Pervasive Sensing and Intelligent Learning of Daily Routiones in Ambient Assisted Living

Abstract

The world’s aged population is increasing by a significant factor due to medical and other facilities like science discoveries. As the number proliferates, requirements of this segment of age are multiplying at a very high rate. The percentage of aged persons living alone also rises at an increased rate due to the inevitable socio-economic changes. This situation demands complete solutions for various problems like loneliness, chronic conditions, social interaction, transportation, day-to-day life, and many more for an independent living person. A large part of the aged population may not be able to interact directly with new technologies. The circumstances sought some serious development towards using intelligent systems, i.e., sensible devices, which help people with their inability to use the available and future solutions. Ambient Assisted Living (AAL) is the answer to these problems. In this thesis, the AAL system’s phases are covered with an indepth analysis of each topic. The research identified the unique challenges of the AAL domain and proposed the solution for each problem in subsequent work. A significant challenge to provide services to the inhabitant in a smart environment resides in implementing models. Most of the proposed models are conceptual and lack practical consideration. Human activity recognition is one of the most challenging tasks to offer the solution for ambient assisted living. In the subsequent work, we explore a time series classification problem - human activity recognition. A total of nine machine learning and deep learning algorithms were implemented and evaluated using an identical dataset. The result is analyzed using different parameters. This solution aims to select a practical machine learning approach for the activity recognition process in ambient assisted living systems. Human behavior monitoring is the next important step that analyzes the person’s behavior in a smart environment. After dealing with the activity recognition problem, the work proposes a novel hybrid framework to monitor human behavior in assisted living facilities. The framework is evaluated using the working prototype incorporating sensing technology and knowledge inference techniques. The daily routine of the occupant is observed during the training phase. Once the inference engine recognizes the routine, all the daily living activities are assigned to any distinct cluster. After observing the routine for a few days, anomalies can be detected using the proposed module. A considerable percentage of the elderly population may not be interested only in the ease of living. Nevertheless, the users of the assisted living system may be more inclined towards health-related services. This situation needs some substantial development in the sphere of intelligent assistive systems. To devise a solution for ambient assisted living, we first decided to take the user’s choice of their requirements based on their needs and behavior in daily life. To record the user’s choice, we conduct a survey that lists a questionnaire requesting the potential AAL users to provide their answers. The questionnaire includes topics related to the day-to-day life of a person. It also incorporates the preferences that they wish to have to improve their quality of life. The work presented next incorporates the user’s perspective in designing an intelligent healthcare system. The proposed approach is evaluated using various datasets utilizing conventional sensor data, sensor data with visual input, and sensor data with human emotion. Different deep learning algorithms are applied to predict personal health. The results show significant improvements while considering both intrusive and nonintrusive systems. Finally, we covered the research efforts available in similar fields. The increasing pervasiveness of age-related health problems is a significant factor driving revenue growth of the market. The ambient assisted living domain gives today’s technocrats ample opportunities to develop solutions for older adults. We have observed numerous efforts by the companies in one or other forms to satisfy the need of the elderly. The case studies included here support our research efforts and their implications in providing commercial solutions for the AAL domain. The research work covers all the essential phases of an ideal ambient assisted living system. Pervasive sensing technology is utilized here to learn human behavior intelligently and offer smart services to the inhabitants.