Excessive nitrate consumption has been linked to potential health risks in humans. Thus, understanding nitrate levels in staple foods such as cow milk can provide insights into their health implications. This study meticulously examined nitrate concentrations in 70 cow milk samples from traditional and industrialized cattle farming systems in Fars province, Iran. A combination of deterministic modeling, a probabilistic approach, and six artificial intelligence algorithms was employed to determine health risk assessments. The data disclosed average nitrate concentrations of 32.63 mg/L in traditional farming and 34.95 mg/L in industrialized systems, presenting no statistically significant difference (p > 0.05). The Hazard Quotient (HQ) was deployed to gauge potential health threats, underscoring heightened vulnerability in children, who exhibited HQ values ranging from 0.05 to 0.58 (mean = 0.19) in contrast to adults, whose values spanned 0.01 to 0.16 (mean = 0.05). Monte Carlo simulations enriched the risk assessment, demarcating the 5th and 95th percentile nitrate concentrations for children at 0.07 and 0.39, respectively. In children, pivotal interactions that influenced HQ encompassed those between nitrate concentration and consumption rate, as well as nitrate concentration and body weight. The interplay between nitrate concentration and consumption rate was most consequential for the adult cohort. Among the algorithms assessed for HQ prediction, Gaussian Naive Bayes (GNB) was optimal for children and eXtreme Gradient Boosting (XGB) for adults, with nitrate concentration being a key determinant. The results underscore the imperative for rigorous oversight of milk nitrate concentrations, highlighting the enhanced susceptibility of children and emphasizing the need for preventive strategies and enlightened consumption.
Keywords: Artificial intelligence; Cow milk; Hazard Quotient; Monte Carlo simulations; Nitrate concentrations.
Copyright © 2023 Elsevier Ltd. All rights reserved.