Application Of Comprehensive Evaluation Model Based On Factor Space In The Selection Of Anti-breast Cancer Drug

With the continuous in-depth research on machine learning and the selection of cancer drug candidates,it has become one of the mainstream of drug research and development to use machine learning to process medical data to help develop drugs.At the same time,the application of factor space theory in information analysis and data processing is also progressing.The selection of anti-breast cancer drug candidates based on factor space theory has great research value.This paper focuses on the feature selection of molecular descriptors and optimization of drug selection of anti-breast cancer drug candidates,and the main achievements are as follows:Firstly,this paper introduced the targets,pharmacokinetic properties and safety of anti-breast cancer drug candidates,introduced and analyzed the above indicators as constraint conditions,and then conducted data preprocessing and preliminary selection of a large number of molecular descriptor data of candidate drugs.Secondly,the feature selection method based on factor space theory was applied to the field of anti-breast cancer drug candidate selection for the first time in this paper.Six kinds of feature data corresponding to six constraint indicators were selected by this feature selection method,which effectively improved the meaning expression of the feature data and provided the analysis of feature data for the selection of anti-breast cancer drug candidate,which was conducive to the targeted drug research and development.Then,this paper uses Light GBM,XGBoost,SVM and random forest prediction models to predict the six indicators.By comparing the accuracy of the prediction results,the six prediction models with the best training effect are selected,and the six prediction models are combined to construct a complex objective function,which makes a foundation for the optimization model for solving the complex objective function proposed below.Finally,a fuzzy multi-population particle swarm optimization model based on factor space is proposed to solve the problem that the optimization model takes too long due to the complexity of the objective function.The fuzzy comprehensive evaluation method is used to provide better convergence direction for particle swarm so that the model can find the optimal solution faster.Meanwhile,the multi-population method is used to improve the model’s defect that it is easy to fall into the local optimal solution.Experiments show that for complex objective function problems,compared with traditional particle swarm optimization algorithm and genetic algorithm,the fuzzy multi-population particle swarm optimization algorithm proposed in this paper has higher solving efficiency.