胃癌是我国高发恶性肿瘤，化疗药物疗效已达瓶颈。抗人表皮生长因子受体2（HER2）曲妥珠单抗是目前晚期胃癌一线治疗唯一获批的靶向药物，极大的改善了HER2阳性胃癌患者生存，但仍存如下难题：胃癌存在时间异质性与空间异质性，单次单部位活检往往不能充分评估肿瘤的HER2表达状态；曲妥珠单抗联合化疗后可能存在原发耐药或继发耐药，治疗受益有待进一步提高。如何精准评估胃癌HER2在不同层面的表达特征，探索其与曲妥珠单抗治疗胃癌疗效及耐药的相关性，建立新的临床决策系统对于进一步提高疗效改善生存至关重要。基于此，本课题拟基于胃癌抗HER2靶向治疗临床资源数据库及生物样本库，融合病理学及影像组学，采用动态液体活检多组学测序技术，最终经人工智能方法融合多学科海量数据，构建多组学融合的胃癌抗HER2靶向治疗疗效和耐药的智能预测模型，并在前瞻性研究中进行验证和优化，最终辅助临床决策，改善预后。

Gastric cancer is the second most commonly diagnosed cancer and the second leading cause of cancer mortality in China. The efficacy of chemotherapy has limitation. Trastuzumab is the only approved targeted therapy in first-line treatment of gastric cancer which prolonged patients survival to a large extent. However, there are still difficultis exists: First, tumor heterogeneity of spatial and timing; Second, biopsy could not evaluate the HER2 status once a time; Third, the primary or secondary resistance of Trastuzumab combined with chemotherapy. There is an urgent need to accurately predict the efficacy and resistance of anti-HER2 targeted therapy in gastric cancer. Our project will base on the established clinical database and biobank of anti-HER2 targeted therapy in our hospital, combinig digital pathology, radiomics, dynamic liquid biopsy using multi-omics sequencing. The multi-disciplinary massive data will be integrated by artificial intelligence method to construct a multi-omics model predicting anti-HER2 targeted therapy efficacy and explore the mechanism of drug resistance. To achieve an effective pre-treatment prediction model, we will do the validation on prospective data. Ultimate, we would like to develop a model which could help doctors make clinical decision, and improve patients outcomes.