大學新生健檢結果的資料挖掘

Mining data from the results of freshmen health checkup

大學部新生體檢資料中，潛藏著一些未曾探索的重要資訊與知識。藉由資料挖掘 技術，可以從中找出學生體檢數值異常項目的組合。本研究以台灣中部某大學2005至 2007年新生體檢資料完整的11,667筆資料做分析，發現近乎一半的大一新生缺乏B型 肝炎表面抗體（48.9%），體重過重者（BMI≧23kg/m2）佔48.1%，收縮壓、舒張壓 偏高者分別為37.8%與15.8%，肝功能異常者（SGPT）佔25.7%，高三酸甘油酯症為 23.5%，高尿酸血症19.6%。運用關聯規則來描述體檢數值異常項目組合間的關聯性， 發現代謝症候群的症狀（體重過重、高血壓、高尿酸、高三酸甘油酯），易合併肝功 能指標SGPT的異常。研究結果建議各大學應對學生開設健康促進課程，以培育大學生 健康的生活方式；再者，對於大一入學時所做健康檢查數據有異常者，學校相關單位 得建檔以定期輔導追蹤，以免演變成疾病的發作或形成。

Lots of important information and knowledge are concealed in huge freshmen health checkup databases. To increase understanding the combination of abnormal items in freshmen health checkup, this study applied data mining techniques to discover those information. Data for this study came from 11,667 health checkup reports, collected in 2005-2007. The results of this study showed that nearly one half of subjects were negative HBsAb (48.9%), 48.1% subjects were overweight (BMI≧23kg/m2), 37.8% and 15.8% subjects had abnormal readings of systolic blood pressure and diastolic blood pressure respectively. Abnormal reading of SGPT was noted in 25.7% of the subjects. Among all of the subjects, 23.5% were noted to have hypertriglyceridemia, and 19.6% had hyperuricemia. According to the association rules analysis of data mining, the symptoms of metabolic syndrome (overweight, abnormal reading in blood pressure, or triglyceride, uric acid) were associated with the elevated SGPT. The results are useful in designing health promotion program to develop healthy lifestyle for young adults. Additionally, to prevent long-term risk of chronic diseases, follow-up the subjects whose health checkup had abnormal items might be warranted.