关键词: 预测毒理学, 动物模型, 类器官芯片, 计算毒理学, 人工智能, 多组学

Abstract:

Traditional toxicological evaluation has long relied on mammalian models such as rodents， which suffers from multiple limitations including high costs， long experimental cycles， uncertain interspecies extrapolation and ethical concerns. Against this backdrop， the development of efficient， reliable and ethically acceptable alternative testing models has become an urgent requirement in this field. Owing to their unique biological features， high reproductive efficiency and low ethical concerns， non-mammalian models have shown great application potential in predictive toxicology research. This review elaborated on typical non-mammalian models， including Artemia， Caenorhabditis elegans， zebrafish， chick embryo chorioallantoic membrane， Drosophila and Hydra. Furthermore， it systematically summarized the current applications of promising emerging technologies such as organ-on-a-chip， computational toxicology， artificial intelligence and multi-omics in toxicity screening and assessment. It focused on analyzing the advantages and challenges of various models with respect to the correlation of toxicological responses， as well as existing issues such as interspecies metabolic differences and insufficient validation of model applicability. This review aims to provide theoretical references for researchers in selecting alternative models and optimizing research strategies， and to facilitate the innovative development of predictive toxicology models.

Key words: predictive toxicology, animal models, organ-on-a-chip, computational toxicology, artificial intelligence, multi-omics