ML-IoT SmartCare: An Explainable ML and IoT based Smart Care for Elderly People

Most elderly people in Norway want to lead a self-sufficient and independent life, live in their own home and receive assistance from the municipal health services (home health nursing, practical assistance, support contact, etc.). The health services are granted by the individual contract between the municipality and the recipient. Around 30%-40% of elderly people in a municipality receive weekly on average 9 to 14 hours of care service. This implies a significant portion of total medical professionals are directly engaged with elder home care. Norwegian health-care industries experience tough time due to shortage of medical professionals as reported by helsedirektoratet [3]. Meanwhile, over 20% of adults aged above 60 suffer from a mental or neurological disorder such as depression, anxiety, dementia, and Alzheimer’s disease [4]. This issue can be more serious for elder people who are living alone than in the nursing home due to lack of engagement from on-duty professionals. Along with increasing the number of medical professionals, advances in emerging technologies that improve the quality of life of the elderly population while reducing strain on healthcare systems and minimizing their operational cost are highly demanded.

Internet of Things (IoT) technologies, increasingly integral to healthcare and smart homes, have the great potential to enhance elderly care significantly. By tracking elders’ movement, biomarkers, and behavior pattern changes, IoT technology can provide early alerts about elders’ physical and mental health to caregivers and family, preventing severe issues and reducing emergency hospital admissions. IoT technologies further support the role of the built environment and caregiving to produce outcomes that enable older people to remain autonomous and stay at home longer safely with dignity [5]. We have seen mobile apps, modern wearable devices, and discrete monitors that continually collect a plethora of biomedical signals. In many cases, physics-based models are used for post-processing of information received from IoT devices that deployed to monitor patient’s health/activities. However, physics-based models are not always proficient in predicting the situation with tolerable precision. For Instance, RSSI or light-based indoor patient localization mechanisms often fail to provide accurate information about the location of a patient. On the contrary, the recently evolving Artificial Intelligence (AI) and Machine Learning (ML) technologies offer great potential with various applications in our daily life. Interest in leveraging ML algorithms for healthcare has grown immensely, including medical imaging, disease diagnosis, supportive clinical decision making, etc. When applied to elderly homecare, one direct application of ML algorithms is to extract sophisticated information collected by IoT devices, provide descriptive insight/summary of collected data, as well as contributing to a qualifiable measurement of elder peoples’ physical and mental health.

Motivated by the aforementioned observations, the underlying idea of the ML-IoT SmartCare project is to further investigate the feasibility and niche of utilizing emerging IoT and ML technologies, and to develop a viable solution for elderly homecare by integrating data collection based on IoT devices and abnormality/emergency detection based on ML algorithms. The overall goal of this study is to build an AI assisted system that support and improve the caregiving quality and capacity of elderly home healthcare.