Complex application of mathematical methods in diagnostics of diffuse liver diseases

Abstract

Aim of investigation. The standard of diffuse liver disease (DLD) diagnostics, allowing to determine etiology, stage of liver fibrosis (LF), histological activity index (HAI) is the liver biopsy (LB). At inaccessibility of LB for diagnostics, LF stage can be assessed by noninvasive methods based on mathematical modelling with plotting of «images» of various etiological forms, LF stages and HAI of disease. An aim of investigation was stepwise application of mathematical analysis methods for improvement of diagnostics of DLD stage and its activity.Methods. Overall 450 patients were investigated: 319 HCV RNA-positive patients and 131 HBV DNApositive patients. To construct the «images» of stages of chronic HCV-and HBV-infections discriminant analysis, neural networks, computer programming were applied. The result of non-invasive diagnostics of patients with HСV-infection in 71% of cases coincided with LB data by LF stage, in 62% – by HAI. At HВV-infection LF stage conformed LB data in 78%, HAI – in 71% of the cases.Conclusions. Combined application of mathematical methods in DLD diagnostics with assessment of LF stages and HAI allows non-invasive estimation of these parameters with sufficient diagnostic accuracy. At revealing of severe liver disease (LF stage ≥2 by METAVIR, HAI by Knodell ≥4 points) LB is recommended.

Keywords

• chronic hepatitis
• liver cirrhosis
• discriminant analysis
• neural networks
• computer programming