[原著］判別分析法を用いたインスリン依存性および非依存性糖尿病の境界領域における病型分類

機関リポジトリ 機関リポジトリ (HANDLE) オープンアクセス

書誌事項

タイトル別名
  • Classification of Diabetes in Patients between IDDM and NIDDM by Discriminant Function Analysis

この論文をさがす

説明

It is very difficult to determine the type of diabetes in patients after they have been treated with insulin. In order to classify diabetes in such patients, clinical manifestations and pancreatic B cell function together with the immunogenetic factors of diabetes were studied. Utilizing WHO criteria, 78 subjects were divided into three groups IDDM (33), NTDDM (19) and unclassified (26). The clinical manifestations at the age of their onset, family history of diabetes, glucagon stimulated peak plasma C-peptide (P-CPR) , insulin dose and HLA type (DR4,DR9,DR2) were studied. These indicators were used as discriminatory variables for classifying diabetes in the Forward Selection Discriminant Function Analysis of IDDM and NIDDM. Among these variables, P-CPR (X1), age at onset (X2), HLA DR4 (X3) and a family history of diabetes (X4) were selected as effective discriminatory variables, and the discriminant coefficients appeared in order of importance. The linear discriminant function was expresed as Z=2.782X1-1.287X2+4.200X3-2.148X4+3.876. The rate of miss-classification was 7.7% in 52 cases when these 4 variables were used. Even if one P-CPR determinant was taken, the incidence of miss-classification was defined as 12.5%. Therefore, if we were to take P-CPR to distingdsh IDDM from NIDDM, about 77% of the unclassified group could be re-classified into the IDDM group. These results suggest that P-CPR might be better than the other variables for discriminating between IDDM and NIDDM in patients whose diabetes had been treated with insulin.

論文

収録刊行物

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ