Deep Neural Network-Empowered Polygenic Disease Prediction on Cardiovascular Diseases

There is a growing area of research showing that complex diseases have been found to be caused by significant genetic variants, that is, multiple changes to the normal genome across multiple locations. Predicting the risk of these diseases is difficult due to the limited knowledge of variant causation and the leading approaches currently focus on gene-disease association. In this work, we propose a cardiovascular disease based analysis using an enhanced indel deep neural network (EI-DNN), comprised of two deep neural networks using novel indel variants alongside conventional variant sites to predict disease risk. This model uses two deep neural networks in series, the first to process indel data and the second to provide the risk score. The experiments were performed on our proposed algorithm using the MGRB database and compared against a conventional PRS calculation and a single DNN algorithm. The experimental results validate the effectiveness of the proposed method and highlight the capabilities with combining indel variants to enhance disease prediction.