| abstract |
A drug molecule screening method and system, comprising: collecting drug molecule data related to a specific disease, preprocessing the data, calculating its encoding vector and drug physicochemical properties; constructing and training an AI model based on a conditional variational autoencoder, The combination of the encoding vector and the drug physicochemical properties of the molecule is used as the input layer of the model, which is converted into the hidden layer encoding vector through the encoding layer of the model, and then generates the possible drug molecular structure through the decoding layer of the model. During the model training process, the gradient descent algorithm is used. The model loss function is minimized, and the weight parameters of the neural network structure of the iterative encoding layer and decoding layer are continuously updated; according to the trained conditional variational autoencoder model, potential drug molecules for curing specific diseases are generated; the above drug molecule screening The method and system also utilize the data of the drug physicochemical properties of the compound molecule, and the drug physicochemical properties have a great correlation with whether the compound can be finally formulated into a drug, so as to improve its druggability. |