Intelligent Integration in Digital Twins: A Deep Meta -Class Sequence Model for Healthy Aging

Supervising elderly people’s daily behavior within their domestic settings can be regarded quite problematic when it comes to delivering healthcare practitioners with the appropriate information to evaluate these actions. This paper presents a new concept in home monitoring by developing an environment model of the elderly person’s home therefore enabling independent monitoring of their activities. To tackle the learning difficulties associated with different home activities the framework employs a deep meta-class sequence model. This model clusters all related activities into one meta-class in view of their concerned nature and this simplifies the learning process. This model draws from the long short-term memory networks to extract feature representations from the data to make sure that every activity is fully understood. It is also possible to analyze activity in detail since each meta-class is divided into the activities of the elderly. During experiments, the proposed model yield more accurate results compared to the baseline models while tested on CASAS dataset. These observations highlight this framework as a sound solution for digitally observing elderly conduct and as a source of rich information for caregivers and health managers regarding the requirements for care for elderly people and possible signals of dangers in their homes. That is why this approach highlights digitalization and the use of sophisticated sensing tools in creating and developing healthcare and improving the quality of the elderly’s life.