Objective:To construct a risk classification model for adverse events of medical consumables,so as to achieve automatic evaluation for risk level of such events,and enhance the capability for risk management of adverse events of medical consumables,and ensure the safety of medical apparatuses.Methods:The data of adverse events of medical consumables of 370 cases of 148 types that were reported by Shandong Provincial Third Hospital from 2020 to 2023 were selected,and they were divided into high-risk and low-risk types.Eight key factors of them,which included the number of cases,injury level,type of registration certificate,with source and without source,high-value and low-value,domestic and imports,product classification,and risk levels,were counted to form a dataset.K-nearest neighbor(KNN),support vector machine(SVM)and decision tree algorithms in machine learning were used to construct a risk classification model for adverse events of medical consumables.The data of 12 adverse events of medical consumables of 5 types of our hospital,which were newly reported in 2024,were integrated for their parameters.Then,the accuracy rate and prediction performance of the model were further analyzed.Results:By comparing the KNN,SVM and decision tree algorithm models,the effect of SVM algorithm model was better,and its accuracy rate was 90.54%,and its area under curve(AUC)value of the receiver operating characteristic(ROC)curve was 0.944,and its Kolmogorov-Smirnov(KS)test value was 0.808.The model had favorable predictive performance.The results,that invoked SVM algorithm model to conduct verification of actual prediction for 12 adverse events of medical consumables of 5 types,indicated it was same between predictive outcomes and risk levels of manual evaluation.Conclusion:The risk classification model of adverse events of medical consumables has established an operational model for assessing the risk level of such events,which can assist monitoring personnel for adverse event of medical apparatuses to quickly and accurately find risk signals of adverse events of medical consumables,and improve the monitoring capability of them for these adverse events.

Objective:To construct a risk classification model for adverse events of medical consumables,so as to achieve automatic evaluation for risk level of such events,and enhance the capability for risk management of adverse events of medical consumables,and ensure the safety of medical apparatuses.Methods:The data of adverse events of medical consumables of 370 cases of 148 types that were reported by Shandong Provincial Third Hospital from 2020 to 2023 were selected,and they were divided into high-risk and low-risk types.Eight key factors of them,which included the number of cases,injury level,type of registration certificate,with source and without source,high-value and low-value,domestic and imports,product classification,and risk levels,were counted to form a dataset.K-nearest neighbor(KNN),support vector machine(SVM)and decision tree algorithms in machine learning were used to construct a risk classification model for adverse events of medical consumables.The data of 12 adverse events of medical consumables of 5 types of our hospital,which were newly reported in 2024,were integrated for their parameters.Then,the accuracy rate and prediction performance of the model were further analyzed.Results:By comparing the KNN,SVM and decision tree algorithm models,the effect of SVM algorithm model was better,and its accuracy rate was 90.54%,and its area under curve(AUC)value of the receiver operating characteristic(ROC)curve was 0.944,and its Kolmogorov-Smirnov(KS)test value was 0.808.The model had favorable predictive performance.The results,that invoked SVM algorithm model to conduct verification of actual prediction for 12 adverse events of medical consumables of 5 types,indicated it was same between predictive outcomes and risk levels of manual evaluation.Conclusion:The risk classification model of adverse events of medical consumables has established an operational model for assessing the risk level of such events,which can assist monitoring personnel for adverse event of medical apparatuses to quickly and accurately find risk signals of adverse events of medical consumables,and improve the monitoring capability of them for these adverse events.