تشخیص بهتر سلامت رانندگان با بهره گیری از شبکه عصبی مصنوعی

Introduction: Uncontrolled health status of drivers, can lead to the death of healthy individuals who are living in their best periods of life in terms of performance and wellness and also it can impose huge financial costs on a country. The purpose of this study was to design an intelligent system using Multilayer perceptron (MLP) and radial basis function (RBF) neural networks in order to diagnose drivers’ health status Methods: In this study, we applied the MLP and RBF networks with some changes in the number of middle layers, neurons, as well as learning algorithms such as Momentum (MOM), Conjugate Gradient (CG), and Levenberg Marquardt (LM) in order to diagnose the health status of the drivers.) Then, the best model was introduced according to the area under receiver operating characteristics (ROC) curve, sensitivity, and precision criteria. Results: In this study, 20 variables were selected as inputs and two variables that include healthy and unhealthy status were determined as output parameters. MLP and RBF neural networks with LM algorithm have the best performance with 66.7% and 29% precision; 97.2% and 100% sensitivity; 91.1% and 86 % accuracy respectively. The area under ROC curve for the nervous system MLP and RBF estimated 91.02 for MLP and 88.1 for RBF. Conclusion: According to this study, the MLP neural network model with the LM learning algorithm compared to the RBF neural network can have an important role in helping physicians in order to diagnose drivers’ health status. Furthermore, such a model can be used in centers of occupational medicine to enhance the accuracy and the speed of diagnosis and reduce costs.