應用隱藏式馬可夫模型於居家被照護者之行為建模

人類少子化及高齡化現象已有逐漸昇高之趨勢，年長者之居家安全與照護即顯得更加重要。本研究係針對本驗室所設計自主式居家照護系統的核心需求，建立一被照護者居家活動時之行為模型，本模型可讓自主式居家照護系統預測被照護者之行為，並在被照護者發生重大意外時，即時向外界請求協助。在居家環境中，尤其是在浴室內，因保有隱私不能使用攝影機作為系統的感測器，需採用非侵入式、非視覺性感測方式。故對一自主式照護系統而言，被照護者的居家活動大多為隱藏狀態。然而研究分析年長者的居家活動特性後發現：居家活動具有習慣性，而每一活動均有其目的(意向)並由按特定順序之連續動作所組成。其居家之動作大多在特定位置執行，且均可觸發在固定位置上之非視覺感測器。最後設計之系統不能反覆詢問被照護者狀況，以避免使用者厭煩且錯誤之詢問容易造成被照護者對系統的不信任感。綜合上述之需求，本研究採用隱藏式馬可夫模型(Hidden Markov Model, HMM)之建模與識別技術於居家環境中，藉由一連續相關特定動作與事件，建立被照護者之個人行為特徵與模型。藉由上述技術加強看護系統推論機制的正確率。

關鍵字

隱藏式馬可夫模型；智慧型居家照護系統；學習演算法；活動模型

並列摘要

Benefited by the modern medical technology, the high-age population is growing up rapidly. In this circumstance of aging, the development of aging-people homecare systems is important and it is forwarding to personalized, autonomous, pervasive, remote, and daylong. In addition, the ubiquitous computing is developed rapidly, applies such technology into the high-age people homecare systems is a more important issue. Those systems can perform the complicated homecare task based on its built-in behavior model and the multiple, heterogeneous and independent sensing agent. In this paper, a learnable behavior model of the aging family is presented and used by our designed intelligent pervasive-care system. This system uses multiple interlaced and non-visual sensors to recognize the behavior of concerned family in independent and concurrent. Based on the feature of habitual activities of aging population, a hidden Markov modeling technology is used to build a habitual behavior models and recognize the purpose of behavior in the environment of family. Model learning and behavior recognition algorithms are also realized in this paper. So the behavior model can empower the intelligent pervasive-care system to monitor the home activities of cared aging family, and assist them to inform suitable peoples when some accidents occurrence.