隨著人類平均壽命提高,目前許多國家均面臨高齡化的社會結構,現有的居家照護多半採人員看護式,但看護人員可能因為疲勞或其他因素無法全天候貼身照護,且發生突發狀況時看護也可能因為過度慌張而影響緊急事件的處理判斷能力。因此,本研究利用智慧型手機平台設計一套適用於居家環境之照護系統,包含提供使用者運動管理功能的計步器,依據使用者運動時的加速度變化頻率估算走路步伐、行走距離,以及消耗的總卡路里。本系統並整合Google Maps,以網頁介面模擬戶外實際散步時的環境,將計步器估算的距離及使用者的面向方位換算成使用者的虛擬位置,呈現於Google Maps上,藉以提升使用者在無人監督環境中努力活動筋骨之意願。此外,本研究以模糊推論方法設計跌倒偵測模組,依據加速度值的變化設計對應的模糊規則庫,辨識使用者日常活動,像是坐下、走路、蹲下、躺下等動作,以及是否發生跌倒之意外狀況,並推測跌倒方向。但若是僅偵測跌倒狀況,尚無法提供完善的居家照護系統。因此,本研究加入緊急通報系統與遠端視訊監控,若偵測到使用者發生跌倒狀況,系統會發送警報訊息,醫護人員便能夠即時掌握使用者受傷狀況並提供緊急醫療協助。實驗結果顯示,計步器步伐偵測準確率約為90%;若使用者沒有發生跌倒狀況,系統能91.5%判斷使用者為正常情況;若發生跌倒狀況,系統偵測準確率為89%。因此,本研究所提之系統適用於居家照護環境。
Many countries nowadays are facing a structure of aging society due to the extension of life expectancy. Home care services are usually provided by health caretakers so far. But it is quite impossible for every caretaker to take care the elderly people all day long. Even worse, caretakers may get flustered when handling emergent situations. To resolve the problems, this study is aimed on designing a home care system on Android smart phone platform that consists of a pedometer for estimating elderly steps, walking distance and the consumption of calories. To encourage the elderly having fun in doing exercise continuously, the information from pedometer is integrated with Google Maps to calculate the elderly virtual location and present the actual outdoor scene on the maps. Furthermore, this study presents fuzzy inference models to identify the actions of sitting, walking, hunkering, lying down, and falling down. However, if the devised system can only detect whether a falling down situation has occurred, it cannot be claimed to be a feasible home care mechanism. This study further integrates emergency alarm mechanism and remote surveillance as parts of the devised home care system. In case a falling down event is detected, the system will send alert messages to doctors immediately so that they can identify the situation about the elderly and provide necessary medical care or treatment. Experimental results show that the proposed system can identify about 90% accuracy of walking steps from pedometer, have 91.5% accuracy of inferring as normal under no falling down situation and 89% accuracy for detecting falling down if the elderly did fall down. The experimental results verify that the proposed system is suitable for home care environment.