A stacked ensemble framework for predicting agrochemicals induced chronic disease risk

Exposure to agrochemicals has raised a major concern for public health because of its potential contribution to the growth and progression of numerous chronic diseases. The study examined agrochemicals associated chronic health issues including depression, cancer, chronic kidney disease (CKD), cardiovascular disease (CVD), Parkinson’s disease, hypertension, diabetes, and respiratory diseases using the National Health and Nutrition Examination Survey dataset (NHANES). This work emphasizes the development and assessment of an ensemble model to predict disease risk due to agrochemical exposure. To achieve accurate and interpretable disease risk prediction, we developed a novel stacked ensemble model (EN-ELM) by integrating Elastic Net (EN) and Extreme Learning Machine (ELM). The predictions from base learners consisting of traditional ML models are fed into meta learners, Elastic Net and Extreme Learning Machine, which are followed by logistic regression to give the final prediction. In contrast to existing studies, the proposed model achieved a higher predictability performance in forecasting various diseases associated with agrochemical exposure. The results indicate that some specific agrochemicals, such as organochlorine, organophosphate, and pyrethroids, had a major association with the disease risk. The work highlights the utilization of ensemble learning approaches with epidemiological data for health risk assessment. The findings provide directions to make strategic policies to minimize the agrochemical exposure and reduce the disease prevalence.