Method based on neural network prediction of biological activity of media network control study of silicosis fibrosis.docVIP

 PAGE \* MERGEFORMAT 14 Method based on neural network prediction of biological activity of media network control study of silicosis fibrosis Author: Du ship Gao Hongsheng, Wang Rui-Chang Wu Yun-Doo [Abstract] Objective: To investigate the biological activity of silicotic fibrosis media relations. Methods: Using Delphi language developed BP artificial neural network model for computer programs, establish and analysis of silicosis, a mathematical model for prediction of collagen fibers. Results: The selection of the network hidden layer nodes is 9, the initial weight threshold value of approximately (-0.2,0.2), the maximum relative error of 4 percent, the smallest relative error is 0.2%. Conclusion: The application of neural network has better prediction results for clinical research studies to provide a good idea. [Keywords:] BP neural network; biological activity of media; silicosis; collagen fibers; Forecast Silicosis is a pneumoconiosis in the most severe type is the result of long-term inhalation of more than a certain concentration of dust containing free silica, of widespread nodular pulmonary fibrosis. Difficult to predict fibrosis silicosis, diagnosis lag. At present, the pathogenesis of silicosis is still not entirely clear, there is no effective early diagnosis (screening) methods, and no specific marker for early diagnosis and specific treatments and methods. Once the traditional post-anterior chest large diagnosed lung lesions have been irreversible. Therefore, the search for early diagnosis (screening) specific combinations of biological media on the prevention, treatment and even the eventual elimination of silicosis is of great significance. The incidence of silicosis and cytokines (Cytokine, CK) are closely linked network of control and high Hongsheng using systems biology approaches such as proof of the cytokine network regulation of relations between silicosis fibrosis [1,2], demonstrated the complexity of the non-cytokine Linear-induced i