Application of decision tree models, logistic regression and pooled logistic regression to predict recovery in patients with sudden sensorineural hearing loss

Document Type : Original Article

Authors

1 Department of Epidemiology and Biostatistics, School of Health, Mashhad University of Medical Sciences, Mashhad, Iran

2 Otorhinolaryngology Department of Qhaem Hospital, Mashhad University of Medical Sciences

Abstract

The present study aimed to predict the recovery of patients with sudden sensorineural deafness using decision tree, logistic regression, and additive logistic regression models. In this study, the data of all patients with sudden sensorineural deafness referred to the ENT department of Ghaem Hospital in Mashhad from 2010 to 2019 were recruited. After preprocessing the data, logistic regression modeling, decision tree, and logistic regression model were performed in the R4.3.2 programming environment. Although the chi-square test in the primary analysis showed that underlying diseases such as high blood pressure and diabetes significantly reduce the chance of recovery, based on the results of the logistic regression model, the variables of age and SRT at the start of treatment were found to be significant. The final variables used in the decision tree also included age, SRT at the start of treatment, and gender of the patients. The additive logistic regression model, like the ordinary logistic regression model, showed that with increasing age and increasing delay in starting treatment, the chance of recovery of patient’s decreases. In addition to these variables, having a recent infection and delay in starting treatment were also identified as significant in the additive logistic regression model. Therefore, in the present study, the additive logistic regression model identified more significant variables than the ordinary logistic regression model and the decision tree, which can be considered in predicting response to treatment.

Keywords

Main Subjects

Beyhagh
Volume 30, Issue 3
November 2025

Files

Share

How to cite

Statistics