Multivariate analysis of skin impedance data in long-term type 1 diabetic patients

The purpose of the present work is to investigate the potential of multi-fr equency skin impedance as a versatile method for early detection of diabete s-related changes in the skin. The method implies application of a small si nusoidal voltage at various frequencies through the outermost layer of the skin. Skin impedance measurement is a quick and easy alternative to the exp ensive and slow study of skin biopsies by microscopy. However, the amount o f data is overwhelming and the measurement noise is rather high and therefo re multivariate data analysis is used. Different models of data transformat ion and data reduction for extraction of the parts of the data that correla te with patient status and other diagnostic parameters are discussed. Multi variate methods make it very easy to give a visual interpretation of the re sults. The results demonstrate how a regression model between skin impedanc e and other diagnostic data for diabetic patients and control groups can be developed into a novel diagnostic tool for the early discovery of possible complications for diabetic patients. (C) 1998 Elsevier Science B.V. All ri ghts reserved.