基于宿主一微生物网络的疾病预测算法

打开文本图片集

中图分类号：TP391 文献标志码：A

Disease Prediction Algorithm Based on Host-microbial Network

LUO Wei-wen，WU Shun-yao

Abstract： An algorithm for disease prediction based on a host-microbe dual-layer network was proposed to address the issue microbial interactions being overlooked in disease prediction methods. Host-microbe networks were constructed by microbial sequencing data phylogenetic trees. The restart rom walk algorithm was applied to propagate signals across the network， disease-associated microbial motifs were identified with prior microbial knowledge. These motifs were converted into predictive features incorporated into a model for disease prediction. The experimental results show that in the detection colorectal cancer，the AUC value logistic regression increases from O.84 to O.90， the AUC value rom forest increases from O.88 to 0.92.

Keywords： network propagation;microbial markers;disease prediction;host state

随着对肠道微生物群研究的深人，对肠道微生物群的作用也有新的了解。（剩余6251字）