以實驗室微生物資料為基礎的院內泌尿道感染監視系統之研究

The Research of A Laboratory based Intelligent Nosocomial Urinary Tract Infection Surveillance System

張博仁(Po-Ren Chang)

指導教授 ： 劉建財

臺北醫學大學/醫學院/醫學資訊研究所/碩士(2007年)

https://doi.org/10.6831/TMU.2007.00132

摘要

醫療技術的進步與醫療設備的推陳創新，使得人類的生命得以延長，但也間接引發不少醫療問題，其中較為嚴重的就為院內感染。院內感染不但造成病患住院天數增加、醫療資源的浪費，更加重病患的不適，甚至造成死亡。而院內感染最常見的就是泌尿道感染，排名為前一、二名。而實驗室資訊系統收集的病原菌資料更能準確描述病情的發展，如病原菌種的群集和區域位置等，若能由實驗室報告提早偵測出有院內泌尿道感染實在有迫切必要。

本研究目的將菌種排名表、菌種陽性率、菌種分離率、菌種與年齡層呈現圖形化及可分析化，此表可知道醫院菌種排名，提供年報、季報與月報同時呈現，讓管染管制人員一目了然院內菌種是否有變化，達到監視的目的。與利用貝氏定理（Bayesian Theorem）算出每支菌種的後置機率來監測護理站每日院內泌尿道感染的機率。與建立全院與護理站院內泌尿道感染的基準線來預測群聚感染的發生，讓感染管制人員到提早調查院內是否有泌尿道感染群聚發生，以免擴大感染而導致院內的大流行。

本研究方法有兩種一是利用每隻菌種的後置機率與菌種數量監視院內泌尿道感染的發生，另一種是以全院與護理站的基準線來監測群聚的發生。而呈現方式則利用資料倉儲以一維或多維分析、視覺化、報表等功能，使其具備可用性、簡單性、親切性來讓感染管制人員使用。

本研究結果對於院內泌尿道感染的群聚，是以全院平均值減1.5個標準差做為基準線，與以護理站平均值與後置機率大於0.75來偵測，結果則Sensitivity 為 0.75，而Specificity 為 0.78。而每日提供每個護理站院內泌尿道感染菌種的機率大小，由大排到小一目了然讓感染管制人員到現場調查，提早找出院內泌尿道感染以防止院內泌尿道感染的增加

關鍵字

院內泌尿道感染 ； 監視系統 ； 貝氏定理

並列摘要

Medical technology progress and medical equipment innovation enable the human life to lengthen。 But also indirect initiation many medical service question ,the comparatively serious one is nosocomial infection。 Not only the nosocomial infection creates sickness in hospital number of days to increase, the medical resources waste, serious illness trouble illness, even causes the death。 Nosocomial infecting the most common one is the urinary tract and infect. The laboratory information system collection disease germ material can describe the condition development accurately. For Example germ cluster and regional setting and so on. An urgent need that Nosocomial Urinary Tract Infection will be dected from the laboratory report.

The purpose of research has bacterial sorting list，bacterial positive rate, bacterial separation rate, between bacterial and age level appear the figure and can analyse that melt, this form can know the bacterial rank of the hospital, offer annual report, quarterly report and monthly magazine to appear at the same time, nosocomial infecting officer very clear institute the bacterial change, achieve the goal of monitoring to in charge of to let. With spend Bayesian Theorem calculate each probability of bacterial to Surveillance probability the urinary tract infect of nurse station.To build the base line hospital and nurse station of the Nosocomial Urinary Tract Infection to divine cluster event to come up, the nosocomial infecting officer can early discover cluster event to happen and stop nosocomial popular.

Research approach have two method ,first utilize posterior probability and bacterial number to surveillance the Nosocomial Urinary Tract Infection , another method utilize baseline to monitor of clustering to come. The way appears to utilize data warehouse one /multi-dimensional Analysis,Visualization, Reporting functions , make it possess usability, simplicity, friendliness with the nosocomial infecting office using.

Result of the Research this to clustering that urinary tract infection. Sensitivity is 0.75 , and Specificity is 0.78

並列關鍵字

Nosocomial Urinary Tract Infection ； Surveillance System ； Bayesian Theorem

參考文獻

甘偉志，黃政文.泌尿道感染。當代醫學.2005;32（1）:p10~16.

黃麗玲.某教學醫院室內空氣之微生物.公共衛生雜誌，2006;25(4): p315-321.

林炯璁，＜臨床微生物檢驗的資料挖掘研究－以偵測鮑氏不動桿菌院內感染為例＞，亞洲大學生物資訊研究所，碩士論文，民94.6。

院內感染經濟分析之回顧.院內感染控制雜誌. Available at: http://www.nics.org.tw/magazine/16/01/16-1-13.htm Accessed May 1 ,2007。

施秀，鍾琴玲，吳劍清，羅世慧，駱惠銘.院內泌尿道感染之調查表.院內感染控制雜誌.1998;8（1）:400-409.