Tracking activity via body-attached sensors is aimed at helping people who suffer from neurodegenerative diseases including Alzheimer’s disease and dementia, or those with some ongoing neurologic-related symptoms like memory loss, limited motion ability with an increased risk of balance loss, or disorientation. S-SMART is proposed to such people as a solution or simply a tool that, by monitoring their routine activities, can support them in their lifestyle.

Twenty-three possible scenarios called action classes have been predefined by the authors in order to model the environment stage of the user based on the user’s probable interactions represented by the semantic maps that position him or her in action using the proposed unified Bayes rule, in which objects such as windows, doors, water taps, and phones associated with the environments are defined as a set of states--for instance, open and closed door--and correlated to activity-conditional probabilities.

According to the authors, S-SMART outperforms some other systems of this kind from the perspective of unsupervised learning mapping, handling the activity observations, avoiding activity disambiguation, and utilizing stateful mapping.

Implementation of the S-SMART system is presented using examples of some people performing their routine activities in a set of environments. The statistical sample seems large enough to evaluate the system efficiently and reliably, especially because the evaluation measures like precision and recall are classical and thus can be compared to other models of the same kind.

However, I feel that the proposed set of activities or states to be recognized should be oriented toward more risky daily situations like taking drugs on time, having a mobile device in hand for emergencies, checking the gas or water tap, not opening the door to unknown people, finding a lost patient, and so on. User privacy should also be considered and thus addressed in this paper.

Although the topic is interesting and the presentation is quite clear, the paper does contain some grammatical errors (for example, incorrect punctuation, occasional omissions of definite or indefinite articles, and some occasional spelling mistakes). Still, I find the paper worth recommending both to the research community and specifically to medical companies as a very pragmatic solution to the problems that more and more senior citizens face nowadays.