[Discriminate function of diabetes mellitus (DM) for screening test by serum 3 lipids and glucose]

The purpose of this study was to establish a program to differentiate diabetic patients in computerized health check data using serum lipid test data. We selected 4 items (TC, FC, TG and Glu) as a marker of clusters for further study on the strength of data obtained from cluster analysis using chemical data obtained on 134 patients. Next we tried principal analysis of the 4 items and computed parameters of component 1 & component 2, and obtained a function (D) to discriminate DM. Function (D) is read from comp. 1 and comp. 2. Comp. 1 = (0.89 x TC + 0.95 x FC + 0.75 x TG + 0.24 x Glu) Comp. 2 = (-0.26 x TC -0.21 x FC + 0.27 x TG + 0.93 x Glu) D = (0.2 x comp. 1 + comp. 2-160) DM patients show D greater than = 0 and non-DM patients D less than 0. We applied data on 587 patients (including of 43 DM) incorporating function (D) to check the property. The results proved it useful as a screening test for diabetic patients. The parameters of function (D) are influenced by inter-laboratory differences. However, our parameters were still effective. Data obtained from other laboratories incorporating our parameter gave a diagnostic quality of more than 70%.