The workplace is one of the environments that most impacts on population psychological health and well-being, and there is growing evidence that investment in work-related well-being derives substantial benefits for employees and companies. The rapid development of technology has fostered the development of new approaches to the promotion of emotional and mental health. Thus, this thesis aims to advance in the research of this field and its application to the workplace. Specifically, we sought to advance the research in three different lines: automatic psychological awareness using non-obtrusive methods; analysis of mental and emotional well-being promotion strategies; and smart environment adaptation.
Regarding automatic psychological awareness, this thesis contributes with advancements in unobtrusive and economical solutions for automatic mental illness recognition based on machine learning. On the one hand, we further early mental disorder detection from text with an approach to detect psychological stress combining a lexicon-based feature framework with distributional representations. On the other hand, we have designed a machine learning approach to predict the current stress level of an individual using their surrounding stress-related data, that is, their previous stress levels along with stress levels from their close colleagues. Both solutions showed positive results during the experimental evaluation.
As for the second line, this thesis contributes with the design of an agent model for occupational stress based on ambient and work conditions. The model has been integrated into a simulation system, enabling the evaluation of different mental health promotion policies at work. The reliability of the system has been validated through several experiments including in-lab experimentation and sensitivity analysis.
Finally, this thesis advances in the research on the intelligent adaptation of the environment. An architecture has been designed for a task automation platform that integrates through semantic technologies sensors and actuators oriented to the detection and regulation of emotions. In this way, the platform promotes the well-being of employees while ensuring the interoperability and scalability of all its components. It has been observed that this platform has positive results in the experimental evaluation, increasing the well-being and productivity of users.