BPNN Applied In The Assessment Of Factors Associated With The Seeking-Care Behavior Of TB Suspects

Objective: The main purpose of this study is to explore the application of Back Propagation Neural Networks (BPNN) in the risk-factor analysis associated with Tuberculosis (TB) prevention and control. Based on the data of epidemiological survey on seeking-care behavior and its related factors of TB suspects in Chongqing, use BPNN to model the data to find valuable factors, which affect TB suspects to seeking-care in earlier time. Compare the result of BPNN to results of multivariate Logistic model and multivariate Cox model to evaluate the potential application of BPNN in the TB prevention and control field and to afford a new solution to the related factors analysis in epidemiology.Method: Based on the survey data of study on the seeking-care behaviors between local and rural-to-urban residents in Chongqing, Health Ministry project of TB implementing study, loaned by World Bank. Use MATLAB Neural Networks Tools as platform to establish BPNN, and calculate mean-impact-value (MIV) for each factor, list the factor order by its Absolute Value of MIV, then compare and synthesize the results with...