糖尿病临床诊断事件序列中频繁模式的发现算法

摘要

With the development of information construction process of the major hospitals, various production systems in the hos-pitals such as HIS (Hospital Information System),EMR (Electronic Medical Record System) have accumulated a large-scale clinical data.These clinical big data have a far-reaching significance for improving the quality of clinical care.Diabetes as a chronic disease, is apt to cause a variety of complications, such as kidney disease, eye disease and so on.In order to find out the rules of diabetes complications, this article firstly changes the historical data of diabetes diagnosis to event sequence, then propo-ses a frequent pattern discovery algorithm NFPS for the event sequence of clinical diabetes diagnosis based on the traditional algo-rithm SPADE.The proposed algorithm takes into account diabetes treatment interval, supports the frequent pattern discovery of the diabetes complications within the set time window.Experimental results show its effectiveness in the frequent pattern discovery of clinical diabetes complications.%随着各大医院信息化建设进程的不断推进，医院中的各生产系统如HIS（医院信息化系统）、EMR（电子病历系统）等已经积累了规模庞大的临床数据，这种临床大数据对于提升临床医疗质量有着深远的意义。糖尿病作为一种慢性病，容易引发多种并发症如肾病、眼病等。为了找出糖尿病并发症出现的规律，本文首先对糖尿病历史临床诊断数据进行事件序列化，然后对传统的SPADE算法进行改进，提出一种糖尿病临床诊断事件序列频繁模式发现算法NFPS，该算法考虑糖尿病治疗时间间隔，通过时间窗口的设定，支持对该时间窗口内糖尿病并发症频繁出现模式的发现。实验结果表明其在临床糖尿病并发症频繁模式发现上的有效性。