應用資料探勘之群集分析法於小兒氣喘表現型態

Identification of Childhood Asthma Phenotypes Using Cluster Analysis: a Data Mining

10.6200/TCMJ.202012_17(4).0007

郭靜燕(Ching-Yen Kuo)；吳炯義(Chiung-Yi Wu)；詹金淦(Chin-Kan Chan)；詹前隆(Chien-Lung Chan)

小兒氣喘 ； 表現型態 ； 群集分析法 ； 資料探勘 ； childhood asthma ； asthma phenotypes ； cluster analysis ； data mining

北市醫學雜誌

17卷4期（2020 / 12 / 30）

425 - 434

繁體中文

目的：本研究以資料探勘之群集分析法來區別小兒氣喘不同表現型態，以提供臨床醫療治療及預防策略參考。方法：以2012至2016某區域醫院參加中央健康保險署「氣喘醫療給付改善方案」之6至18歲小兒氣喘收案病人為本研究對象，以資料探勘之二階段群集分析法進行小兒氣喘分群，以區別不同氣喘表現型態。結果：2012～2016年6至18歲氣喘收案病患共334人，男性140人（41.9%），女性194人（58.1%）；平均年齡8.4歲；以群集分析法分析小兒氣喘表現型態可分為3群，第一群有129個個案，嚴重度為輕中度，嗜伊紅白血球陽離子蛋白（ECP）≧15；第二群有90個個案，屬嚴重個案；第三群有115個個案，嚴重度為極輕、輕中度，ECP＜15。結論：本研究將小兒氣喘分群3種不同表現型態，嚴重度是分群的依據，本研究結果可提供臨床醫療治療及預防策略參考，並能加強氣喘病人之追蹤管理及衛教服務，提供完整且連續性的照護模式。

Objective: To explore asthma phenotypes in children using cluster analysis and thus, provide a reference for the treatment and prevention. Methods: Asthma children aged 6-18 years who joined the Asthma Improvement Program of National Health Insurance between 2012 and 2016 in a regional hospital were the research samples. A two-stage cluster analysis was used to differentiate the asthma phenotypes. Results: A total of 334 asthma patients with an average age of 8.4 years were admitted, among which 140 were male (41.9%) and 194 (58.1%) were female. The pediatric asthma phenotypes were divided into three groups. The first group consisted 129 cases with a mild-to-moderate disease severity and an ECP≧15; the second group consisted 90 severe cases; the third group consisted 115 cases with extremely low-to-medium disease severity and an ECP ＜15. Conclusion: This study categorized pediatric asthma phenotypes into three groups based on the disease severity.