資料採礦技術應用於全民健保資料庫分析勃起功能障礙病患心房顫動相關因素之研究

民國102年國人十大死因心臟疾病居第二位佔國人死因的15%，許多研究指出患有心房顫動者之缺血性中風的發生率及心血管疾病的致死率皆高於未患有心房顫動的人，且國內外皆有研究指出勃起功能障礙發生在許多心血管疾病之前也提高男性罹患心血管疾病與腦中風的機會。本研究以2001年至2009年發行之健保資料庫中18歲以上的男性為資料分析及建模之來源，並建立一套全民健康保險研究資料庫分析的標準作業流程，應用存活分析、類神經網路、決策樹、支援向量機與隨機森林等方法，找出勃起功能障礙與心房顫動有無相關性及男性罹患心房顫動之影響因子。由實證分析結果得知，勃起功能障礙不會影響男性心房顫動之發生，建模方法則以隨機森林較佳，本研究之研究結果希望能提供醫學方面的研究做為參考。

關鍵字

全民健康保險研究資料庫；資料採礦；心房顫動；勃起功能障礙

並列摘要

In 2013, heart disease was the second of the top ten leading causes of death and occupied 15% of the cause of death in Taiwan. Many researches indicate that patients with atrial fibrillation are easier to get ischemic stroke also have higher chance to die from cardiovascular disease compare with people without atrial fibrillation. There are also some studies show that the erectile dysfunction happened earlier than cardiovascular diseases and also increased the chance to ill cardiovascular diseases and stroke. In this study, the data came from National Health Insurance Research Database and selected the men who are 18 years old and above from 2001 to 2009 as the sample of data analyze and modeling. This study also used data mining technology to establish Standard Operation Procedure of National Health Insurance Research Database and built various models such as survival analysis, artificial neural network, decision tree, support vector machine and random forest to find out the correlation between erectile dysfunction and atrial fibrillation. And find influential factors for male suffer from atrial fibrillation. The results showed that erectile dysfunction won't cause any affection to atrial fibrillation and the random forest is the best method to predict men who will suffer from atrial fibrillation. In conclusion, we hope the result of the study can provide the reference for medical research.