A state of the art mathematical model of cancer microenvironments that describes the proliferation of cancer cells and their interaction with four types of immune cells namely the natural killer (NK) cells, the CD8+T-cells, the conventional CD4+T-cells and the regulatory CD4+T cells (Treg) and two types of Tumor Associated Macrophage M1-like and M2-like, immersed in a biphasic solid/fluid environment described by a mechanical model of cancer and healthy tissue was introduced in [1]. Coupling with time dependent PDEs modeling the conservation of oxygen, the drug delivery and the delivery of immunotherapy finalizes the parametrized coupled multi-physics model that is solved in the context of a stochatic simulation to reveal the influence of each factor on the growth of the tumor and provide guidelines for designing effective combinatorial therapeutic strategies. The intractable computational cost of the uncertainty quantification study is aleviated by application of a novel surrogate modeling strategy that uses a convolutional autoencoder in conjunction with a feed forward neural network to establish a mapping from the problem’s parametric space to its solution space[2]. [1] F, Mpekris et al. Combining microenvironment normalization strategies to improve cancer immunotherapy. Proc Natl Acad Sci U S A . 2020 Feb 18;117(7):3728-3737. doi: 10.1073/pnas.1919764117. [2] S. Nikolopoulos et al. , Non-intrusive surrogate modeling for parametrized time-dependent partial differential equations using convolutional autoencoders , Engineering Applications of Artificial Intelligence, Volume 109, 2022, https://doi.org/10.1016/j.engappai.2021.104652.