's determinations of postural states with researchers’ recordings of postural changes. The test–retest reliability of the system's ability to determine postures was analyzed using the interclass correlation coefficient ICC(3,1). Results The system was found to exhibit high validity and accuracy (r = 0.928, p < 0.001; accuracy rate: 87.9%) in determining in-bed displacement, turning over, sitting up, and getting off the bed. The system was particularly accurate in detecting motion rightward (90%), turning over leftward (83%), sitting up leftward or rightward (87–93%), and getting off the bed (100%). The test–retest reliability ICC(3,1) value was 0.968 (p < 0.001). Conclusions The system developed in this study exhibits satisfactory validity and reliability in detecting changes in-bed body postures and can be beneficial in assisting caregivers and clinical nursing staff in detecting the in-bed physical activities of bedridden patients and in developing fall prevention warning systems.

Original languageEnglish
Pages (from-to)11-17
Number of pages7
JournalComputer Methods and Programs in Biomedicine
Volume147
DOIs
Publication statusPublished - Aug 1 2017

    Fingerprint

Keywords

  • Bed activities
  • Bedridden
  • Computer-aided system
  • Fall
  • Long-term care

ASJC Scopus subject areas

  • Software
  • Computer Science Applications
  • Health Informatics

Cite this