Context-aware-based Location Recommendation for Elderly Care

Abstract

As adults age, the body declines. Living independently at home can be a significant challenge for the elderly, particularly for those who suffer from dementia or who have memory impairment. Assisting the elderly to live independently and safely in their own homes by providing appropriate services for them and ensuring that caregivers are immediately alerted in the event of an emergency is crucial. Utilizing context in the recommendation process will make recommendations more appropriate. A model of a context-aware-based location recommender system that can seamlessly monitor the location of the elderly and deliver appropriate location recommendations by considering context is proposed as our contribution. Two scenarios are investigated: (1) we classify location as follows: bedroom (class 1), dining room (class 2), and living room (class 3); (2) we classify location as follows: inside (class 1) and outside (class 2) the bedroom. We evaluate our proposed model using a distance measure concept by employing the cosine distance method. We compare the cosine distance method with fuzzy inference system (FIS) rules on labeled data. The results of the experiments for the first scenario show that the cosine distance has better average accuracy than the fuzzy inference system. For the second scenario, fuzzy c-means (FCM) has the same average accuracy as cosine distance. FCM has slightly better accuracy in class 1 compared to cosine distance (1% difference in accuracy), whereas cosine distance has slightly better accuracy in class 2 compared to the FCM (1% difference in accuracy). In general, we can draw the conclusion that, on this dataset, cosine distance which uses a simple algorithm produced better results than the fuzzy inference system which uses a more complex algorithm.