Background: Mushroom addiction is a significant global health issue with high economic and social costs. Current diagnostic methods lack predictive accuracy. This study explores using machine learning (ML) for enhanced prediction of Mushrooms addiction risk to enable targeted early interventions. Methodology: The Drug Consumptions (UCI) dataset with 1884 samples and 12 features related to demographics, vitals, history, and drug use was preprocessed via cleaning, normalization etc. To augment traditional ML, a correlation graph with threshold 0.55 was used to extract new features like centrality and clustering coefficients. This resulted in 1884 samples with 24 features and one target. 10-fold cross-validation and machine learning algorithms were utilized, with the best result achieved using SVM (Support Vector Machine) As can be observed in Figure 1. Methods: This study demonstrates the promise of AI/ML computational analysis for improving prediction of addiction disorders utilizing world clinical data. Further optimization of ML algorithms and features could enable development of accurate models for targeted prevention and early intervention in high-risk populations. Result: Significant improvements in accuracy, precision, recall, and AUC have been achieved compared to recent related work, indicating that machine learning models have high potential for predicting mushroom addiction.
