摘要
目的通过构建基于临床数据中心的可疑呼吸道传染病发现与预测模型,实现对可疑传染病的发现与预测。方法选取某三甲医院的临床数据,基于历史传染病数据进行病历结构化建模,构建呼吸道传染病知识图谱,利用XGboost算法和知识图谱推理技术形成发现与预测合并决策模型,并使用医院历史数据做交叉验证,得到准确度较高的模型。结果发现与预测模型的平均查准率为92.55%,查全率为91.49%,综合F1值为92.01%,均优于单独的知识图谱模型或XGboost模型,将模型与医院的电子病历系统和临床辅助决策系统进行集成,应用于对真实临床病例的预测。结论该方法能够很好地针对新发可疑呼吸道传染病进行预测,辅助医院及时启动传染病应急预案,减少传染病发生早期时医务人员的感染概率。
Objective To construct a detection and prediction model for suspected respiratory infectious diseases(RIDs)based on clinical data center,and achieve the detection and prediction of suspected infectious diseases.Methods Clinical data were selected from a tertiary first-class hospital,structural modeling of medical records was constructed based on historical data of infectious diseases,knowledge map of RIDs was formulated,a combined decision model of detection and prediction was formed through XGboost algorithm and knowledge map reasoning technology,and cross validation based on hospital historical data was performed,a model with high accuracy was obtained.Results The average precision ratio of the detection and prediction model was 92.55%,with recall ratio of 91.49%and the comprehensive F1 test value of 92.01%,which were superior to the individual knowledge map model or XGboost model.The model was integrated with the hospital’s electronic medical record system and clinical decision support system for predicting real clinical cases.Conclusion This method can effectively predict emerging suspected RIDs,assist hospitals to initiate emergency plans timely for infectious diseases,and reduce the probability of infection among health care workers at the early stage of infectious diseases.
作者
陈廷寅
冯嵩
CHEN Ting-yin;FENG Song(Network Information Center,Xiangya Hospital,Central South University,Changsha 410008,China;National Clinical Research Center for Geriatric Disorders,Changsha 410008,China)
出处
《中国感染控制杂志》
CAS
CSCD
北大核心
2023年第8期964-971,共8页
Chinese Journal of Infection Control
基金
长株潭一体化医疗大数据应用基础支撑体系研究试点示范(202102-001)
移动医疗教育部中国移动联合实验室课题(2020MHL02015)。
关键词
可疑传染病
人工智能
知识图谱
预测模型
suspected infectious disease
artificial intelligence
knowledge mapping
prediction model
