The discipline of disease prognosis has recently garnered significant interest. In order to convert the input prediction signals into the estimated diseases for each patient, it is necessary to train a suitable classifier. However, to achieve accurate prediction outcomes, existing machine learning methods primarily depend on a substantial quantity of EMR training data that has been manually labeled. On the other hand process of converting data from different fields into graph topologies has become increasingly popular in recent years. Graph Neural Networks (GNNs) have become the widely accepted and commonly used tool for machine learning problems involving graphs. Additionally, neural network models such as the Multi-Layer Perceptron (MLP) can be represented as graphs. Here Graph Convolution Network (GCN) model, along with its applications like link prediction and node classification, is experimented on medical records that consist of symptoms and disease information. Results are assessed on accuracy, precision, and recall on all verticals, like train, validate, and test.