Scientific journal
Научное обозрение. Медицинские науки
ISSN 2500-0780
ПИ №ФС77-57452

METHOD OF NEURAL NETWORK MODELLING AND ITS USAGE IN FORECASTING SOMATIC PATHOLOGY DEVELOPMENT AT PATIENTS OF SENIOR AGE GROUPS

Prokhorenko I.O. 1
1 Medical Institute «REAVIZ»
A special neural network model has been created in order to forecast development and severity of a widespread somatic pathology at patients of senior age groups. This model lets us define the probability of depression, metabolic syndrome, myocardial infarction (including non Q-wave MI), osteoporosis development with 66-89% accuracy on the ground of patients’ background data and psychological profile (according to Cattell questionnaire). Collected data let us draw a conclusion that usage of neural network algorithm with the purpose of for forecasting for the short-term is feasible and can achieve acceptable forecast mistake. Apart from accurate forecast computer program can define level of confidence in each diagnosis individually. This allows a doctor optimise treatment plan and take all the necessary prophylactic measures beforehand.