Health risk assessment of heavy metals in Iranian cereal products using ICP-AES and a hybrid machine learning framework

This research examines the health risks posed by heavy metal contamination in 13 popular cereal-based products from Arak, Iran. The team measured arsenic (As), cadmium (Cd), mercury (Hg), lead (Pb), and copper (Cu) levels using inductively coupled plasma atomic emission spectrometry (ICP-AES). Quantification revealed ranges from 0–0.834 mg/kg for As, 0.0002–0.014 mg/kg for Cd, 0–0.502 mg/kg for Hg, 0–2.0 mg/kg for Pb, and 0–1.203 mg/kg for Cu.

The highest hazard quotients (HQ) were identified for Tak (HQ = 1.657) and Zhik (HQ = 0.928) brands, primarily due to mercury (Hg) and arsenic (As) respectively, suggesting potential non-carcinogenic health risks. In comparison, Zar, Mak, and Jahan brands had HQ values under 0.01. Researchers created a hybrid machine learning framework, merging Random Forest and Gradient Boosting, to predict HQ.

The model demonstrated high accuracy (R2 = 0.97, RMSE = 0.042) and highlighted mercury (34%), arsenic (28%), copper (16%), cadmium (12%), and lead (10%) as the primary risk factors. While the study's small sample size (n=13) necessitates further investigation, the hybrid ensemble approach proves to be a reliable and efficient method for rapid food safety assessment. Funding was provided by the Research Council of Arak University.