運用資料探勘分析代謝症候群相關疾病

Abstract

隨著社會型態的改變，國人飲食攝取問題從營養不足轉變成營養過剩，加上現代人工作忙碌，造成新的健康問題接著出現，而「代謝症候群」就是其中之一。代謝症候群是指腰圍過粗、血壓和血糖偏高、血脂有點異常的群聚現象，但數值尚未達到慢性疾病的標準；換句話說，若有代謝症候群的警訊出現可說是腦血管疾病、心臟病、糖尿病、腎病變、高血壓等慢性疾病的前兆。 本研究運用資料探勘技術，針對國內某醫院 H 醫院的健康檢查中心，藉由 H 醫院的健康檢查中心之顧客健康檢查資料進行研究，由顧客個人健檢項目、後續追蹤就醫檢查報告資料，定義出與代謝症候群指標等屬性的關係，再分別用羅吉斯迴歸分析、決策樹、類神經網路來建立模型，最後，依據研究結果找出相關疾病後再擬定建議，建立健康管理模式，作為 H 醫院後續經營之參考。

Along with the social state change, people's dietary intakes have been changed from the lack of nutrition into excess nutrients; coupled with the busy work of modern people, new healthy problems then appear, and "metabolic syndrome" is one of them. Metabolic syndrome refers to the following: the waist is too thick; high blood pressure and blood sugar; blood lipid is unusual, but the value has not reached the standard of chronic diseases yet. In other words, the metabolic syndrome indicates danger signal for blood vessel of brain disease (cardiovascular disease), heart diseases, diabetes, the nephrosis changes, hypertension and so on, which are the omens of chronic diseases. This study collects the data of a domestic hospital, H hospital health check (examination) center, including the customer health check data and follow-up medical examination report data. The collected data are analyzed to discover the relationship between metabolic syndrome and other attributes. Moreover, this study applies the Logistic regression analysis, decision tree, neural network to build the prediction model for discovering the relevant disease. The results will be used to develop recommendations to establish a health management model, as a reference for the follow-up operation of H Hospital.