Missing value filling method of juvenile myopia prediction system and system using method

Abstract

The invention discloses a missing value filling method of a juvenile myopia prediction system. The missing value filling method comprises the following steps: filling non-diopter data in ophthalmic data; screening a first-grade diopter data sample and a second-grade diopter non-missing data sample; taking the second-grade diopter as label data to be fitted, and selecting features from the first-grade diopter data to obtain a feature subset; utilizing a machine learning method to construct a regression model for fitting; selecting a machine learning model with the optimal fitting effect; inputting the data sample with the second-grade diopter missing into the model, and performing filling with the predicted value to obtain second-grade complete diopter data, and so on; and filling the diopter data of the next grade by utilizing the diopter data of the current grade. The invention also provides the juvenile myopia prediction system using the method, and a management platform. The methodis characterized in that the GBRT is introduced to fill the missing data, so the filling result is closer to the real condition, and the prediction accuracy after many years can be improved.

Description

technical field [0001] The invention belongs to the technical field of data mining and machine learning, and in particular relates to a method for filling missing values ​​of a myopia prediction system for adolescents and children and a system using the method. Background technique [0002] The prediction of myopia in adolescents is based on the data analysis of children's ophthalmology cohort data. Since this data is a prospective population cohort data, the data involves the case investigation data of children in each year of primary school. In practice, except for the data of the first grade of primary school In addition to completeness, the data of the other grades are seriously missing, and a large amount of available data information has been lost. In terms of cohort data, there is no effective and complete method for filling missing data. In the existing medical data research, most of the data filling methods such as mean value, mode, and multi-digit number are used, ...

AI Technical Summary

Problems solved by technology

[0002]Myopia prediction in adolescents and children is based on the data analysis of children ophthalmology cohort data, because this data is a prospective population cohort data, the data involves children in every year of primary school In practice, except for the relatively complete data of the first grade of elementary school, the data of other grades are seriously missing, and a large amount of available data information has been lost. In terms of cohort data, there is no effective and complete data. Missing value imputation method

In the existing medical data research, most of the data filling methods such as mean value, mode, and multi-digit number are used, but they are not suitable for children's eye data filling, because the diopter data is the vision detection data of children after mydriasis, which belongs to Objective data, using conventional methods to fill has no practical significance, and the effect is not ideal

Images