Optimization of the hypertension course prognosis at the primary level based on correlation indices and multiparameter neural network clasterization