近年来，液体活检与人工智能技术相结合为肝细胞癌的早期诊断提供了一个新的研究视角。液体活检通过检测循环肿瘤DNA的基因突变、甲基化变化和片段组学特征，以及循环肿瘤细胞和细胞外囊泡等生物标记物，从而无创性地发现肿瘤早期分子事件；人工智能技术，尤其是机器学习与深度学习算法可以改善液体活检中低丰度信号和背景噪声的技术短板，提高诊断准确率，并且可以将液体活检数据与临床蛋白标记物、影像组学特征和病理组学特征等整合构建多模态分析模型，从而进一步提高模型性能，具有广阔的临床应用前景。但技术标准化不足、算法可解释性欠缺、外部验证有限及临床转化成本较高是阻碍液体活检临床推广的主要问题。

Recently, liquid biopsy and artificial intelligence (AI) technology offer promising early HCC detection opportunities. Liquid biopsy can provide non-invasive early molecular events from circulating tumour DNA mutations, methylation alterations and fragmentomic features, circulating tumour cells or extracellular vesicles. AI technologies such as machine learning and deep learning algorithm can overcome technical issues such as low-abundance signals and background noises for improved diagnostic performance. Building multimodal analytic models using liquid biopsies and clinical protein biomarkers, radiomics and pathomics features can improve model performance, which has substantial potential for clinical application. However, several barriers still impede clinical translation including poor technical standardization, low-interpretability of algorithms, poor external validation, and high implementation costs.