Logistic Regression Model and Bayesian Network of Factors Related to Drug Use Tendency

Abstract

The aim of this study is to compare the likelihood of drug use tendency among individuals in addiction rehabilitation camps and members of Narcotics Anonymous (NA) using the logistic regression model and to identify the Bayesian network model of factors associated with drug use tendency. This research follows a descriptive-correlational design using the logistic regression model and Bayesian network. The study population includes individuals who, in the second half of 2020, attended NA meetings or one of the medium-term residential treatment centers in Isfahan Province. A total of 823 questionnaires were distributed to these centers using convenience sampling, and after eliminating incomplete responses, 769 questionnaires were analyzed. Drug use tendency was assessed using the Weed and Butcher Addiction Potential Scale, which had an alpha coefficient of 0.90. The logistic regression model indicated that the likelihood of drug use tendency was lower among NA members compared to others. The accuracy of the Bayesian network model algorithm was 93.37%, demonstrating its strong predictive capability for drug use tendency.