DeepFlu: a deep learning approach for forecasting symptomatic influenza A infection based on pre-exposure gene expression

Zan, Anna; Xie, Zhong-Ru; Hsu, Yi-Chen; Chen, Yu-Hao; Lin, Tsung-Hsien; Chang, Yong-Shan; Chang, Kuan Y.

Background and Objective Not everyone gets sick after an exposure to influenza A viruses (IAV). Although KLRD1 has been identified as a potential biomarker for influenza susceptibility, it remains unclear whether forecasting symptomatic flu infection based on pre-exposure host gene expression might be possible. Method To examine this hypothesis, we developed DeepFlu using the state-of-the-art deep learning approach on the human gene expression data infected with IAV subtype H1N1 or H3N2 viruses to forecast who would catch the flu prior to an exposure to IAV. Results The results indicated that such forecast is possible and, in other words, gene expression could reflect the strength of host immunity. In the leave-one-person-out cross-validation, DeepFlu based on deep neural network outperformed the models using convolutional neural network, random forest, or support vector machine, achieving 70.0% accuracy, 0.787 AUROC, and 0.758 AUPR for H1N1 and 73.8% accuracy, 0.847 AUROC, and 0.901 AUPR for H3N2. In the external validation, DeepFlu also reached 71.4% accuracy, 0.700 AUROC, and 0.723 AUPR for H1N1 and 73.5% accuracy, 0.732 AUROC, and 0.749 AUPR for H3N2, surpassing the KLRD1 biomarker. In addition, DeepFlu which was trained only by pre-exposure data worked the best than by other time spans and mixed training data of H1N1 and H3N2 did not necessarily enhance prediction. DeepFlu is available at https://github.com/ntou-compbio/DeepFlu. Conclusions DeepFlu is a prognostic tool that can moderately recognize individuals susceptible to the flu and may help prevent the spread of IAV.