Information prediction model and construction method and construction device thereof

Abstract

The invention provides an information prediction model and a construction method and device thereof. The construction method of the information prediction model comprises the steps of obtaining a sample set; defining event result parameters and independent variables; based on the sample set, non-linear independent variables having a non-linear relationship with the event result parameters are determined through a single-variable spline regression model, and the node number of each non-linear independent variable is determined; based on the sample set, determining at least one linear independent variable related to the event result parameter through a linear regression mode; and constructing an information prediction model based on the determined non-linear independent variables, the node number of each non-linear independent variable and the linear independent variables. According to the technical scheme, by screening out the independent variables with linear or nonlinear influence and comprehensively establishing the model, compared with an existing model established independently through linear variables or nonlinear variables, the established model greatly improves the prediction sensitivity and specificity.

Description

technical field [0001] The present invention relates to the field of information prediction, in particular, to an information prediction model, a construction method and a construction device thereof. Background technique [0002] In statistics, Linear Regression is a regression analysis that uses a least squares function called a linear regression equation to model the relationship between one or more independent variables and event outcome parameters. In bioinformatics, linear regression analysis is widely used in disease diagnosis, efficacy prediction, and prognosis prediction. According to the different types of event outcome variables, linear regression can be divided into cox model that can fit prognosis and logistic regression model that the outcome is a binary variable. In fact, an important assumption of most linear regression models is that the independent variables and event outcome parameters are linearly related, which is difficult to meet in practice. A commo...

AI Technical Summary

Problems solved by technology

[0003] However, the multiple regression model established by the linear regression analysis method can only reflect a single linear relationship between multiple independent variables and event result parameters, while restricted cubic splines can only fit the nonlinear relationship between a single independent variable and event result parameters. relation

Therefore, the accuracy of the information prediction model established only through linear regression analysis and restricted cubic splines is low, which cannot meet daily needs.

Images