應用資料探勘技術於高血壓疾病之預測 ─以門診資料為例

Application of Data Mining Techniques on the Prediction of Hypertension: Using National Health Insurance Outpatient Database.

林欣慧(HSIN-HUI LIN)

指導教授 ： 李緒東 ； 吳泰熙

國立臺北大學/商學院/企業管理學系/碩士(2014年)

摘要

高血壓為常見於國人之慢性疾病，且容易併發其他疾病，若能根據全民健保資料庫過往之病例關聯，分析出與自發性高血壓高度關聯之疾病類別，將有助於民眾或醫事人員提早注意罹病風險。本研究旨在透過資料探勘技術，應用關聯分析、序列分析與決策樹分析來探討自發性高血壓與其潛在之疾病關聯，並結合醫療領域知識，提出相關建議。
本研究使用資料探勘之關聯法則、序列分析與決策樹方法，進行逐步分析健保資料庫門診檔。首先以關聯法則找出與自發性高血壓相關之疾病，再以關聯法則之結果進一步進行時序分析，最後挑選特定疾病作為決策樹分析之對象。
研究結果顯示：關聯分析得與自發性高血壓相關之十類疾病為：「其他內分泌腺疾病」、「其他新陳代謝失調及免疫失調」、「精神官能症、人格違常及其他非精神病心理疾患」、「缺血性心臟病」、「腦血管疾病」、「急性呼吸道感染」、「食道、胃及十二指腸之疾病」、「消化系統之其他疾病」、「關節病變及有關疾患」、「背部除外之風濕症」。序列分析顯示「其他內分泌腺疾病」之第二型糖尿病相關疾病與「腦血管疾病」之腦血管疾病後遺症患者為罹患自發性高血壓之高危險群，且上述二者多發生於自發性高血壓疾病之前。

關鍵字

健保資料庫 ； 高血壓 ； 決策樹 ； 序列分析 ； 關聯法則 ； 資料探勘

並列摘要

Objectives: Hypertension is a common chronic disease in Taiwan, which leads to many kinds of complications. Knowing the diseases which are highly related to hypertension would be important for patients to decrease the risk of having hypertension. The present study shows the association between hypertension and other chronic hypertension complications. Based on association rules, sequential analysis and decision tree, this study performs data mining on the National Health Insurance research database, which developed by the National Health Research Institute in Taiwan. Furthermore, this study discusses the results of the data mining rules with medical practical.

Methods: The study samples are collected from the National Health Insurance research database from 2006 to 2010. After data cleaning, the total samples are 3583,645. Association rules are first used in the study. Based on the results of association rules, sequential analysis is developed in the following part. And decision tree is applied to particular hypertension complication.

Results and conclusions: According to the results of association rules, the diseases related to hypertension are cerebrovascular disease, Ischemic heart disease, type II diabetes mellitus, etc. Besides, according to the results of sequential analysis, type II diabetes mellitus and cerebrovascular disease are highly related to hypertension, and they usually occurre before hypertension.

並列關鍵字

Hypertension ； Decision tree ； Sequential analysis ； Association rule ； Data mining ； NHIRD

參考文獻

周歆凱、蘇喜、黃興進、蔡明足、翁林仲(2006)。運用決策樹技術探討急診病患醫療費用之消耗。台灣衛誌，25(5)。

莊宗南、龔榮源、陳俊龍(2006)。以資料探勘技術建立病患就醫導引─以胃腸科病患為例。醫療資訊雜誌，15(1)，17-34。

陳銘樹、王建智、王麗燕(2008)。應用決策樹演算法以探究高科技員工潛在的糖尿病危險因子。健康管理學刊，6(2)。

曾淵如(1998)。高血壓及其併發症之診斷與處置。取自http://catalog.digitalarchives.tw/item/00/63/37/d3.html

Agrawal, R., Imieliński, T., & Swami, A. (1993, June). Mining association rules between sets of items in large databases. In ACM SIGMOD Record (Vol. 22, No. 2, pp. 207-216). ACM.