Gene network relationship fused anti-cancer cell sensitivity prediction method, system and equipment and medium

A method, system, device and medium for predicting anti-cancer cell sensitivity by fusing a gene network relationship, the method comprising: using a vector composed of expression quantities of each gene of a cell line as an original gene expression feature of the cell line, and using a self-encoder to compress a low-dimensional implicit vector as a gene expression feature of the cell line, performing graph self-coding on the gene interaction network to obtain gene interaction features, calculating gene network features according to the gene expression features and the gene interaction features, establishing a drug compound molecular graph to obtain features of atoms in drug compound molecules, and obtaining predicted anti-cancer cell sensitivity according to an EIGA model; the system, the equipment and the medium are used for realizing the anti-cancer cell sensitivity prediction method fusing the gene network relationship. According to the method, by designing a new algorithm, the interaction relationship be