Machine Learning-Based Spatial Mapping, Source Apportionment and Health Risk Prediction of Heavy Metal Contamination in Zanjan Province, Iran

The ground beneath our feet is often treated as a silent, static stage for human life, but in many parts of the world, it is an active reservoir of invisible chemical history. When heavy metals like lead, zinc, or cadmium enter the soil through mining, industrial waste, or vehicle emissions, they do not break down or disappear. Instead, they persist for decades, waiting to be picked up by wind, washed by rain, or, most dangerously, ingested by people and animals. The risk these metals pose depends heavily on how much of them is present, where they are located, and who is exposed to them. Children are uniquely vulnerable; their smaller bodies, developing organs, and natural tendency to play in the dirt mean they absorb these toxins more readily than adults. Understanding the precise map of this contamination and predicting who is in danger requires more than just taking a few samples and guessing the rest. It demands a way to see the invisible patterns connecting the chemistry of the earth to the health of its inhabitants.

In the northwestern province of Zanjan, Iran, a team of researchers set out to map this hidden danger with unprecedented detail. The region is home to the Angouran mine, one of the largest zinc deposits in the Middle East, which has been operating since the 1960s alongside active smelters and industrial towns. For decades, these operations have left a chemical footprint on the surrounding land. To understand the full scope of the problem, the researchers collected soil from 357 different locations across the province, creating a dense grid of data points that covered the entire area. They measured the levels of six specific heavy metals: zinc, copper, cadmium, nickel, iron, and lead. Rather than relying on a single method to interpret this massive amount of data, they turned to a suite of advanced computer programs known as machine learning. These programs act like highly trained pattern-recognition engines, capable of finding complex relationships between the metal concentrations and the potential health risks they pose to humans.

The study revealed a landscape of stark contrasts. While some areas showed metal levels close to the natural background found in the region's rocks, others were saturated with contamination. Zinc was the most widespread offender, with nearly 96 percent of the sampled sites showing levels higher than the natural background, and some spots containing up to thirty-four times the expected amount. Cadmium, though present in smaller absolute quantities, was found to be the most dangerous in terms of health impact because it is highly toxic even in tiny amounts and stays in the human body for decades. The researchers used their computer models to trace the origins of this pollution. They identified three distinct sources: a mining and smelting signature dominated by zinc, cadmium, and lead, concentrated around the Angouran mine; an industrial signature of copper and iron linked to the Zanjan Industrial Town; and a natural, geological background of nickel and iron that came from the local rocks themselves.

What made this research particularly powerful was its ability to predict health risks with high precision using different types of computer models for different goals. The team found that no single computer program was perfect for every job. For predicting general pollution loads and non-cancer health risks, a model called Gradient Boosting Regression worked best, accurately mapping out where the soil was dangerous. However, for predicting the risk of developing cancer over a lifetime, a different model called Support Vector Regression proved superior. This distinction mattered because it allowed the researchers to build a more accurate picture of the threat than if they had forced one tool to do everything. Their analysis confirmed that cadmium and lead were the primary drivers of immediate, non-cancer health threats, while copper and nickel were the main contributors to long-term cancer risks.

Perhaps the most striking finding of the study was the dramatic difference in risk between children and adults. The researchers calculated that while only a tiny fraction of the sites posed a non-cancer health risk to adults, a massive 35 percent of the sites exceeded the safety threshold for children. In other words, a child playing in the soil at these locations faces a risk of harm that is more than twelve times higher than an adult standing in the same spot. This disparity is driven by the fact that children ingest more soil relative to their body weight and are more sensitive to toxins. The study pinpointed five specific "hotspot" sites, located right next to the mine tailings, where the contamination was so severe that immediate action, such as cleaning or capping the soil, is necessary to protect the local population.

The researchers also used a technique that breaks down exactly which metal is responsible for the risk at any given location. They found that in the most dangerous areas, cadmium was the single biggest factor pushing health risks over the safety limit. This means that if the goal is to make the soil safe for children, removing or neutralizing the cadmium would be the most effective first step, even if other metals remain. The study did not just identify the problem; it provided a clear, data-driven roadmap for solving it. By combining detailed spatial maps with advanced computer predictions, the team showed that the threat is not uniform but concentrated in specific zones driven by specific industrial activities.

This work serves as a blueprint for how modern science can tackle environmental health crises. It moves beyond simple measurements to create a dynamic, predictive understanding of how pollution moves through the environment and affects human health. The findings suggest that in mining-intensive regions, the protection of children requires a targeted approach that recognizes their unique vulnerability. The study concludes that immediate intervention is needed in the five identified hotspot zones, along with regular monitoring of the soil around industrial towns and the mine itself. By focusing on the specific metals that drive the highest risks, authorities can prioritize their efforts to clean the air and soil, ensuring that the ground beneath the province becomes a safe foundation for the next generation.