| Summary: | 碩士 === 國防大學理工學院 === 光電工程碩士班 === 107 === In view of the year-on-year increase in the number of elderly people, Taiwan has officially entered the advanced society in 2018. The slow response to growth with age is often accompanied by many chronic diseases. These factors are particularly prone to falls, and falls often have irreversible consequences, resulting in families and The society must bear the burden of losing loved ones or disability care, so detecting the occurrence of falls and providing timely medical assistance to alleviate the damage caused by falls has become an important issue”.
This study is in line with the current trend of smart bracelets. It integrates the micro-control unit, accelerometer and LoRa wireless network module into the wrist, and collects four kinds of fall behavior data such as front, back, left and right falls through the LoRa wireless network. And the Siginal Vector Magnitude (SVM) single threshold combined with the Dynamic Window Approach (DWA) method to extract the accelerometer fall and daily life behavior characteristics information, and then use the inverse transfer neural network Back Propagation Neural Network (BPNN) algorithm learning, supplemented by walking, sitting, standing up, going upstairs, going downstairs and other five activities of daily living (ADL) to verify the accuracy of the fall event, the experiment proves the use of BPNN The classification method can successfully realize the occurrence of the identification of fall behavior. The accuracy of the method for offline detection of fall is 100%. In real-time detection, 100% fall sensitivity can also be obtained. All fall behaviors can be detected and the false positive rate is 0.8%. In the process of system operation, there will be no extra brain trouble caused by false positives. This is a satisfactory level of false positives. ADL behavior of different age groups, verify that the system can also be used to make fall judgments in different age groups; after the fall occurs, the fall alarm will be transmitted to the personal computer or mobile device via Ethernet via 4G/Wi-Fi, so that the caregiver can provide the immediate care. Medical care to reduce the damage caused by falls.
|
|---|
