论文部分内容阅读
目的比较决策树和神经网络模型在苦碟子注射液(碟脉灵)自发呈报系统(SRS)数据库信号预警分析中的应用准确性和价值,明确碟脉灵ADR信号的发生特征。方法选取2007年1月至2012年12月国家不良反应监测中心SRS数据库中的碟脉灵ADR信息,以此为基础构建了碟脉灵不良反应数据库,分别采用决策树和神经网络两种方法对碟脉灵ADR信号进行预测模型建构和提升度检验。结果 (1)碟脉灵ADR发生的可能相关因素,决策树模型显示按重要性排序依次为发生年度、剂量、原发疾病影响、触发时间和年龄,神经网络模型显示重要性排序依次为发生季节、发生年度、年龄、用药次数和触发时间;(2)预测准确度方面,决策树模型为53.93%,神经网络模型50.4%;(3)模型提升度方面,以最常见的不良反应“皮疹”为例进行检测,决策树模型比神经网络模型提升度阈值下降更平缓且表现相对稳定。结论基于现有数据,苦碟子注射液ADR的发生很可能与年龄和触发时间相关;在预测准确度和提升度方面,决策树模型均优于神经网络模型;两者比较决策树可能更适合用于对于碟脉灵现有SRS系统数据的分析。
Objective To compare the accuracy and value of decision tree and neural network model in early warning analysis of SRD database and to identify the occurrence characteristics of ADR signal in Dishouling. Methods The ADR information of Diemailin from the National Adverse Reaction Monitoring Center (SRS) database from January 2007 to December 2012 was selected. Based on this data, an ADR database was constructed. Two methods, decision tree and neural network, Prediction model construction and ascension test of Diemail Ling ADR signal. Results (1) The possible related factors of ADR occurrence in Diemailing, the decision tree model showed that the order of importance was followed by occurrence year, dose, the influence of primary disease, triggering time and age, and the neural network model showed the order of importance was occurrence season (2) In prediction accuracy, the decision tree model was 53.93% and the neural network model was 50.4%. (3) In terms of model improvement, the most common adverse reactions were rash "As an example for testing, the decision tree model is more stable and the performance is more stable than the neural network model. Conclusion Based on the available data, the occurrence of ADR in Kudiezi injection is likely to be related to age and triggering time. The decision tree model is superior to the neural network model in predicting the accuracy and promotion degree. The comparison between decision tree may be more suitable Analysis of existing SRS system data for Diemailing.