Risk Prediction of Heroin Drug Consumption using Machine Learning Algorithms

Heroin, an illegal and extremely addictive opioid derived from morphine, is commonly administered via injection, sniffing, snorting, or smoking. Users initially experience an intense "rush" of pleasure, followed by effects such as skin flushing, dry mouth, limb heaviness, severe itching, nausea, vomiting, and prolonged drowsiness. Severe health problems, including as mental health disorders, abscesses, and illnesses of the liver, kidneys, and lungs, are linked to chronic usage. Additionally, injecting heroin greatly heightens the risk of developing infectious illnesses like as HIV and hepatitis, and can lead to blocked veins. In this work, we applied multiple machine learning algorithms to accurately detect heroin consumptions. Among these models, CatBoost classifier demonstrated superior performance, achieving an accuracy of 88.66%. This highlights CatBoost’s efficacy in the accurate detection of heroin consumption, highlighting its potential to improve detection systems in both forensic and clinical contexts.