Ciencias Exactas y Ciencias de la Salud
Permanent URI for this collectionhttps://hdl.handle.net/11285/551039
Pertenecen a esta colección Tesis y Trabajos de grado de las Maestrías correspondientes a las Escuelas de Ingeniería y Ciencias así como a Medicina y Ciencias de la Salud.
Browse
Search Results
- Stress levels detection from EMG signals in virtual reality: a machine learning approach(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2022) Orozco Mora, Carmen Elisa; Fuentes Aguilar, Rita Quetziquel; emijzarate, emipsanchez; Tovar Corona, Blanca; González Mendoza, Miguel; Escuela de Ingeniería y Ciencias; Campus Estado de México; Hernández Melgarejo, GustavoVirtual Reality technology has had an outbreak in mainstream usage in the last decade, nowadays it is being used in areas like entertainment, education, rehabilitation, etc. In the entertainment area, Virtual Reality is implemented in video games to enhance the enjoyment of the player and has become an important asset in the industry since there exists a constant search for ways to improve gameplay and immersion experience. The difficulty level of a video game has a direct impact on the enjoyment and feeling of flow of the players, if they find the game too hard or too easy they can lose interest and leave with a bad impression of the game, but if the difficulty was to adapt to their levels of stress then the flow, enjoyment and thus, feeling of presence of the player would be enhanced. The analysis of some physiological signals that have a direct correlation with the emotional state of a person and could help with the assessment of their stress level. There exist previous studies where they managed the latter using different Machine Learning techniques and obtaining different results and levels of accuracy. Nonetheless, there does not exist a forearm-based electromyogram signal classifier that determines the level of stress of a Virtual Reality video game player. In this study, a Virtual Reality video game was built, it was designed to induce three different levels of stress in the player. This game was used to compile a dataset containing features extracted from the forearm EMG signals from volunteers while playing said video game. Different Machine Learning algorithms were trained and tested for classification. We obtained up to 66.6% accuracy in the 4-class classification between three levels of stress and a resting stage and 100% accuracy in the binary classification of a resting stage and the different levels of stress. The best-performing models were embedded in an external system to make predictions online. These were validated through experimental and statistical tests. Finally, a system was built which allowed to dynamically change the difficulty of a VR video game accordingly to the predicted level of stress of the player. We concluded that the system was capable of reacting as expected and that the difficulty of a game can indeed be modified online with the input of the emotional state of the player.