The dissertation is devoted to theoretical generalization and solution of the actual scientific problem – improvement of diagnostics and prognostication of the development of diabetic retinopathy (DR) in patients with diabetes mellitus (DM) type 2 on the basis of the analysis of individual mechanisms of regulation of the platelet linkage hemostasis. For the first time, it was found that in DR patients the response of TC to agonists adrenaline, collagen, angiotensin 2, FAT and ADP changes ambiguously, which is due to the presence of different clusters of functional activity of receptors. This phenomenon is associated with individual reactivity of the organism and is manifested by various influences of humoral pathogenetic factors of diabetes on intracellular Tc signaling systems, which are associated with GPVI-receptors for collagen, α2-adreno- and AT1-receptors, purine (P2Y1 P2Y12) and FAT. The interaction of pathogenetic factors of type 2 diabetes mellitus (expression of collagen and adrenaline) in the modulation of TC function has been proved, which allows to specify the induction of thrombogenesis and retinal microcirculation disorders in patients with DR. For the first time in patients with nonproliferative DR, a prothrombogenic (hyperadrenoreactive) platelet phenotype was detected, which can create conditions for disease progression under the action of collagen, angiotensin-2 and FAT. In patients with proliferative DR, hemorrhage progression, inflammation, and neovascularization have been shown for the first time to involve the prothrombogenic (hyperangiotensin) Tc phenotype, which is characterized by a high response to ADP and FAT. For the first time, neural network models for predicting DR stages have been developed and implemented in practice. The linear model for predicting the stages of DR is based on the analysis of two factor features - aggregation of TC, induced by collagen and ADP. The prediction of the DR stage in a two-factor linear neural network model was based on the ATP-induced ADP and collagen in vitro, i.e., the analysis of the progregant status of Tc. Of the 99 cases of mistakes predicted for 18 patients, the prediction accuracy was 81.8% (95% CI 73.5% -88.8%). The individual thrombocytes reactivity with regard to the influence of pathogenetic factors of the central nervous system (collagen exposure, activation of sympatho-adrenal and renin-angiotensin systems, thrombocytes autocrine stimulation and inflammation) has been shown to justify the development of models for predicting the stages of DM and the risk of macular edema. To determine the leading determinants of the development of DR stages, a statistical analysis was carried out using methods for constructing multi-factor neural network and logistic regression models.
