Enhanced Prediction Of Anabolic-Androgenic Steroid-Induced Side Effects On Human Body And Adverse Impact On Environment

Abstract

This research investigates the application of machine learning to predict severe side effects, including cardiovascular diseases (CVDs), pulmonary diseases, hormonal imbalances, and liver damage associated with anabolic-androgenic steroid (AAS) consumption in human bodies. It further analyses how the steroids consumed by humans are not metabolized fully and are excreted through urine and feces which eventually hampers the environment through water and land pollution.  Utilizing datasets focused on common AAS drugs (Anadrol, Oxandrolone, Clenbuterol, Deca Durabolin, and Dianabol), the study developed and optimized predictive models to address critical AAS-related health risks. A novel hybrid algorithm, combining Grid Search Cross-Validation with Support Vector Machines (SVM) and Multilayer Perceptron (MLP), achieved a maximum accuracy of 91% for predicting hormonal imbalances and 88% for CVD and Pulmonary diseases, outperforming baseline models. For liver damage prediction, Gradient Boosting and a hybrid RNN+Gradient Boosting approach demonstrated superior performance. Analysis of hormonal imbalances further highlighted the efficacy of RNN+LSTM and MLP models in capturing non-linear and temporal dependencies, surpassing PLSR. These findings determine the potential of machine learning to identify individuals at high risk for AAS-induced health complications, facilitating timely interventions. Future research should expand the scope of AAS drugs, incorporate additional risk factors, and also determine the impact of steroids on environment in terms of water and land pollutants. This study is validating the results with respect to diverse datasets to enhance clinical applicability and improve the prevention and management of AAS-related adverse health outcomes and environmental impacts.