Development of an accident detection system for care service users by image and motion analysis

In the field of nursing care for the elderly or handicapped people, few service users to 1 care givers as legal standards of personnel positioning at every establishment in Japan. However, there is some lax attention caused by repititive job routine. The program is focused on the specific time and situation, which is guiding the service users to the dining room from each room and waiting for them until everybody is ready to eat. The program will watch each service users´ face with a couple of observation points. One is to indentify a pain facial expression which is the main point and the other one is head motion. Both evaluations are analyzed one in different ways. In this study, firstly, a pain stimulation test where healthy adult subjects dipped their hands into ice water was carried out. A strong correlation between pain face factor calculated from the facial expressions digitized by the system and sensory evaluation shown by a pain score, VAS (visual analog scale) value, was observed for young adults. It indicates the validity of this system for the estimation of pain intensity. Secondly, a specific position and motion pattern was regarded as a critical region in each person was decided according to range of motion which was measured in each service user´s head by an OT (occupational therapist) and a BS (bornsetter), in order to tell care givers when the service users´ head moved in to a critical region or shifted unnaturally. Because there is a grat variation from individual to individual, the motion analysis requires an initial setting for each person. The study indicates the capability of the system by head traking accuracy.