Lying and sitting posture recognition and transition detection using a pressure sensor array

This paper demonstrates the use of a bed-based optical pressure sensor array to unobtrusively recognize sitting and lying postures as well as lie-to-sit postural transitions. Young healthy, older healthy, older post-stroke, and older post-hip-fracture participants performed a bed entry and exit routine. Data was collected using a pressure sensor array and video cameras. Lying and sitting postures and transitions were analyzed by our system and compared to video analysis from two medical students. For posture identification, eight pressure signal features and three classification techniques were compared. For transition detection, a movement detection algorithm was implemented and combined with the posture identification system. Postural detection accuracy of 100% was achievable using a combination of pressure features. Postural transition detection held a very low miss rate. Differences in measurement of transition duration between our system and video analysis were statistically insignificant.