嵌入式微機電系統在生醫電子的應用

Abstract

跌倒及跌倒以致傷殘的發生機率在老年人中非常的普遍。在台灣，每年約13.6％之老年人曾經因為跌倒而導致骨折。通常那些因跌倒而骨折的老年人無法自行爬起就醫，而需要其他人幫忙及照護。跌到偵測系統的發展正是為了回應老年看護與通報的殷切需求。過去通常使用微機電(MEMS)系統提供老年人的跌倒偵測系統，然而這套系統卻要即時偵測真實跌倒情況，卻有其難度。在這篇論文中，我們提出一種演算法，透過使用簡單便宜的三軸加速儀和陀螺儀作為感測器，和Arduino以及手機做結合，用以估測人體姿態。藉由將裝置微小化並且放置於手腕上，以及經由藍芽將資料傳輸至手機上，可獲得老年人身體姿勢，可用以偵測可能跌倒的警訊 本系統跌倒偵測的平均敏感度為97%，平均變異度為99.111%，判斷跌倒偵測所取的測量點數為30點決策一次。 因為傳統型血壓計和血氧機體積大且不容易攜帶，我們利用MEMS將兩者皆微小化於手腕錶帶上，以提升準確度，使其能夠和傳統型的偵測器抗衡。我們發展出的偵測器能將所得到的數據傳到手機上，再藉由手機送回控管中心，使病患的生理特徵能夠被及時掌握，不但可提升危機時將病患救回的機率，更能防患於未然。

The occurrence of falls and fall-related injuries are very common. Each year in Taiwan, approximately 13.6% of the elder people experiences injuries and even suffers a fracture or sprain, when they fall. Unsurprisingly, once they fall and sustain a fracture, they cannot stand up on their own; instead, they need the immediate help offered by a third party, either a family member or a caretaker. It is with the aim to offer timely help to elderly people who experience accidental falls or slips that a fall detection system has been developed. However, the fall detection system developed in the past, which utilizes the micro-electromechanical systems (MEMS), does not provide real-time detection of a fall, thereby limiting its value and efficacy. In this thesis, we propose an algorithm, which makes use of cheap and simple accelerometer and gyroscope as sensors, and we also employ an Arduino and android phone to estimate the posture and positions of the human body at the moment of a fall. By miniaturizing our device so that it is wearable on the wrist, we design and create a fall detection system that can transmit data through a Bluetooth serial port to the android phone to determine the postures of the elder people who wear the device. In so doing, we can easily determine if a fall has occurred, and, if so, in what posture the elderly person falls or slips. The average sensitivity of our fall detection system is 97% and the average specificity is 99.111%. We make our falling detection decision every 30 measurement point. Because the traditional sphygmomanometer and blood oxygen machine are too big and not easy to carry, we have decided to use MEMS to miniaturize these two devices into an integrated wristband. Besides, not only do we improve the accuracy of our device, but we also make it a smart device that can transmit the data we catch to an android phone which then immediately sends the relevant data to the medical control center for further analysis. By thus relaying the vital signs of the elderly person who wears the device to the control center, our detection system can not only provide the vital information needed to save an elderly when a fall occurs, but it can also help to plot preventive actions; that is, it may detect a fall even before it actually occurs.