The Possibilities of Neural Network Analysis to Evaluate the Prognosis of Chronic Heart Failure in Elderly Patients
New diagnostic approaches to establish the severity of chronic heart failure as a widespread syndrome on a background of cardiovascular diseases should integrate the results of various studies of the pathogenesis and create a basis for risk assessment of its progression, estimation of the individual prognosis. To develop an algorithm of integrated assessment and prediction of functional disorders of the cardio-vascular system, a neural network analysis of echo- and Doppler-cardiography indicators, markers of subclinical inflammation, lipid disorders, oxidative stress, apoptosis, interstitial fibrosis in the myocardium, reflecting the severity of the major pathogenetic processes in the progression of heart failure in elderly hypertensive patients was carried out. The use of neural network analysis by means of neuro-imitator NeuroPro 0,25 on the basis of a consultation of neural networks has provided a highly accurate assessment of the risk of cardiovascular disorders. As results of the experiment were 15 neural networks of minimum structure with their simplification by reducing the number of input signals, allowed to accurately predict the functional class of heart failure. The highest factor importance of reducing serum levels of tissue inhibitor of matrix metalloproteinase-1 less than 500 pg/ml, the increasing end-diastolic dimensions of the left ventricle over 5 cm, the activity level of high-sensitivity C-reactive protein more than 5 mg/l in determining the prognosis of progression of chronic heart failure were identified