A framework for developing a smart and adaptive environment for aging population
Nirmalya Thakur, Chia Yung Han · Institution of Engineering and Technology eBooks · 2020
This work presents a framework that takes a holistic approach toward fostering active aging of elderly people in the future of smart homes and empowering them to perform ADLs in an independent manner. Two specific functionalities presented are (1) a personalized task recommendation system that takes into account the various micro- and macro -actions that a user performs in the context of any given activity and recommends them actions and tasks to complete the same and (2) a fall prediction model that can analyze several dynamics of human motion and predict whether the person is going to fall, even before the fall occurs. These functionalities of the framework have been tested on several datasets and the results are presented and discussed. Comparison studies were also performed for each of these two functionalities where several learning models were implemented. Based on the performance characteristics of these learning models, a learning model that achieves the highest performance accuracy for each of the respective functionalities of this framework was adopted for our study. To the best knowledge of the authors, no similar approach has been done in this field yet. The results presented uphold the relevance and demonstrate the feasibility for practical implementation of this framework in the future of smart homes. This framework is envisioned to lead toward the development of the future of adaptive behavior intervention technologies in health care and social welfare that will be able to analyze, assess, model, profile and provide efficient personalized behavior interventions to enhance user experiences and foster elderly care while addressing the varying diversities of the elderly population. Future work would involve setting up an IoT-based environment through a host of both wireless and wearable sensors to implement this framework in a real-time context.