多组学与人工智能整合在血液肿瘤诊疗中的研究进展
孙瑞霞 1,2,赵邢力 1,2*
作者信息
- 1 天津市人民医院,南开大学第一附属医院,肿瘤诊疗中心血液科,天津 300121
- 2 天津市人民医院,南开大学第一附属医院,中西医结合肿瘤研究所,天津 300121
*通信作者
摘要
血液肿瘤的生物学行为及临床预后具有较为复杂的异质性,传统基于形态学和有限分子标志物的分类体系难以满足精准诊疗的需求。 随着高通量测序、单细胞、空间组学等技术的成熟,涵盖基因组、表观基因组、转录组、蛋白质组与代谢组的多维数据得以系统解析,为揭示克隆演化轨迹、耐药形成路径及免疫生态重塑提供了全新视角。 与此同时,机器学习与深度学习等人工智能方法在处理高维、多模态生物医学数据、自动特征提取和构建预测模型方面展现出显著优势,逐渐从辅助分析工具演变为推动诊断、风险分层和疗效预测的核心方法。 本文综述了多组学数据特征及人工智能方法原理,系统总结了两者在血液肿瘤分子分型、复发风险评估以及治疗反应预测中的整合应用进展,并分析了当前存在的局限性和面临的关键挑战,旨在为完善血液肿瘤风险分层体系、推动血液疾病精准诊疗提供新的方法学参考与研究方向。
Research advances in integrating multi-omics and artificial intelligence for the diagnosis and treatment of hematological malignancies
Ruixia SUN 1,2, Xingli ZHAO 1,2*
Author Information
- 1 Department of Hematology, Oncology Center, Tianjin Union Medical Center, The First Affiliated Hospital of NanKai University, NanKai University, Tianjin 300121, China
- 2 Tianjin Cancer Institute of Integrative Traditional Chinese and Western Medicine, Tianjin Union Medical Center, The First Affiliated Hospital of Nankai University, NanKai University, Tianjin 300121, China
*Corresponding author
Abstract
Hematological malignancies are characterized by pronounced molecular heterogeneity and diverse clinical outcomes, rendering traditional classification systems based on morphology and a limited set of molecular markers insufficient for the requirements of precision medicine. With the maturation of high-throughput sequencing, single-cell and spatial omics technologies, multidimensional data encompassing the genome, epigenome, transcriptome, proteome, and metabolome can now be systematically interrogated, providing new insights into elucidating clonal evolutionary trajectories, mechanisms of drug resistance, and the remodeling of the immune microenvironment. In parallel, artificial intelligence (AI) methods, particularly machine learning and deep learning, have demonstrated substantial advantages in handling high-dimensional, multi-modal biomedical data, performing automated feature extraction, and constructing predictive models. They are evolving from auxiliary analytic tools into core approaches that drive diagnosis, risk stratification, and prediction of treatment response. This review outlines the characteristics of multi-omics data and the fundamental principles of AI methodologies, systematically summarizes the advances in their integrated application to molecular classification, relapse risk assessment and treatment response prediction in hematological malignancies,and analyses the existing limitations and key challenges, thereby providing new methodological perspectives and research directions for refining risk stratification systems and promoting precision diagnosis and treatment of hematological diseases.
