The approaches to detecting blood system pathologies based on non-invasive methods of spectrophotometry in the visible and near infrared ranges and intelligent data analysis are proposed. Blood system pathologies are diseases that affect blood components, such as red and white blood cells, platelets and plasma. A group of patients with various pathologies, such as erythrocytosis, anemia and leukemia, was studied. The effectiveness of a combined approach combining multichannel optical spectroscopic analysis with modern machine learning algorithms for processing broadband spectral characteristics of biological tissues in order to differentiate pathological and physiological conditions is shown.. Keywords: spectrophotometry, blood system pathology, data mining, linear discriminant analysis.