NCH-DDA: Neighborhood contrastive learning heterogeneous network for drug-disease association prediction

Peiliang Zhang, Chao Che, Bo Jin, Jingling Yuan, Ruixin Li, Yongjun Zhu

Abstract: Highlights•Neighborhood contrastive model for heterogeneous representation and DDA prediction.•MNFEM and SNFEM capture the important features in multiple potential spaces.•Contrast learning aggregated generic neighborhood features of drugs and diseases.•Presented a mixing optimization strategy to reduce the dependence on prior knowledge.•Four of the top five drugs recommended for AD by NCH-DDA were proved to be effective.