Application of a machine learning-based food risk framework to assess the public dietary risk of cadmium in China

The daunting challenges of heavy metal pollution and soil acidification seriously threaten human health and the development of sustainable agriculture worldwide. However, effective approaches for identifying and controlling public health risks in a large geographic space are urgently needed. Here, we propose a proof-of-concept machine learning-based food risk (MFR) framework and accurately recognize that soil pH is the key driver affecting cadmium accumulation in wheat and rice at the national scale. From the 1980s to the 2000s, under soil acidification, the dietary risk of cadmium through rice and wheat increased by approximately 10 % in Central South China and East China. Geospatial trade increased the uncertainty of risk. In contrast, the mitigation of soil acidification alleviated cadmium accumulation in crops and the associated public dietary risks from the 2000s to the 2010s. The dietary risk of cadmium is 1.61 (rice) and 1.59 (wheat) times greater for children than for adults. Because the dietary risks of cadmium in rice are 3.38 (adults) and 3.41 (children) times greater than those for wheat, grain-based dietary structure adjustment is a possible pathway for reducing public health risks. Taken together, soil-based solutions for controlling soil acidification and optimizing dietary structures are large geospatial strategies for alleviating the public dietary risk of heavy metals (e.g., cadmium).