Intelligent Data Analysis for the Diagnosis of Alcohol Dependence Syndrome

Abstract

Alcohol dependence syndrome is hard to diagnose because none of the existing laboratory markers alone has sufficient specificity and sensitivity. This study investigated whether combinations of markers would improve the identification of patients with alcohol dependence syndrome. Intelligent data analysis was carried out using the decision tree induction method with training and test data from 244 healthy volunteers and 238 patients with alcohol dependence syndrome. The results showed that a combination of two or three laboratory markers can identify alcohol dependence syndrome with almost 85% accuracy. It must be noted that induced decision trees offer a qualitatively different diagnostic evaluation of laboratory findings that varies from common practice, because they set up their own new borders and criteria that are different to generally accepted or set reference values. Tests for all of the selected laboratory markers are widely available, inexpensive to perform and usually form part of a routine laboratory examination.