基于人工智能构建非危重型新型冠状病毒肺炎的中医辅助决策模型

Building a Traditional Chinese Medicine-assisted Decision-making Model for Non-severe Cases of COVID-19 Based on Artificial Intelligence

目的使用人工智能模型构建非危重型新型冠状病毒肺炎的中医辅助决策模型,协助医生进行诊断。方法收集2022年12月至2023年6月于天津中医药大学第一附属医院呼吸科、感染科、急诊科就诊的新冠病毒肺炎患者病历314份,以此建立数据集。将314个数据样本按7∶3比例分为训练集和测试集,利用随机森林、支持向量机、LightGBM和K最近邻分别构建非危重型新型冠状病毒肺炎辨证诊断模型,以精确率、召回率、f1分数、AUC值等作为模型的评价指标。利用4种人工智能模型构建中医辅助决策软件。结果随机森林、支持向量机、LightGBM、K最近邻的精确率分别为0.94、0.95、0.89、0.90;召回率分别为0.94、0.95、0.88、0.90;f1分数分别为0.94、0.95、0.88、0.90;AUC值分别为0.99、0.99、0.99、0.99。结论支持向量机模型准确率最高,更适用于构建非危重型新型冠状病毒肺炎的中医辅助决策模型,中医辅助决策软件能提高临床诊断效率。

Objective:To build a traditional Chinese medicine(TCM)-assisted decision-making model for nonsevere cases of COVID-19 using artificial intelligence.Methods:A dataset was created by collecting medical re⁃cords of 314 patients with COVID-19 from the Respiratory Department,Infectious Disease Department,and Emer⁃gency Department of the First Affiliated Hospital of Tianjin University of Traditional Chinese Medicine from De⁃cember 2022 to June 2023.The dataset was split into a training set and a test set in a 7∶3 ratio.Non-severe COV⁃ID-19 diagnostic models were built using Random Forest,Support Vector Machine(SVM),LightGBM,and K-Near⁃est Neighbors,with evaluation metrics including precision,recall,F1 score,and AUC value.Results:The preci⁃sion of Random Forest,SVM,LightGBM,and K-Nearest Neighbors were 0.94,0.95,0.89,and 0.90,respectively.The recall values were 0.94,0.95,0.88,and 0.90,respectively,while f1 scores were 0.94,0.95,0.88,and 0.90,re⁃spectively.The AUC values were 0.99 for all models.Conclusion:The SVM model demonstrated the highest accu⁃racy and is more suitable for building a TCM-assisted decision-making model for non-severe cases of COVID-19.