Immersive environments are digital spaces that surround the user, creating a sense of presence and participation in the virtual and augmented world. Unlike traditional interfaces, which present information in two-dimensional form, immersive environments seek to generate three-dimensional and multi-sensory experiences that stimulate iterations with the simulated environment. From these iterations between users and their environment over time, a lot of data is obtained that, when analyzed, generates enough useful information to improve their experiences within these immersive environments. On the other hand, deep learning models are considered the state-of-the-art in various offline machine learning tasks. Traditionally, training deep neural network models requires that all data be available at the start of training in an offline environment. However, many of the techniques that have been developed are not considered suitable for online learning scenarios. That is, this learning scheme is not suitable for many practical situations where data arrives over time. To study the methods and models of deep learning suitable for working in these scenarios, the area known as Online Deep Learning was created. To use this type of models within immersive environments, the key lies in the ability to continuously learn and adapt to new experiences within the environment, without having to retrain the entire model from scratch each time. This work aims to study the feasibility of adapting and integrating some of the main techniques of Online Deep Learning within immersive environments. In addition, it studies the possibility that some methods and models in this area can be adjusted to the limitations of the different devices that support extended reality technologies. These methods should provide real-time responses to user interactions with promising levels of accuracy.