The primary aim of this work is to utilize Convolutional Neural Network (CNN) models, which are based on deep learning, to analyze numerical data frequently employed in the medical field. This endeavour is crucial because liver failure ranks as the second most prevalent disease worldwide, and the number of donors is insufficient to meet the demand from recipients. It is essential to identify patients who have a high likelihood of survival and improvement following liver transplantation. To accomplish this, a dataset containing only numerical values for liver transplantation must be transformed into image data to take advantage of Convolutional Neural Network's capabilities. To achieve this, the raw data is first normalized and then subjected to a logarithmic transformation. Each normalized feature is then assigned to a specific area of the image grid. This generates images with different brightness zones based on the numerical value of each feature. The Grey Wolf Optimizer is for training the hyperparameters of the Convolutional Neural Network models. The first Convolutional Neural Network model has an accuracy rate of 91%, making it the best model. Eventually, the Ensemble Exponential Weighted Average Grey Wolf Optimization-based Convolutional Neural Network approach successfully classifies the survival rate of patients with accuracy of 93% when ensemble learning is employed.