智慧型手機與微型化裝置於一般及失智年長者跌倒早期偵測兼通報之應用

Abstract

鑒於台灣人口高齡化與少子化對社會帶來之巨大衝擊,老夫老妻及獨居老人比例急劇上升,年長者醫療照護已是迫在眉睫之議題。跌倒為年長者日常生活中相當普遍但卻帶來嚴重後果的意外事件。在老化過程中，由於肌肉萎縮導致與平衡相關的能力下降，進而使得年長者跌倒發生機率上升；跌倒後最常見的情況便是骨折，將導致年長者行動能力下降而長期躺臥在床，進一步使得新陳代謝率下降及免疫功能降低；此外在日常生活大小事皆需他人協助的情況下，也失去其生活自理能力。為減少跌倒對年長者所帶來的傷害，即時偵測跌倒並快速給予適當醫療照護是重要的課題。本研究為一般健康年長者及安養院所失智症年長者發展兩套跌倒偵測兼通報之系統，考慮到廣泛應用於居家照護之時效性,及安養照護的個資安全與偵測嚴謹，使用了不同材料及方法來建立兩項跌倒偵測系統。 在本研究所提出的兩項系統中，其一為智慧型手機應用程式之居家照護跌倒通報系統，具備易於推廣至社會大眾，以及後續維護及更新之各項特質。此外，在智慧型手機用跌倒偵測演算法中，所擷取的特徵值能考量到手機日常使用情境，即使不將手機固定在單一位置，也能有相當高的跌倒偵測準確度。另一則是為了安養院所失智症年長者，所開發之高精準度微型化跌倒裝置，此微型化裝置具備高雜訊比以及低功耗的特質；而考量到此類跌倒裝置隸屬於醫療器材範疇，為保有高嚴謹度將搭配安養院內之資料傳輸系統。此外，所開發之高精準度跌倒偵測演算法使用真實跌倒事件訊號做驗證，仍然具備相當高的跌倒偵測靈敏度及準確度。 本研究中所建立之兩項跌倒偵測系統，針對個別年長者族群使用了不同的方法來記錄及分析年長者的活動訊號，如此一來提高了實際使用之可行性，將此雙系統應用於一般及失智症年長者跌倒早期偵測之居家及安養照護，應可有效減少跌倒對年長者所帶來的傷害。

Considering the huge impact from the increasing aging rate of population as well as decreasing birth rate in Taiwan, health care of the elderly is an issue that must be addressed without delay. Falling is a common accidence among the elderly and has serious consequences. The most common consequence of falling among the elderly is bone fracture, which leads to hospitalization and decreased activity levels. With low activity levels, metabolic rates will decline and thus the elderly people can become susceptible to diseases. To reduce the injuries from falling, detecting fall events early by an automatic fall detection system and sending suitable medical care are widely adopted. In this investigation, we implemented two kinds of fall detection systems aimed for home care of the elderly living alone and nursing care of the elderly with dementia. To obtain widespread acceptance in the elderly community and reach high security and accuracy of nursing care at the same time, two kinds of fall detection systems were developed by using different materials and methods. The fall alarm system for home care is implemented with the smartphone APP and the analyzing clouding platform that can report the fall events of the independent elderly immediately. By using the smartphone APP and clouding platform, the designed home-care fall alarm system is widespread acceptance among the elderly, and also maintains and upgrades subsequently. In addition, the cooperative smartphone-based fall detection algorithm contained novel features which enhance the performance of fall detection in real condition. The other designed high accuracy miniature fall detection device is for the elderly with dementia in nursing centers. The miniature device is designed with high SN ratio, low power consumption and potential of miniaturization as a small chip. Additionally, the designed miniature device-based fall detection algorithm is evaluated by the real-world fall repository and performs higher sensitivity and specificity than previous studies. In this dissertation, two kinds of fall detection and alarm systems with distinct monitoring and analyzing techniques are developed. These two systems not only meet the requirements of application in real condition but also show higher performance in fall detection than previous studies. With these characteristics, our proposed two systems applied to early fall detection in home care and nursing care could be feasible.