Precision Medical Research Based On Heterogeneity Model

Everything has its own uniqueness.Every individual has both commonness and uniqueness.With the rise of precision medicine,machine learning model has been introduced into the field of medical diagnosis,but this model only abstractly summarizes the overall trend of development.When the overall trend is restored to the individual,the conclusion will often be biased.This paper proposes a heterogeneity model based on hybrid model and machine learning model,which not only can effectively explain the heterogeneous population,but also can make full use of the efficient and accurate characteristics of machine learning,and has a good fitting effect on the data.Based on the data set of infant birth indicators,this paper uses regression analysis,machine learning,hybrid model and other analysis methods.The results of model error scale and discriminant analysis show that the discriminant accuracy of random forest reaches 91%,and the classification effect is the best.Statistical descriptive analysis shows that there is a high degree of heterogeneity in each subject’s mother,so a linear mixed model is introduced to deal with individual heterogeneity.Maximum likelihood estimation combined with EM algorithm is used to estimate the parameters of linear mixed model,and random effects ib are used as individual heterogeneity.Similarly,the generalized linear mixed model is introduced to deal with classification data,and the penalized likelihood estimation is used to estimate the parameters.By introducing heterogeneity into machine learning model,all heterogeneity models are better than the original model.The square error of model and SSE are only 1/4 or less of the original model.Support vector machine regression has the best improvement effect,and the results of cross validation are the smallest,the error is the smallest and only 1/8 of the original model.Machine learning model is introduced to solve individual heterogeneity.Heterogeneity model is greatly improved in discriminant analysis.Decision tree is upgraded the most,from 69% to 91%.Random forest classification has the best effect and the discriminant accuracy is 100%.Therefore,the extracted heterogeneous variables play an important role in optimizing the model,significantly improving the error variance and the first-class errors.Clustering analysis of the population is carried out according to heterogeneous variables,and clustering with heterogeneous variables is added to make clustering more concise and effective.