Research On Activity Recognition And Daily Behavior Of Home-based Elderly Based On Machine Learning

The problem of population aging has intensified,making the proportion of the elderly population larger and larger.The elderly care service industry needs to meet the growing needs of the elderly.However,due to the lack of domestic resources for the elderly and the inadequate elderly care service system,the health care of the elderly faces severe challenges,especially the elderly living alone.The development of sensor technology and pervasive computing enables various sensors to be worn on the elderly or deployed in the elderly homes,thereby generating a large amount of data through the elderly sports,daily activities and changes in their physiological conditions.At the same time,the development of big data processing and artificial intelligence technology provides efficient means and methods for analyzing various data of the elderly at home and studying the behavior of the elderly.The main research work of this article includes the following three parts:(1)Research on activity recognition of the elderly based on deep learning.In order to solve the problem of the imbalance of the data of the elderly daily activities,this paper re-samples the data and uses different neural network models to identify the sensor data generated by the elderly daily activities.Research shows that the hybrid CNN+LSTM model can well identify activities of the elderly.(2)Research on Alzheimer’s Disease in the elderly based on machine learning.Through the voice data collected by the microphone device in the smart home,the relevant voice features are extracted and analyzed,taking the case of Alzheimer’s disease as an example,to identify the health status of the elderly.Research shows that the LogisticRegressionCV model can well identify the elderly illness status.(3)Research on the discovery of the elderly daily behavior rules based on multiple conditional association restrictions.The daily behavior of the elderly is a coarse-grained representation of their daily activities.It is not limited to a specific activity.It does not require the sensor ID,trigger time,and location triggered by the activity to be consistent.It is based on long-term daily activity data.Abstract the general behavioral laws of the elderly activities.In this paper,by correlating the three conditions of time distance,optimal path and sensor distance,the daily behavior patterns of the elderly are studied for a period of time,and it is found that the daily behavior patterns of the elderly can be well separated by EM clustering algorithm.Figure[39]Table[12]Reference[72]...