PO.BCS02.05 · 生物信息与计算

Genie-ADLA：一种用于基于甲基化的多癌种早期检测（MCED）的深度学习算法

Genie-ADLA: A deep learning algorithm for methylation-based multiple cancer early detection (MCED)

编号 5471 展板 7 时间 4/21 02:00–05:00 区域 Section 2 主讲 Guoqiang Zhao, MS

分会场 Deep Learning in Cancer

查看 PDF 下载 PDF 🔒 查看 / 下载完整 PDF 需登录并开通下载套餐 · 查看套餐 / 开通 AACR 官方页面

作者与单位 Authors & Affiliations

Kezhong Chen¹, Ziyu Li², Xiaojian Wu³, Jian Huang⁴, Guoyue Lv⁵, Weiping Wen⁶, Dahong Zhang⁷, Xiangyu Zhao⁸, Danbo Wang⁹, Zhihua Liu¹⁰, Lixin Sun¹¹, Shu Wang¹², Xiangnan Li¹³, Zhigang Li¹⁴, Jiandong Tai¹⁵, Jiayin Yang¹⁶, Zhentong Wei¹⁷, Ming Cai¹⁸, Qiang Zhang⁹, Songbing He¹⁹, Shuhua Yi²⁰, Shenhong Qu²¹, Wenhui Zhao²², Xianjun Yu²³, Ruixia Guo¹³, Jianhong Lian¹¹, Desong Yang²⁴, Huaiwu Lu²⁵, Xi Guo²⁶, Yan Zhang²⁷, Zhuowei Liu²⁸, Yingjiang Ye²⁹, Chang Lin³⁰, Jie Gao³¹, Xuanhui Liu³², Yushu Guo³³, Suying Ding¹³, Guoqiang Zhao³⁴, Yanzhan Yang³⁴, Jiangyu Li³⁴, Shiqing Chen³⁴, Hui Yu³⁴, Fang Liu³⁴, Yang Wang³⁴, Min Li³⁴, Baoliang Zhu³⁴, Yonghui Li³⁴, Xiaohui Wu³⁴, Fan Yang¹, Jun Wang¹

¹Thoracic Oncology Institute and Department of Thoracic Surgery, Peking University People’s Hospital, Beijing, China,²Peking University Cancer Hospital and Institute, Beijing, China,³Department of General Surgery (Colorectal Surgery), The Sixth Affiliated Hospital, Sun Yat-sen University, Guangzhou, China,⁴The Department of Breast Surgery, Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China,⁵Department of Hepatobiliary and Pancreatic Surgery, General Surgery Center, First Hospital of Jilin University, Changchun, China,⁶Department of Otolaryngology, The First Affiliated Hospital, Sun Yat-sen University, Guangzhou, China,⁷Urology & Nephrology Center, Department of Urology, Zhejiang Provincial People's Hospital (Affiliated People's Hospital), Hangzhou Medical College, Hangzhou, China,⁸Peking University People's Hospital, Peking Universtiy Institute of Hematology, Beijing, China,⁹Liaoning Provincial Cancer Hospital, Shenyang, China,¹⁰Cancer Hospital, Chinese Academy of Medical Sciences, Beijing, China,¹¹Department of General Surgery, Shanxi Cancer Hospital, Taiyuan, China,¹²Breast Disease Center, Peking University People's Hospital, Beijing, China,¹³The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China,¹⁴Department of Thoracic Surgery, Shanghai Chest Hospital, Shanghai Jiao Tong University School of medicine, Shanghai, China,¹⁵Department of Colorectal&anal Surgery, General Surgery Center, First Hospital of Jilin University, Changchun, China,¹⁶The Department of Liver Surgery of West China Hospital, Sichuan University, Chengdu, China,¹⁷Department of Obstetrics and Gynecology, The First Hospital of Jilin University, Changchun, China,¹⁸Department of Urology, The Second Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, China,¹⁹The First Affiliated Hospital of Soochow University Department of General Surgery, Suzhou, China,²⁰National Clinical Research Center for Blood Diseases, Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China,²¹Department of Otolaryngology-Head and Neck Surgery, Guangxi Zhuang Autonomous Region People's Hospital, Nanning, China,²²Harbin Medical University Cancer Hospital, Harbin, China,²³Department of Pancreatic Surgery, Fudan University Shanghai Cancer Center, Shanghai, China,²⁴Hunan Cancer Hospital & The Affiliated Cancer Hospital of Xiangya School of Medicine, Central South University, Changsha, China,²⁵Department of Gynecologic Oncology, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, Guangzhou, China,²⁶Department of Urology, Hunan Provincial People's Hospital, the First Affiliated Hospital of Hunan Normal University, Changsha, China,²⁷Department of Oncology, Shijiazhuang People’s Hospital, Shijiazhuang, China,²⁸Sun Yat-sen University Cancer Center, Guangzhou, China,²⁹Department of Gastrointestinal Surgery, Peking University People's Hospital, Beijing, China,³⁰Otorhinolaryngology Department of the First Affiliated Hospital of Fujian Medical University, Fuzhou, China,³¹Department of Hepatobiliary Surgery, Peking University Organ Transplantation Institute, Peking University People's Hospital, Beijing, China,³²The Sixth Affiliated Hospital, Sun Yat-sen University, Guangzhou, China,³³Health Management Center, Peking University People’s Hospital, Beijing, China,³⁴Shanghai Xiaohe Medical Laboratory Co. Ltd., Shanghai, China

摘要 Abstract

中文摘要

背景：基于甲基化的游离DNA（cfDNA）分析已成为MCED的一项关键技术。然而，现有方法依赖传统的机器学习算法，这本质上限制了检测性能。随着人工智能（AI）的快速发展，我们开发了Genie-ADLA，一种专为MCED设计的深度学习算法。通过将最先进的深度神经网络架构与甲基化数据固有的内在模式相结合，Genie-ADLA显著提升了MCED性能。方法：Genie-ADLA在一个由4,781名40-75岁参与者组成的数据集上进行了训练和评估，包括横跨16种癌症类型的2,702例经病理确诊的癌症病例和2,079例非癌症对照（NCT06217900）。训练集包含3,217份样本（1,756例癌症病例和1,461例非癌症对照），模型性能在一个由1,564份样本（618例非癌症对照和946例癌症病例）组成的独立测试集上进行评估。为应对甲基化数据固有的挑战——高维性、稀疏性和噪声——我们应用了特征降维和嵌入策略，减轻了计算负担，缓解了过拟合，并提高了学习效率。集成学习方法进一步增强了稳健性和泛化能力。结果：在16种癌症类型的所有分期中，Genie-ADLA在测试队列中于99.3%（612/618，95% CI：[97.90%，99.64%]）特异性下实现了63.43%（600/946，95% CI：[60.26%，66.50%]）的总体灵敏度。与在同一数据集上训练的XGBoost模型相比，Genie-ADLA在16种癌症类型中的11种中展现出改善的总体灵敏度，平均提升4.86%。对于I-III期癌症患者，在99.3%特异性下的灵敏度相比XGBoost有显著提升：结直肠癌达到76.98%（97/126，95% CI：[68.65%，84.01%]），从67.46%提升了9.52%；食管癌达到80.95%（51/63，95% CI：[69.09%，89.75%]），从74.60%提升了6.35%；乳腺癌达到37.14%（26/70，95% CI：[25.89%，49.52%]），提升了5.71%（从31.43%）。肺癌被细分为腺癌和非腺癌，I-III期灵敏度在腺癌中为40.90%（27/66，95% CI：[28.95%，53.71%]），提升了10.6%，在非腺癌中为84.44%（38/45，95% CI：[70.54%，93.51%]），提升了2.22%。结论：Genie-ADLA利用先进的深度神经网络架构和数据处理策略，大幅提升了基于甲基化的癌症早期检测的性能上限，为AI驱动的癌症筛查提供了新范式。

查看英文原文 English abstract

Background: Methylation-based analysis of cell-free DNA (cfDNA) has emerged as a key technology for MCED. However, existing approaches rely on traditional machine learning algorithms, which inherently limit detection performance. With the rapid advancement of artificial intelligence (AI), we have developed Genie-ADLA, a deep learning algorithm designed specifically for MCED. By integrating state-of-the-art deep neural network architectures with the intrinsic patterns inherent in methylation data, Genie-ADLA significantly enhanced MCED performance. Methods: Genie-ADLA was trained and evaluated on a dataset of 4,781 participants aged 40-75 years, including 2,702 pathologically confirmed cancer cases across 16 cancer types and 2,079 non-cancer controls (NCT06217900). The training set comprised 3,217 samples (1,756 cancer cases and 1,461 non-cancer controls), and the model's performance was evaluated on an independent test set of 1,564 samples (618 non-cancer controls and 946 cancer cases). To address challenges inherent to methylation data-high dimensionality, sparsity, and noise-we applied feature dimensionality reduction and embedding strategies, reducing computational burden, mitigating overfitting, and improving learning efficiency. An ensemble learning approach further strengthened robustness and generalization. Results: Across all stages of 16 cancer types, Genie-ADLA achieved an overall sensitivity of 63.43% (600/946, 95% CI: [60.26%, 66.50%]) at 99.3% (612/618, 95% CI: [97.90%, 99.64%]) specificity in the test cohort. Compared with the XGBoost model trained on the same dataset, Genie-ADLA demonstrated improved overall sensitivity in 11 of the 16 cancer types, with an average increase of 4.86%.For stage I-III cancer patients, the sensitivities at 99.3% specificity showed notable gains over XGBoost: colorectal cancer achieved 76.98% (97/126, 95% CI: [68.65%, 84.01%]), an improvement of 9.52% from 67.46%; esophageal cancer reached 80.95% (51/63, 95% CI: [69.09%, 89.75%]), up 6.35% from 74.60%; breast cancer reached 37.14% (26/70, 95% CI: [25.89%, 49.52%]), improving by 5.71% from 31.43%. Lung cancer was subdivided into adenocarcinoma and non-adenocarcinoma, with stage I-III sensitivities of 40.90% (27/66, 95% CI: [28.95%, 53.71%]) in adenocarcinoma, an increase of 10.6%, and 84.44% (38/45, 95% CI: [70.54%, 93.51%]) in non-adenocarcinoma, improving by 2.22%. Conclusions: Genie-ADLA, leveraging advanced deep neural network architectures and data processing strategies, substantially elevates the performance ceiling of methylation-based early cancer detection, offering a new paradigm for AI-driven cancer screening.

利益披露 Disclosure

K. Chen, None.. Z. Li, None.. X. Wu, None.. J. Huang, None.. G. Lv, None.. W. Wen, None.. D. Zhang, None.. X. Zhao, None.. D. Wang, None.. Z. Liu, None.. L. Sun, None.. S. Wang, None.. X. Li, None.. Z. Li, None.. J. Tai, None.. J. Yang, None.. Z. Wei, None.. M. Cai, None.. Q. Zhang, None.. S. He, None.. S. Yi, None.. S. Qu, None.. W. Zhao, None.. X. Yu, None.. R. Guo, None.. J. Lian, None.. D. Yang, None.. H. Lu, None.. X. Guo, None.. Y. Zhang, None.. Z. Liu, None.. Y. Ye, None.. C. Lin, None.. J. Gao, None.. X. Liu, None.. Y. Guo, None.. S. Ding, None. G. Zhao, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. Y. Yang, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. J. Li, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. S. Chen, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. H. Yu, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. F. Liu, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. Y. Wang, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. M. Li, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. B. Zhu, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. Y. Li, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. X. Wu, Shanghai Xiaohe Medical Laboratory Co. Ltd. Employment. F. Yang, None.. J. Wang, None.

← 返回 AACR 2026 检索