建構具有危險預測能力的安全智慧家庭環境

Construct a Safe and Intelligent Family Environment with Dangerous Predictive Ability

近年來，由於人口老化的問題越來越嚴重，因此家庭設備的自動化及居家安全的自動偵測就顯得越來越重要。本研究提出一個能即時提供居家安全的智能環境「Be-Health Home」，透過觀察家中長輩日常生活的行為（Behavior），藉以實踐電子化健康照護（Health Care） 的效果。透過「Be-Health Home」所實現的智慧家庭（Smart Home），能夠在任何地方任何時間依據長輩的日常生活行為，預測長輩是否可能進行危險的行為，因而可以照護家中的長輩避免發生危險。研究中，將所觀察到的行為分成正常與非正常兩種。若所觀察到的行為是正常，則繼續追蹤行為。反之，若觀察到的行為是非正常的行為，則繼續判斷該行為是異常的行為或是危險的行為。為了避免觀察到的行為與其他正常行為很相近，所以另外設置了一個持續追蹤的機制，讓在遠方的家人可以觀看家中長輩此時此刻的行為是否具有危險性。

In an ageing society, the living safety at home of elderly person has gained attention and concern of the parties in the field. However, the cost of providing a safety smart home environment is very high. Therefore, according to the elderly person's behavior patterns, the “Be-Health Home” is proposed, which can provide more safety environment in any time. Through to observe activities of daily living and cross validation with behavior, the health care in smart home can be reached. The observed behaviors by “Be-Health Home” can be divided into normal and non-normal. If the behavior is detected as normal, then continue to track monitoring. If the behavior is classified as non-normal, the behavior will further be analyzed to distinguish abnormal or dangerous behavior. If the behavior is judged to be abnormal state, then the frequency of detection of tracking and monitoring is increased to prevent dangerous behavior. If the behavior is determined as dangerous, warning is issued and the subsequent processing immediately. However, through the proposed model, the safety living environment can be obtained.