Objectives: Continuous renal replacement therapy (CRRT) is an essential treatment for uncontrolled severe metabolic acidosis. However, CRRT is a relatively invasive treatment that requires the establishment of a new central line and may lead to complications. Therefore, selectively applying CRRT to patients with significant treatment benefits is crucial. This study aims to investigate the therapeutic effect of CRRT in patients with severe acidosis by utilizing a deep learning-based causal inference model to assess its potential impact on in-hospital mortality. Methods: The MIMIC-III dataset was utilized, and subjects with data available within the first 48 hours after intensive care unit (ICU) admission were selected. Patients experiencing severe acidosis with a pH < 7.2 within the initial 48 hours were selected. Treatment was defined as the application of CRRT within 48 hours of ICU admission, and the outcome was defined as in-hospital mortality. The dataset was randomly divided into a 85:15 ratio for training and testing. The Generative Adversarial Nets for Inference of Individualized Treatment Effects (GANITE) model was trained using the training dataset, and the model's performance was evaluated using the test dataset. Results: In the training set, the model demonstrated an accuracy and AUROC of 0.88 and 0.89, respectively, while in the test set, it showed 0.84 and 0.82. The probability change of average in-hospital mortality with CRRT treatment for all severe acidosis patients was +15% and +14% in the train and test sets, respectively. However, in the group that underwent CRRT, the application of CRRT resulted in an average reduction of in-hospital mortality probability by -13% in both the train and test sets. Age and creatinine levels in subjects experiencing a reduction in in-hospital mortality with CRRT were higher than the overall population data. Conclusions: Developed model could be expected to aid decision-making in the future application of CRRT treatment.