Research on Classification Model of Visual Function Abnormality for Auxiliary Diagnosis Based on Machine Learning

Abstract

In order to help ophthalmology clinical staff make correct judgments on patients' visual function abnormalities in a shorter period of time, and to alleviate to a certain extent the pressure of consultation due to the mismatch of medical resources, this paper presents a study on the design of a classification model based on machine learning to assist in diagnosing visual function abnormalities. Using visual function data from major optometric medical centers, a predictive classification model was developed using six algorithms (K-nearest neighbor algorithm, decision tree, random forest, support vector machine, plain Bayes, and XGBoost), and the accuracy, precision, recall, and F1 parameters were calculated to compare the training effect of the model under each algorithm, and the data set was validated using cross-validation. The results showed that the models trained with random forest could classify the “set” and “adjustment” labels with clinical applicability.