ARTIFICIAL NEURAL NETWORKS APPLICATION FOR PREDICTION OF CANCER RISK IN PATIENTS WITH THYROID BENIGN PATHOLOGY

On base of a comprehensive survey of 248 patients with different thyroid pathology, artificial neural network has been created. Its task was to predict the possibility of cancer risk in patients with thyroid benign pathology. Neural network training based on the data of surveys carried out 120 patients, testing according to 128. Trained artificial neural network is highly flexible with regard to the different variants of the background benign pathology and capable with high sensitivity (89,5 %) and specificity (80,3 %) forecast the existence of hidden currents or probability of occurrence in future thyroid cancer in patients with benign pathology.