毒性病理學(xué)是藥物臨床前安全性評(píng)價(jià)的重要環(huán)節(jié)，由于計(jì)算機(jī)技術(shù)和醫(yī)療技術(shù)的飛速發(fā)展，毒性病理學(xué)正在從模擬方法向數(shù)字方法過渡。人工智能（AI）作為一種新型的技術(shù)，已經(jīng)廣泛應(yīng)用于藥物研發(fā)的各個(gè)階段，包括臨床前安全性評(píng)價(jià)階段的毒性病理學(xué)檢查。但是，AI作為一種新興的技術(shù)，在毒性病理學(xué)中的應(yīng)用也面臨巨大的挑戰(zhàn)。目前，美國(guó)食品藥品監(jiān)督管理局（FDA）已經(jīng)批準(zhǔn)了多種基于AI或機(jī)器學(xué)習(xí)（ML）的醫(yī)療器械用于臨床輔助診斷。為了加強(qiáng)這類設(shè)備的審查和監(jiān)管，國(guó)內(nèi)外發(fā)布了一系列法規(guī)和指導(dǎo)原則?？偨Y(jié)AI和數(shù)字病理學(xué)的發(fā)展、AI在毒性病理學(xué)中的應(yīng)用、FDA批準(zhǔn)的基于AI和ML的醫(yī)療器械，以及國(guó)內(nèi)外相關(guān)的指導(dǎo)原則和法規(guī)，以期為AI在毒性病理學(xué)中的應(yīng)用和監(jiān)管提供新的思考。

Toxicologic pathology plays a crucial role in the preclinical safety evaluation of drugs. With the rapid advances in computer and medical technologies, toxicologic pathology is transiting from traditional simulation methods to digital approaches. Artificial intelligence (AI), as an emerging technology, has been widely applied across various stages of drug development, including toxicologic pathology assessments during the preclinical safety evaluation phase. However, the application of AI in toxicologic pathology faces significant challenges due to its novelty. Currently, U.S. FDA has approved several AI or machine learning (ML)-based medical devices for clinical diagnostic assistance. To strengthen the review and regulation of these medical devices, a series of regulations guidelines have been issued both domestically and internationally. This article provides an overview of the development of AI and digital pathology, application progress of AI in toxicologic pathology, AI and ML-based medical devices approved by U. S. FDA, and the relevant regulatory guidelines and regulations from both domestic and international perspectives, aiming to offer new insights into the application and regulation of AI in toxicologic pathology.

R965

藥品監(jiān)管科學(xué)全國(guó)重點(diǎn)實(shí)驗(yàn)室課題（2023SKLDRS0127）