• 作者： 楊炳炘; 陳光和; 林瑞雄
• 作者服務機構： 財團法人中華血液基金會研究處; 國立臺灣大學公共衛生學院公共衛生研究所
• 中文摘要： 以血清總蛋白質、白蛋白質、總膽固醇、尿素氮、尿酸、肌酸酐、總膽紅質、鹼性磷酸?、丙胺酸轉胺?、天門冬
  胺酸轉胺?等10項血清生化學檢驗數據為依據，經過主成分統計分析整合為三個新的獨立指標，分別定義為肝細胞損
  壞、肝臟代謝功能、肝以外檢驗等三項指標。再利用這三次元的新指標，藉類群分析可以將890位青少年類聚為幾個子
  群體；不同子群體間性別、或B型肝炎標誌分布情形，都具有顯著的差異(P<0.001)。890位青少年每人都接受兩次的血
  清生化學篩檢，其間間隔3.5年；第一次篩檢時的年齡為14歲至17歲間(平均值15.5歲，標準差0.7歲)。兩次篩檢數據
  個別分析的結果相當一致，證實本研究得到一個可行的方案：利用多變項統計分析將有助於提高群體的物質性，進而改
  善群體正常參考值，以利健康監視的成效。
  i
• 英文摘要： All 890 study subjects attended two serum biochemistry screenings at an interval 3.5 years, the firstwhile they were aged between 14 to 17 years (mean 15.5 and standard deviation 0.7). Three mutuallyindependent laboratory indices-hepatocellular damage, liver metabolism, and outer-liver measurement-were identified as physiological scales by principal component analysis on ten common serum biochemicaltests. Existence of a subgroup for an entity of healthy adolescents were supported by sum-of-squaresclustering analysis on these three-dimensional indices. The consistency between two time-point repeatedscreenings for each of the same subjects confirmed the practicability of clustering schemes. The resultingcluster pattern for each screening bore a relation to gender or hepatitis B surface antigen (HBsAg) classifica-tions. These features are expected to provide insight into normal references with enhanced group homo-geneity, and lead to an improved multivariate protocol that will aid in health monitoring of subgroupswith high risk.
• 中文關鍵字： cluster analysis; principal component analysis; group homgeneity; serum biochemistry; adolescent.
• 英文關鍵字： --