零膨胀模型在药品安全性计数资料分析中的应用

Application of Zero-inflated Models in Drug Safety Analysis of Count Data

零膨胀模型是一类用于处理存在零膨胀现象计数资料的统计模型,广泛应用于保险费率厘定、车祸发生频次等研究。药品安全性分析中的计数资料可能存在零膨胀现象。poisson回归模型或负二项回归模型用于拟合存在零膨胀现象的计数数据可能存在拟合不良的问题。零膨胀模型可能是一类更优的替代模型,且已有相关实例研究报道。将零膨胀模型用于不良事件/反应发生频次的影响因素分析,可以有效处理数据中可能存在的零膨胀现象,对于提高参数估计精度、探索药品(特别是中药注射剂)安全性事件的发生规律、识别重要危险因素以及预测等均具有重要意义。

Zero-inflated model is a statistical model for handling zero-inflated count data.Its widely used in modeling insurance data,investi- gating the influence of curbs on single-vehicle crash injury severity and so on.Zero-inflated phenomenon may existed in drug safety analysis of count data.There is the problem of poor fit in modeling zero-inflated count data by poisson regression or negative binomial regression model. Zero-inflated model may be a better alternative and the related examples have been reported.The zero-inflated models could be used to model adverse events or adverse reactions count data.It consider the zero-inflated phenomenon and Improve accuracy of parameter estimation.There is great significance in exploring drugs(especially traditional Chinese medicine injections)security events occurrence,identifing important risk factors and so on.