干眼的中医智能辅助诊断模型研究OA

Objective The accuracy of traditional Chinese medicine syndrome differentiation for dry eye needs to be improved,and there are shortcomings.This study aimed to construct a traditional Chinese medicine intelligence-assisted diagnosis model for dry eye based on objective information.Methods By utilizing 14 machine learning techniques,the model can provide the corre-sponding diagnostic judgments for new data,thereby optimizing the diagnosis and treatment strategies of traditional Chinese medi-cine for dry eye.Results The XGBoost algorithm demonstrated excellent performance on the binary classification model of dry eye deficiency and excess syndromes.The classification model for specific dry eye syndrome types constructed by using 14 machine learning methods had poor performance while the Extra-Trees algorithm showed better performance in the tertiary classification model of dry eye with solid evidence types,and the CatBoost algorithm showed better performance in the tertiary classification model of dry eye with deficiency syndromes.Conclusion Compared with the direct use of machine learning to construct a six-classification model for the specific evidence type of dry eye,on the basis of the two-classification model for the deficiency and excess syndromes,the most appropriate algorithm should be used to construct the model for the different objectives of the dry eye syndrome classification task to ensure that the performance of the model achieves the best results.