Abstract

Info

Title:: Prediction model of pediatric thyroid disease based on back-propagation neural network

Author(s):: TIAN Juan1; ZHU Shujing2; LU Qiang1; LI Kun1; ZHANG Xixue1; 1. School of Medical Information Engineering, Shandong First Medical University, Taian 271016, China 2. Department of Laboratory, Taian Center for Diseases Prevention and Control, Taian 271016, China

Keywords:: Keywords: pediatric thyroid disease back-propagation neural network disease prediction classification accuracy

Abstract:: Abstract: Objective To construct a prediction model of thyroid disease in pediatric patients. Methods The physical examination data and preliminary clinical diagnoses of 1 400 children aged 8-11 years were collected from 2013 to 2016 in a Center for Disease Prevention and Control. One thousand out of 1 400 children were randomly selected as training samples, and the remaining 400 were used as test samples. A three-layer back-propagation neural network model was constructed by MATLAB R2018b software. Results The classification accuracy of the model reached 91.43% when the combination of log&log was selected as the transfer function of the hidden layer and the output layer, with 8 nodes in the hidden layer. Conclusion Back-propagation neural network can be applied to the prediction of thyroid diseases in children, and provide a theoretical basis for the prevention and treatment of diseases.