Multivariate and predictive modelling of arsenic and cadmium in rice: Influence of origin, grain type, and processing

Rice is a global dietary staple and understanding the accumulation of toxic and essential elements in rice grains is vital for public health, particularly in high-consumption regions. While prior studies have assessed elemental contamination, applying advanced machine learning (ML) to predict and analyze geographic patterns remains limited. This study analyzed arsenic (As), cadmium (Cd), and six essential elements (Zn, Fe, Cu, Mn, Se, and Mo) in 46 rice samples sold in Sydney, Australia, using ICP-MS. Multivariate statistics and various ML models were used to assess the influences of origin, rice type, and grain type on elemental profiles. As and Cd levels showed limited variation between groups, but notable differences within groups. PCA indicated moderate geographical clustering of As and Cd. Inorganic As predominated, especially in rice from Thailand, India, Pakistan, and others. Brown and purple rice had higher elemental concentrations. This combined profiling and ML approach supports improved food safety surveillance and origin-specific risk mitigation.