An artificial neural network comparison with logistic regression in predicting post-traumatic mental disorders in mild brain injury patients

Subject Areas :

Elham  Shafiei ¹ (Prevention of Psychosocial Injuries, Research Centre, Ilam University of Medical Sciences, Ilam, Iran.)
arash  nademi ² (2. Department of Statistics, Ilam Branch Islamic Azad University, Ilam, IR Iran)
Esmaeil  l Fakharian ³ (Trauma Research Center, Kashan University of Medical Science, Kashan, IR Iran)
abdollah  omidi ⁴ (مدیرگروه روان شناسی بالینی و عضو هیئت علمی دانشگاه علوم پزشکی کاشان)

Received: 2018-02-25 Accepted : 2018-02-25 Published : 2017-12-22

Keywords: artificial neural network, logistic regression, post-traumatic mental disorders,

Abstract :

Although severe brain injury can make people susceptible to mental disorders, there is still debate about traumatic brain injury. The purpose of this study was to compare the power of artificial neural network in predicting post-traumatic mental disorder in mild brain injury patients and logistic regression. For this purpose, in a prospective cohort study, 100 trauma patients referred to the trauma center of Shahid Beheshti Hospital of Kashan during 6 months were compared with 100 healthy people. For modeling, the data were randomly divided into two educational groups (100) and experimental (100 people). The Rock's curve and classification accuracy were used to estimate the predictive power of mental disorder. The results showed that there is a significant difference between the two groups of mild traumatic patients and healthy subjects in terms of mental disorders, and artificial neural network models have better efficiency than logistic regression models. This study showed that in order to predict mental disorder, the diagnostic indices of this factor should be considered at the beginning of the traumatic brain injury patients and then, using the artificial neural network model, predict this factor. The necessity of using this technology in demographic screening is useful in treating patients with trauma and preventing possible problems for such patients.