Development of a Machine Learning System for Predicting Stress at the Workplace

Pablo Sainz San Juan. (2022). Development of a Machine Learning System for Predicting Stress at the Workplace. Trabajo Fin de Titulación. Universidad Politécnica de Madrid.

Nowadays stress is a widespread phenomenon that appears in our lives consciously or unconsciously and it is considered that a large proportion of work-related sick leave is stress related. Although a certain level of stress could act as a positive factor, prolonged exposure to high levels of stress could have detrimental effects on health and can trigger or worsen diseases such as: depression, panic attacks, hypertension, diabetes and heart problems. It is possible to predict the level of stress non-invasively with machine learning techniques such as: text analysis, body posture, pulse, pupil diameter... In addition, these stress detection techniques avoid the need to acquire specific hardware for its use since we can find all the necessary elements in a computer. This TFG focuses on the development of an application that helps us to analyze and detect stress in a non-invasive way by using some of the methods mentioned above; with the aim of being able to deploy it in academic or work environments. The objective is to develop a Machine Learning model capable of predicting stress from text, keyboard use and mouse movement. To achieve this, the state-of-the-art of the research will be analyzed, then datasets will be collected allowing us to develop and validate different Machine Learning models. Finally, a stress detection application integrating these models will be developed. To ensure a better follow-up, weekly meetings with the tutor have been established, which in case of not being able to be face-to-face, will be held by videoconference or e-mail.