ABSTRACT
For the condition to be caught early, it is essential to find the genes linked to pediatric cardiomyopathy. In recent years, machine learning approaches have been applied to predict disease-related genes in various ways. It is necessary to increase the effectiveness of current techniques for predicting disease genes. Furthermore, the bulk of machine learning-based illness gene prediction algorithms are unable to identify indirect correlations between gene features. While understanding the relationship between the genes can improve the diagnosis of disease at the earliest. We propose a deep learning framework in this paper, named N2V- LSTM (Node2Vec – Long Short-Term Memory) to predict genes associated with pediatric cardiomyopathy disease. The Microarray gene data was converted to a graph and information related to genes was obtained using the Node2Vec method. The Embedded Vector was generated with learned features for each gene trained with LSTM. N2V-LSTM framework outperforms existing methods significantly.
