- 學位論文
應用資料探勘技術分析疾病間共現與因果關係–以失眠與心腦血管疾病為例
Applying Data Mining in Analyzing the Co-occurrence and the Causality between Insomnia and Cardiovascular Disease or Cerebrovascular Disease
摘要
睡眠是一種本能及基本需求，但國內外的研究，都有數據顯示有越來越多的人飽受失眠困擾。長期睡眠不足者除了影響心理層面，更可能會影響身體基本運作，導致罹患各式疾病機率增加。一般進行睡眠研究常運用睡眠日誌進行小規模資料收集資料庫，但此方式容易受限於區域或有樣本人數過少的問題，使得結論受到侷限。本研究以全民健康保險資料庫的抽樣歸人檔為資料來源，探討台灣地區失眠與心腦血管疾病之間的關聯。該資料庫取樣自全國民眾的就診資料，且具10年以上的紀錄，讓研究結果更具可靠性及代表性。 本研究分為兩大部分：流行病學分析及資料探勘分析。由健保資料庫中選取8,406病患當作實驗組，並以1:5的方式在其他病患中，選出42,030符合年齡及性別者當作對照組。流行病學方面，以生物統計方式分析台灣失眠人口的各項描述性資料，並選擇精神疾病、心血管疾病及腦血管疾病等目標疾病，探討失眠者對目標疾病的患病風險。資料探勘部分則嘗試運用貝氏網路技術，探討兩兩疾病間關聯程度，以及失眠與其他目標疾病間的因果關係。 研究發現失眠患者女性人數顯著高於男性，且女性罹患失眠的機率約為男性的1.32倍，高危險群主要介於51-60歲，而罹患失眠的男性平均年齡顯著高於女性。失眠與精神疾病高度相關(OR=5~9)，而腦血管疾病以輕微型中風關聯程度最高(OR=2.386)，心血管疾病則以冠狀動脈疾病最顯著相關(OR=1.578)。無論從流行病學或資料探勘分析，皆發現失眠與精神疾病和心腦血管疾病關係密切。
並列摘要
Sleep is a kind of instinct and basic needs. However, there were investigations showing that more and more people were suffering from insomnia. Chronic insomnia were more likely to affect the basic operation of the body, leading to increased risk of suffering from various diseases. Therefore, we aimed to provide the demographic data of insomnia in Taiwan and to determine whether insomnia is associated with cardiovascular disease or cerebrovascular disease, based on National Health Insurance Research Database which is a nationwide population-based database in Taiwan. This study was divided into two parts: Epidemiological Analysis and Data Mining Analysis. We identified a total of 8,406 patients with insomnia were included in our study cohort. We randomly selected 42,030 subjects (5 for each insomnia patient) from the remaining patients in the data set as our control group that matched with the study group by sex and age. We provided the demographic data of insomnia in the part of epidemiological analysis. In the part of data mining analysis, we used bayesian network as a tool to analyze the causality between insomnia and cardiovascular disease or cerebrovascular disease. There was significant gender difference in the prevalence of insomnia. Female/male ratios for insomnia are about 1.32. Female between 51 and 60 years are the high-risk group, and the average age of male is significantly greater than female. We concluded that patients with insomnia are at an increased risk for mental illness (OR=5~9), transient ischemic attack and vertebrobasilar insufficiency (OR=2.386), and coronary artery disease (OR=1.578).