A Recognized System of Bedridden Behavior for Disabled Elders

• Main Authors: Tsung-Min Liu, 劉宗旻
• Other Authors: Chun-Ju Hou
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/75477826843194797066

碩士 === 南台科技大學 === 電機工程系 === 97 === With the change of social environment and the advance of medical technology, the rate of the aged people has been dramatically increased. The function degeneration of the aged people results in the decay of the ability in activities of daily life. However, in the limit of human resource, clinical nurses bend with a heavy burden on physiological and psychological aspects for the elders’ safety care. It is necessary for applying healthcare technologies to assist clinical nurses in patient’s care. The purpose of this study is to establish a recognized system of bedridden behavior for disabled elders. The recognized system can detect body movement, position and posture in bed, and fall from a bed. Five specific aims are 1) to design a pressure sensing mattress, 2) to build a calibration system for pressure sensing mattress, 3) to develop recognized algorithms of bedridden behavior in bed, 4) to build a database and graphic user’s interface, and 5) to integrate hardware and software systems, and verify the feasibility of the integrated system. Algorithms of bedridden behavior analysis in bed include the analysis of 1) grey level images transformed from pressure distribution, 2) body movements such as change of posture, restlessness and quiet recognized by using activity score and time duration, 3) body postures such as left, supine and right lying, and positions such as left, middle and right sides, and 4) Out-of-bed analysis using the sum of projection estimated from pressure distribution image. System integration and experimental verification include pressure sensor calibration, system response time, and the accuracy of bedridden behavior recognition. The results for the test of system response time show the proportion of positive response frequencies to total test frequencies is 0.83 after four seconds and 0.96 after five seconds. The accuracy is 71.8% in body posture recognition, 98.5% in body position recognition, 70% in the restlessness recognition, and 100% in the out-of-bed recognition. The recognized system has been verified the feasibility of bedridden behavior recognition using pressure sensing mattress and the establishment of recognition algorithm. In the future, improvements in the system include reducing numbers of pressure sensor, modifying recognition algorithms, and considering object interference such as pillow and quilt.