计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (9): 25-36.doi: 10.11896/jsjkx.250600104

• 智能医学工程 • 上一篇    下一篇

临床数据建模中的多域自适应问题研究进展

陈秀1, 张馨匀1, 程煜婷1, 陈伟1, 黄正行3, 刘振宇4, 张远鹏1,2   

  1. 1 南通大学医学院医学信息学系 江苏 南通 226001
    2 香港理工大学医疗科技与资讯学系 香港 999077
    3 浙江大学计算机科学与技术学院 杭州 310058
    4 中国科学院自动化研究所 北京 100190
  • 收稿日期:2025-06-14 修回日期:2025-08-01 出版日期:2025-09-15 发布日期:2025-09-11
  • 通讯作者: 张远鹏(y.p.zhang@ieee.org)
  • 作者简介:(2431310054@stmail.ntu.edu.cn)
  • 基金资助:
    江苏高校“青蓝工程”

Research Progress on Multi-domain Adaptation Problems in Clinical Data Modeling

CHEN Xiu1, ZHANG Xinyun1, CHENG Yuting1, CHEN Wei1, HUANG Zhengxing3, LIU Zhenyu4, ZHANG Yuanpeng1,2   

  1. 1 Department of Medical Informatics,School of Medicine,Nantong University,Nantong,Jiangsu 226001,China
    2 Department of Health Technology and Informatics,The Hong Kong Polytechnic University,Hong Kong 999077,China
    3 College of Computer Science and Technology,Zhejiang University,Hangzhou 310058,China
    4 Institute of Automation,Chinese Academy of Sciences,Beijing 100190,China
  • Received:2025-06-14 Revised:2025-08-01 Online:2025-09-15 Published:2025-09-11
  • About author:CHEN Xiu,born in 2001,master.Her main research interest is intelligent medical engineering.
    ZHANG Yuanpeng,born in 1984,professor,Ph.D supervisor.His main research interest is medical artificial intelligence.
  • Supported by:
    “Qinglan Project” of Jiangsu Higher Education Institutions.

PDF (PC)

8

摘要： 随着人工智能与医疗健康的深度融合,临床数据正经历从“辅助决策”到“驱动决策”的范式转变。临床数据包括患者症状、诊断影像、治疗记录等结构化与非结构化信息,为医疗决策提供重要支撑。然而,由于“领域偏移”现象的普遍存在,临床AI模型训练评估依赖的数据独立同分布假设(i.i.d.)失效,模型的跨域泛化能力被严重制约。域适应和域泛化技术可有效提升模型跨域表现。前者利用无标注目标域数据调整模型,使其适配新环境;后者基于源域数据学习域不变特征,实现无目标域数据下的泛化。针对两类技术在临床数据建模中的应用进展,按浅层、深层方法分类,展示其在不同数据类型中的应用场景,并总结了当前各类方法在泛化性能、数据依赖性与可解释性等方面的表现差异。

关键词: 临床数据, 领域偏移, 域适应, 域泛化, 数据依赖

Abstract: With the deep integration of artificial intelligence and healthcare,clinical data is undergoing a paradigm shift from “aiding decision-making” to “driving decision-making”.Clinical data encompasses both structured and unstructured information such as patient symptoms,diagnostic images,and treatment records,providing crucial support for medical decision-making.However,due to the prevalent “domain shift” phenomenon,the independent and identically distributed(i.i.d.) assumption,which clinical AI models rely on for training and evaluation,is invalidated,severely restricting the models’ cross-domain generalization ability.Domain adaptation and domain generalization techniques can effectively enhance the cross-domain performance of models.The former adjusts models use unlabeled target domain data to adapt them to new environments,the latter learns domain-invariant features based on source domain data to achieve generalization without target domain data.Regarding the application progress of these two types of techniques in clinical data modeling,this paper classifies them into shallow and deep methods,demonstrates their application scenarios across different data types,and summarizes the current performance differences of various methods in terms of generalization performance,data dependency,and interpretability.

Key words: Clinical data, Domain shift, Domain adaptation, Domain generalization, Data dependence

中图分类号: 

  • R319
[1]HAH H,GOLDIN D S.How Clinicians Perceive Artificial Intel-ligence-Assisted Technologies in Diagnostic Decision Making:Mixed Methods Approach[J].Journal of Medical Internet Research,2021,23(12):e33540.
[2]LIANG J Y,QIAN Y H,LI D Y,et al.Granular Computing Theory and Methods for Big Data Mining[J].Scientia Sinica:Informationis,2015,45(11):1355-1369.
[3]GAO S,HUANG X Y,WEI X Y,et al.Insulator Defect Detection Method Based on Domain Knowledge Transfer[J].Science Technology and Engineering,2023,23(14):6054-6062.
[4]TAO J,XU H.Discovering Domain-Invariant Subspace for Depression Recognition by Jointly Exploiting Appearance and Dynamics Feature Representations[J].IEEE Access,2019,7:186417-186436.
[5]GOETZ M,WEBER C,BINCZYK F,et al.DALSA:DomainAdaptation for Supervised Learning From Sparsely Annotated MR Images[J].IEEE Transactions on Medical Imaging,2016,35(1):184-196.
[6]CHEN K,LIU J,WAN R,et al.Unsupervised Domain Adaptation for Low-Dose CT Reconstruction via Bayesian Uncertainty Alignment[J].IEEE Transactions on Neural Networks and Learning Systems,2024,36(5):8525-8539.
[7]ZHAO X,SICILIA A,MINHAS D S,et al.Robust White Matter Hyperintensity Segmentation On Unseen Domain[C]//2021 IEEE 18th International Symposium on Biomedical Imaging(ISBI).IEEE,2021:1047-1051.
[8]XIA J,ZENG L,LI T,et al.PKDG:Prior Knowledge based Domain Generalization Model for fundus image segmentation[J/OL].Research Square,2024.https://doi.org/10.21203/rs.3.rs-4703557/v1.
[9]CARLONI G,TSAFTARIS S A,COLANTONIO S.CROCO-DILE:Causality Aids RObustness via COntrastive DIsentangled LEarning[C]//International Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging.Cham:Springer,2025:105-116.
[10]WACHINGER C,REUTER M.Domain Adaptation for Alzheimer’s Disease Diagnostics[J].NeuroImage,2016,139:470-479.
[11]CHEPLYINA V,PENA I P,PEDERSEN J H,et al.TransferLearning for Multicenter Classification of Chronic Obstructive Pulmonary Disease[J].IEEE Journal of Biomedical and Health Informatics,2018,22(5):1486-1496.
[12]XIA K,NI T,YIN H,et al.Cross-Domain Classification Model With Knowledge Utilization Maximization for Recognition of Epileptic EEG Signals[J].IEEE/ACM Transactions on Computational Biology and Bioinformatics,2021,18(1):53-61.
[13]LI W,ZHAO Y,CHEN X,et al.Detecting Alzheimer’s Disease on Small Dataset:A Knowledge Transfer Perspective[J].IEEE Journal of Biomedical and Health Informatics,2019,23(3):1234-1242.
[14]JIANG L,LIU S,MA Z,et al.Regularized RKHS-Based Subspace Learning for Motor Imagery Classification[J].Entropy,2022,24(2):195.
[15]PENG Y,WANG W,KONG W,et al.Joint feature adaptation and graph adaptive label propagation for cross-subject emotion recognition from EEG signals[J].IEEE Transactions on Affective Computing,2022,13(4):1941-1958.
[16]CUI J,JIN X,HU H,et al.Dynamic Distribution AlignmentWith Dual-Subspace Mapping for Cross-Subject Driver Mental State Detection[J].IEEE Transactions on Cognitive and Deve-lopmental Systems,2022,14(4):1705-1716.
[17]GANIN Y,USTINOVA E,AJAKAN H,et al.Domain-Adversarial Training of Neural Networks[M]//Domain Adaptation in Computer Vision Applications.Cham:Springer,2017:189-209.
[18]JIE L,LIANG P,ZHAO Z,et al.ADAN:An Adversarial Do-main Adaptation Neural Network for Early Gastric Cancer Prediction[C]//2022 44th Annual International Conference of the IEEE Engineering in Medicine & Biology Society(EMBC).2022:2169-2172.
[19]REN P,SHI X,YU Z,et al.Single-cell assignment using multiple-adversarial domain adaptation network with large-scale references[J].Cell Reports Methods,2023,3(9):100577.
[20]SOMERS P,HOLDENRIED-KRAFFT S,ZAHN J,et al.Cystoscopic depth estimation using gated adversarial domain adaptation[J].Biomedical Engineering Letters,2023,13(2):141-151.
[21]QIN X,BUI F,HAN Z.Semantically preserving adversarial unsupervised domain adaptation network for improving disease recognition from chest x-rays[J].Computerized Medical Imaging and Graphics,2023,107:102232.
[22]BAO G,ZHUANG N,TONG L,et al.Two-Level Domain Adaptation Neural Network for EEG-Based Emotion Recognition[J].Frontiers in Human Neuroscience,2021,14:605246.
[23]LIU W,NI Z,CHEN Q,et al.Attention-Guided Partial Domain Adaptation for Automated Pneumonia Diagnosis From Chest X-Ray Images[J].IEEE Journal of Biomedical and Health Informatics,2023,27(12):5848-5859.
[24]HONG X,ZHENG Q,LIU L,et al.Dynamic Joint Domain Adaptation Network for Motor Imagery Classification[J].IEEE Transactions on Neural Systems and Rehabilitation Enginee-ring,2021,29:556-565.
[25]OFER A,OPHIR A,YOAV N,et al.Supervised autoencoderdenoiser for non-stationarity in multi-session EEG-based BCI[J].Journal of Neural Engineering,2025,22(2):026043.
[26]CUI H,LI Y,WANG Y,et al.Toward Accurate Cardiac MRI Segmentation With Variational Autoencoder-Based Unsupervised Domain Adaptation[J].IEEE Transactions on Medical Imaging,2024,43(8):2924-2936.
[27]JIANG K,QUAN L,GONG T.Disentangled representation and cross-modality image translation based unsupervised domain adaptation method for abdominal organ segmentation[J].International Journal of Computer Assisted Radiology and Surgery,2022,17(6):1101-1113.
[28]WANG J,LAN C,LIU C,et al.Generalizing to Unseen Domains:A Survey on Domain Generalization[J].IEEE Transactions on Knowledge and Data Engineering,2023,35(8):8052-8072.
[29]GUO B,LU D,SZUMEL G,et al.The impact of scanner domain shift on deep learning performance in medical imaging:an experimental study[J].arXiv:2409.04368,2024.
[30]ATHALYE C,ARNAOUT R.Domain-guided dataaugmenta-tion for deep learning on medical imaging[J].PLOS One,2023,18(3):e0282532.
[31]KAISTI M,LAITALA J,WONG D,et al.Domain randomization using synthetic electrocardiograms for training neural networks[J].Artificial Intelligence in Medicine,2023,143:102583.
[32]AYODELE K P,IKEZOGWO W O,KOMOLAFE M A,et al.Supervised domain generalization for integration of disparate scalp EEG datasets for automatic epileptic seizure detection[J].Computers in Biology and Medicine,2020,120:103757.
[33]HE Y,SHEN X,XU R,et al.Covariate-shift generalization via random sample weighting[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2023:11828-11836.
[34]LIU B,ZHANG Y,WANG S,et al.DGSSA:Domain generalization with structural and stylistic augmentation for retinal vessel segmentation[J].arXiv:2501.03466,2025.
[35]SOLEIMANI M,TOOSI M H,MOHAMMADI S,et al.Using Test-Time Data Augmentation for Cross-Domain Atrial Fibrillation Detection from ECG Signals[J].arXiv:2503.13483,2025.
[36]YE Z.SSL-DG:rethinking and fusing semi-supervised learning and domain generalization in medical image segmentation[J].arXiv:2311.02583,2023.
[37]PATIL A,MEHTO A,NALBAND S.Enhancing skin lesion diagnosis with data augmentation techniques:a review of the state-of-the-art[J].Multimedia Tools and Applications,2025,84:25325-25364.
[38]APELLANIZ P A,PARRAS J,ZAZO S.Improving SyntheticData Generation Through Federated Learning in Scarce and Heterogeneous Data Scenarios[J].Big Data and Cognitive Computing,Multidisciplinary Digital Publishing Institute,2025,9(2):18.
[39]YU J,ZHU Y,FU P,et al.Robust Polyp Detection and Diagnosis through Compositional Prompt-Guided Diffusion Models[J].arXiv:2502.17951,2025.
[40]CHE H,WU Y,JIN H,et al.FedDAG:Federated Domain Adversarial Generation Towards Generalizable Medical Image Analysis[J].arXiv:2501.13967,2025.
[41]GUO L L,STEINBERG E,FLEMING S L,et al.EHR foundation models improve robustness in the presence of temporal distribution shift[J].Scientific Reports,2023,13(1):3767.
[42]WU Z,YAO H,LIEBOVITZ D,et al.An Iterative Self-Learning Framework for Medical Domain Generalization[J].Advances in Neural Information Processing Systems,2023,36:54833-54854.
[43]HONG J,LIU B,LONG G,et al.Rethinking domain generalization in medical image segmentation:One image as one domain[J].arXiv:2501.04741,2025.
[44]KIM H,SHIN Y,HWANG D.DiMix:Disentangle-and-MixBased Domain Generalizable Medical Image Segmentation[C]//Medical Image Computing and Computer Assisted Interven-tion－MICCAI 2023.Cham:Springer,2023:242-251.
[45]QIN W Q,BAO X,CHEN X,et al.A method for predicting the synergy of anticancer drug combinations based on graph attention network [J].Journal of Nantong University(Natural Science Edition),2025,24(1):10-17.
[46]GU R,ZHANG J,HUANG R,et al.Domain Composition and Attention for Unseen-Domain Generalizable Medical Image Segmentation[C]//Medical Image Computing and Computer Assisted Intervention－MICCAI 2021.Cham:Springer,2021:241-250.
[47]XU Y,ZHANG T.Boundless Across Domains:A New Paradigm of Adaptive Feature and Cross-Attention for Domain Generalization in Medical Image Segmentation[J].arXiv:2411.14883,2024.
[48]LI C,LIN X,MAO Y,et al.Domain generalization on medical imaging classification using episodic training with task augmentation[J].Computers in Biology and Medicine,2022,141:105144.
[49]KREUTER D,TULL S,GILBEY J,et al.Dis-AE:Multi-do-main & Multi-task Generalisation on Real-World Clinical Data[J].arXiv:2306.09177,2023.
[50]MATSUN A,MOHAMED D O,CHOKUWA S,et al.DGM-DR:Domain Generalization with Mutual Information Regularized Diabetic Retinopathy Classification[C]//Domain Adaptation and Representation Transfer.Cham:Springer,2024:115-125.
[51]RADWAN A,OSMAN I,SHEHATA M S.Universal medical imaging model for domain generalization with data privacy[J].arXiv:2407.14719,2024.
[52]DENG Z,XU Z,ISSHIKI T,et al.FedSemiDG:Domain Generalized Federated Semi-supervised Medical Image Segmentation[J].arXiv:2501.07378,2025.
[53]CHEN K,WANG X H,SHI X L,et al.A graph neural network method for predicting rutting in asphalt pavements based on causal inference [J].Journal of Nantong University(Natural Science Edition),2025,24(1):18-27,50.
[54]FEDER A,WALD Y,SHI C,et al.Data augmentations for improved(large) language model generalization[J].Advances in Neural Information Processing Systems,2023,36:70638-70653.
[55]CARVALHO J,ZHANG M,GEYER R,et al.Invariant anomaly detection under distribution shifts:a causal perspective[J].Advances in Neural Information Processing Systems,2023,36:56310-56337.
[56]SHETH P,MORAFFAH R,CANDAN K S,et al.Domain Generalization--A Causal Perspective[J].arXiv:2209.15177,2022.
[57]CUI X,CAO J,LAI X,et al.Cluster Embedding Joint-Probability-Discrepancy Transfer for Cross-Subject Seizure Detection[J].IEEE Transactions on Neural Systems and Rehabilitation Engineering,2023,31:593-605.
[58]CHAI X,WANG Q,ZHAO Y,et al.Multi-subject subspace alignment for non-stationary EEG-based emotion recognition[J].Technology and Health Care:Official Journal of the European Society for Engineering and Medicine,2018,26(S1):327-335.
[59]JIANG L,LIU S,MA Z,et al.Regularized RKHS-Based Subspace Learning for Motor Imagery Classification[J].Entropy,2022,24(2):195.
[60]EKONG F,YU Y,PATAMIA R A,et al.RetVes segmentation:A pseudo-labeling and feature knowledge distillation optimization technique for retinal vessel channel enhancement[J].Computers in Biology and Medicine,2024,182:109150.
[61]CUI H,YUWEN C,JIANG L,et al.Bidirectional cross-modality unsupervised domain adaptation using generative adversarial networks for cardiac image segmentation[J].Computers in Bio-logy and Medicine,2021,136:104726.
[62]KAUR S,GUMP A,XIAO Y,et al.CRoP:Context-wise Robust Static Human-Sensing Personalization[C]//Proceedings of the ACM on Interactive,Mobile,Wearable and Ubiquitous Techno-logies.2025:1-34.
[63]BERNECKER T,PETERS A,SCHLETT C L,et al.Fednorm:Modality-based normalization in federated learning for multi-modal liver segmentation[J].arXiv:2205.11096,2022.
[64]WANG S,YU L,LI K,et al.DoFE:Domain-Oriented FeatureEmbedding for Generalizable Fundus Image Segmentation on Unseen Datasets[J].IEEE Transactions on Medical Imaging,2020,39(12):4237-4248.
[65]WONG A,OTLES E,DONNELLY J P,et al.External validation of a widely implemented proprietary sepsis prediction model in hospitalized patients[J].JAMA Internal Medicine,2021,181(8):1065-1070.
[66]GUO L L,PFOHL S R,FRIES J,et al.Evaluation of domain generalization and adaptation on improving model robustness to temporal dataset shift in clinical medicine[J].Scientific Reports,2022,12(1):2726.
[67]COHEN S N,FOSTER J,FOSTER P,et al.Subtle variation in sepsis-III definitions markedly influences predictive performance within and across methods[J].Scientific Reports,2024,14(1):1920.
[68]ZHENG J,CAI X,QIU S,et al.Spurious forgetting in continual learning of language models[J].arXiv:2501.13453,2025.
[69]ZECH J R,BADGELEY M A,LIU M,et al.Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs:a cross-sectional study[J].PLoS Medicine,2018,15(11):e1002683.
[70]CARUANA R,LOU Y,GEHRKE J,et al.Intelligible Modelsfor HealthCare:Predicting Pneumonia Risk and Hospital 30-day Readmission[C]//Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.ACM,2015:1721-1730.
[71]DEGRAVE A J,JANIZEK J D,LEE S I.AI for radiographic COVID-19 detection selects shortcuts over signal[J].Nature Machine Intelligence,2021,3(7):610-619.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发