Machine learning-based health risk assessment of fluoride and nitrate in indian drinking water: A systematic review.

Abstract

Introduction

The availability of clean, safe drinking water is a significant problem for millions worldwide. Groundwater is the principal source of drinking water in rural and semi-arid locations, making fluoride pollution a significant public health concern. Although fluoride is a naturally occurring element beneficial in small amounts for preventing dental caries, its excessive intake may result in irreversible dental and skeletal fluorosis, which can have severe fiscal implications (Raza et al., 2017, Chen et al., 2024). The problem looks acute, especially in developing countries like India, where rapid industrialization and agricultural activities have further spoiled the situation (Zhu et al., 2023). Studies have shown fluoride contamination is prevalent in groundwater in many parts of India, including cases where the concentration frequently exceeds the permissible limits recommended by the WHO (Ali et al., 2023). Consequently, it poses a serious health risk to the local population, particularly in the rural setting where alternative water sources are meager. Health risks associated with fluoride contamination have been well-documented (Bazeli et al., 2022, Ali et al., 2021, Ghaderpoori et al., 2019). Prolonged intake to high concentrations of fluoride can result in dental fluorosis, teeth mottling, and skeletal fluorosis that imposes effects on the bone and joints, manifesting in pains and damage to the skeletal system. However, their severity differs according to fluoride concentration in drinking water, the duration of exposure, and age and nutritional status. Beyond dental and skeletal fluorosis, chronic fluoride exposure has been linked to neurological effects like impaired cognitive skills and reduced IQ in children, thyroid dysfunction, and gastrointestinal issues. India, especially states like Uttar Pradesh, Rajasthan, Gujarat, and Andhra Pradesh, faces significant challenges with fluoride contamination in groundwater, often exceeding the WHO’s recommended limit of 1.5 mg/L (Ali et al., 2023, Ali et al., 2022a, Ma et al., 2023). There have been numerous research studies during recent years to estimate health risks because of fluoride contamination in huge areas of India (Jin et al., 2024). For instance, a study conducted in one part of India, that of the Bichpuri block in Agra, Uttar Pradesh, reported more than 92 % of groundwater samples to have fluoride content more than the safe limit, thus exposing the local population to a substantial risk related to non-carcinogenic health effects (Ali et al., 2023, Sahu et al., 2017, Ayoob and Gupta, 2006, Mirzabeygi et al., 2017). Similar results have been reported by other studies carried out in various parts of the country, and it characterizes the problem. The safety of Agra region groundwater and the health of its inhabitants depend critically on the results of health risk assessments and monitoring of fluoride concentrations (Sahu et al., 2017, Ali et al., 2017, Ali et al., 2021, Ali et al., 2022a, Ali et al., 2022b, Ali et al., 2023). The outcomes of the study have revealed a critical concern that may greatly influence public health: A total of 27.27 % of the fluoride samples (27 out of 110) and 45.45 % of the nitrate samples (44 out of 110) exceeded the safe levels specified by the WHO. The average HQ values for adults were 3.02, 1.57, and 1.45; for children and teenagers, they were 1.88, 0.98, and 0.90, respectively, in the Mathura region of Uttar Pradesh (Ali et al., 2024). Thus inclusion of efficient defluoridation methods is mandatory for access to potable water for mitigation of the health hazards of fluoride contamination (Ayoob and Gupta, 2006, Mirzabeygi et al., 2017). Damage to human health and toxicity has resulted from water contamination caused by contaminants present at more than acceptable levels. With this context, activated alumina, bone char, and reverse osmosis have been ruled out for the extraction of overdosed fluoride from drinking water (Mohammadi et al., 2017, Zafarzadeh et al., 2022). Globally, human health suffers due to the high levels of nitrogen compounds (nitrates (NO₃^–) (Jalili et al., 2018), nitrites (NO₂^–)), fluorides (F^–), herbicides, disruption chemicals, drugs, and heavy metals in groundwater (Ali et al., 2022a, Ma et al., 2023, Ali et al., 2017). The Water Quality Index functions as a dependable device for assessing water quality. It is then possible to use Geographical Information System (GIS) interpolation techniques that are used for preparation of groundwater quality maps and identification of vulnerable locations. Integrating WQI with GIS gives complete ready results quickly making it possible for policymakers to come up with timely strategies for water quality management (Soleimani et al., 2018, Ali et al., 2021, Ali et al., 2022a, Mirzabeygi et al., 2017).

The present paper makes an honest attempt to review the current status of fluoride and nitrate contamination in groundwater across India and the health risks posed by chronic exposure to high levels of fluoride and nitrate. The proposed study estimates a Water Quality Index to evaluate the suitability of groundwater for drinking purposes, utilizing a Geographic Information System. Additionally, it assesses the non-carcinogenic human health risks linked to exposure to nitrate and fluoride through a risk assessment methodology within the affected population group. The research was conducted to determine groundwater contamination with fluoride, nitrate, and nitrite and assess the health risks of using groundwater as a drinking water source (Yousefi et al., 2018, Wang et al., 2024).

Despite their widespread use for groundwater quality estimation, there is a significant gap in research on the application of ANN models to forecast the HQ of fluoride contamination. The HQ necessitates an accurate risk assessment, a crucial step in understanding the non-carcinogenic health impacts of fluoride. While ANN models have shown promise in predicting fluoride levels, the area of HQ prediction remains under-researched (Mohammadi et al., 2016, Hussein et al., 2020, Yousefi et al., 2019). To address this issue, ANN offers an innovative approach to risk modeling. Artificial neural networks (ANNs) are appropriate because they can recognize and forecast intricate patterns in data and capture the nonlinear correlations between fluoride and nitrate exposure characteristics and the risk of health impacts (Islam et al., 2024, Zafarzadeh et al., 2021). Investigators may exploit this gap by developing and testing ANN-based and XGBoost models for HQ estimation; this will improve public health and management of fluoride and nitrate contamination. This review paper presents a novel method using machine learning methods—especially ANN and XGBoost—to evaluate the health concerns of fluoride and nitrate contamination in groundwater of India. To forecast groundwater quality and related health hazards for various population groups, it especially blends advanced machine learning models with geographic information system (GIS) data. The key objectives of the present research are outlined that follow: (a) To perform an extensive literature review of fluoride and nitrate pollution of groundwater resources in India between 2015 and 2024. (b) To determine the geographical pattern of the level of fluoride and nitrate concentrations with Geographic Information System (GIS) and map the most contaminated areas. (c) To assess the health risks of fluoride and nitrate exposure using a health risk assessment approach, with a focus on children, adolescents, and adults. (d) To evaluate the performance of machine learning algorithms, i.e., Artificial Neural Networks (ANN) and Extreme Gradient Boosting (XGBoost), to predict fluoride and nitrate concentrations and health risk estimates.