A facial expression recognition system for eHealth intervention platforms: A proof of concept

Resumen

Adherence is essential in any kind of treatment. The noncompliance or lack of adherence is a complex problem that comprises a wide variety of situations and has multiple causes. It is necessary to know the level of follow-up in order to increase it as much as possible and thus improve the patient's health, both physical and mental. People?s mood and motivation are key aspects. In order to help to deduce these variables, a system capable of recognizing emotions from facial expressions has been designed, oriented to people who are carrying out any off-site therapy. Thus, thanks to the automatic detection of their emotional state, knowing how the treatment affect them will be possible, in order to personalize it and maximize its adherence and effectiveness. Public datasets have been used, containing a large number of images with the specific characteristics that are necessary to obtain the desired effect on the network. In the different webpages that the application contains, on the one hand the registration of a session is possible, period during which the facial expressions shown by the user will be detected and stored. The evolution of these emotions can be graphically visualized, for any of the users and the sessions registered in the database. This system has been submitted to some tests with different subjects to demonstrate that its operation is correct, and that the detection is independent of the user.