基于耦合关系的医生用药异常分析

Abstract

近些年来，医患关系受到广泛关注。如何准确地挖掘异常用药是制约医生和缓减医患关系的关键。本文提出了一种检测医生用药异常的总体框架。该框架集成处方数据的耦合相似度度量和变色龙聚类算法，并分为三个阶段，定性分析，定量分析和异常检测。在真实处方数据上进行了充分的实验，实验验证了该框架能够有效地检测出医生用药异常。 In recent years, the doctor-patient relationship has received extensive attention. How to exactly mine the abnormal use of drugs is the vital to constrain doctors and relieve the doctor-patient relationship. This paper presents a general framework for detecting the abnormal use of drugs. The framework integrates coupled similarity and Chameleon clustering algorithm, and consists of three stages, qualitative analysis, quantitative analysis and abnormal detection. We conduct extensive experiments on real-world prescription data. The experiments evaluate that the framework can efficiently detect abnormal use of drugs.