A toxicity risk priority control list of FLCMs and their transformation products developed through machine-learning...

Fluorinated liquid crystal monomers (FLCMs) are classified as emerging contaminants due to their persistence, bioaccumulation, and toxicity, with significant concern regarding their transformation products (TPs). A total of 2650 TPs from six main reaction pathways of FLCMs were deduced based on experimental results from experimental literature, and 37 risk indicators covering environmental, biological, and human health risks were computed for 2674 chemicals (24 FLCMs and 2650 TPs). A priority control list was established, categorizing 1542 substances as high risks, 804 as medium risk, and 328 as low risk based on the comprehensive toxic effects quantified through Min-Max Scaling, entropy weighting, and comprehensive index methods. Nine machine-learning and one deep-learning model were constructed, and TabNet was selected for the mechanistic analysis due to its superior performance. Hydroxyl radical oxidation and organismal metabolic pathways demonstrated high-risk potential. Toxicity correlations were identified with molecular lipophilicity, number of acidic groups, molecular structural complexity, electronegativity of fluorinated functional groups, and connection modes. This study represents the first systematic evaluation of the toxic effects of FLCMs and their TPs, providing computationally derived regulatory support for emerging contaminant management.