A Deeper Understanding of Modular DNN in Predicting Ageing-Related Disease

Abstract

Ageing is a significant process happening in all humans and close related to health and lifetime. However, the mechanism of ageing is poorly understood. Getting to know about which specific genes control ageing-related diseases can be a great help of this mechanism. This paper focuses on using one of the most advanced machine learning methods nowadays to predict ageing related disease with large amount of genes. This paper finds a deeper relation behind the different datasets and encoders of modular DNN raised by Fabio Fabris’ group. With a deeper understanding of modular DNN, this paper is able to find a model with AUC value equal to 0.9732, which has a 10.65% improvement compared with former paper. With the results and final model of this paper, this paper can help scientists identify high-possible ageing-related genes with higher accuracy.