基于图卷积神经网络的孤独症谱系障碍多模态数据融合与诊断模型研究

doi:10. 20223 / j. cnki. 1000-8535. 2026. 01. 006

广州医药 ›› 2026, Vol. 57 ›› Issue (1): 39-45.DOI: 10. 20223 / j. cnki. 1000-8535. 2026. 01. 006

基于图卷积神经网络的孤独症谱系障碍多模态数据融合与诊断模型研究

黄仲羽^1,2,3,4, 吕子畔⁵, 沈峰², 严瀚³, 查彩慧⁶

1 华南理工大学附属第二医院（广州市第一人民医院）临床医学研究所（广东广州 510180）;
2 华南理工大学附属第二医院（广州市第一人民医院）营养科（广东广州 510180）;
3 华南理工大学附属第二医院（广州市第一人民医院）骨科（广东广州 510180）;
4 暨南大学中医学院（广东广州 510632）;
5 新加坡南洋理工大学生命科学学院（新加坡 639798）;
6 广州市妇女儿童医疗中心心理科（广东广州 510632）

收稿日期:2025-09-15 发布日期:2026-02-03
通讯作者: 黄仲羽,E-mail：zy1717086@163.com
基金资助:
国家自然科学基金（82004256）; 广东省基础与应用基础研究基金（2023A1515011432）

Development of an interpretable graph convolutional neural network for multimodal evidence integration and quantitative diagnosis of autism spectrum disorder

HUANG Zhongyu^1,2,3,4, LYU Zipan⁵, SHEN Feng², YAN Han³, ZHA Caihui⁶

1 Institute of Clinical Medicine,Guangzhou First People' s Hospital,the Second Affiliated Hospital,School of Medicine,South China University of Technology,Guangzhou 510180,China;
2 Department of Nutrition,Guangzhou First People' s Hospital,the Second Affiliated Hospital,School of Medicine,South China University of Technology,Guangzhou 510180,China;
3 Department of Orthopedics,Guangzhou First People' s Hospital,the Second Affiliated Hospital,School of Medicine,South China University of Technology,Guangzhou 510180,China;
4 College of Chinese Medicine,Jinan University,Guangzhou 510632,China;
5 School of Biological Sciences,Nanyang Technology University,Singapore Avenue 639798,Singapore;
6 Department of Psychology,Guangzhou Women and Children' s Medical Center,Guangzhou 510632,China

Received:2025-09-15 Published:2026-02-03

摘要/Abstract

摘要： 目的针对孤独症多模态证据融合与定量化辨识的关键问题,本研究提出基于图卷积神经网络（GCN）的孤独症谱系障碍（ASD）诊断模型研究思路。方法通过对来源于ABIDE的ASD儿童脑部fMRI数据进行整理和筛选,提取脑区功能连接矩阵作为图结构的邻接矩阵,并融合临床表型数据,构建了ASD多模态关联网络。通过网络特征比较分析,识别出了ASD与典型发育组的脑功能连接网络组间差异。进一步地构建一个端到端的GCN模型,并尝试引入注意力机制,提高模型决策的可解释性。结果该模型在诊断性能指标优于传统机器学习方法（准确率=0.710,精确率=0.709,召回率=0.780,F1=0.743,曲线下面积=0.746）。背侧注意网络与边缘系统-颞极枢纽的功能连接减弱是模型做出判断的最主要依据。结论以异质图为多模态数据整合的基本架构,本模型为ASD的潜在病理机制探索提供了新的方法学范例。

关键词: 孤独症谱系障碍, 图卷积网络, 多模态, 可解释性, 脑连接网络

Abstract: Objective To develop a quantitative model for autism spectrum disorder（ASD）integration multimodal evidences. Methods The fMRI dataset from ABIDE was used for extracting connectivity function network of ASD after data preprocessing.Difference between ASD and typical development about their brain connectivity function was evaluated with t-test. Integrating phenotypic data and fMRI dataset,an graph convolutional neural network （GCN）with attention module was estimated and compared against benchmark models about their efficiency and interpretability. Results The GCN model was evaluated outperformed other models with better accuracy indices.And regions from Dorsal Attention Network and Limbic-Temporal Pole were ranked as the highest weights for the differentiation in the model. Conclusions This study provided a novel paradigm for quantitative diagnosis and exploring pathogenesis of ASD.

Key words: autism spectrum disorder, graph convolutional neural network, multimodal, interpretability, brain connectome

黄仲羽, 吕子畔, 沈峰, 严瀚, 查彩慧. 基于图卷积神经网络的孤独症谱系障碍多模态数据融合与诊断模型研究[J]. 广州医药, 2026, 57(1): 39-45.

HUANG Zhongyu, LYU Zipan, SHEN Feng, YAN Han, ZHA Caihui. Development of an interpretable graph convolutional neural network for multimodal evidence integration and quantitative diagnosis of autism spectrum disorder[J]. Guangzhou Medical Journal, 2026, 57(1): 39-45.

参考文献

[1] CRUZ PUERTO M,SANDÍN VÁZQUEZ M.Understanding heterogeneity within autism spectrum disorder:A scoping review[J].Adv Autism,2024,10(4):314-322.
[2] FRYE R E,VASSALL S,KAUR G,et al.Emerging biomarkers in autism spectrum disorder:A systematic review[J].Ann Transl Med,2019,7(23):792.
[3] HAGHIGHAT H,MIRZAREZAEE M,ARAABI B N,et al.Functional networks abnormalities in autism spectrum disorder:Age-related hypo and hyper connectivity[J].Brain Topogr,2021,34(3):306-322.
[4] 张淑婷,韩超,徐靖尧,等.自闭症的神经影像学研究进展[J].磁共振成像,2025,16(7):91-96,128.
[5] NOMURA J,MARDO M,TAKUMI T.Molecular signatures from multi‐omics of autism spectrum disorders and schizophrenia[J].J Neurochem,2021,159(4):647-659.
[6] TANG X,FENG C,ZHAO Y,et al.A study of genetic heterogeneity in autism spectrum disorders based on plasma proteomic and metabolomic analysis:Multiomics study of autism heterogeneity[J].MedComm(2020),2023,4(5):e380.
[7] IYORTSUUN N K,KIM S H,JHON M,et al.A review of machine learning and deep learning approaches on mental health diagnosis[J].Healthcare(Basel),2023,11(3):285.
[8] KARAMPASI A,KAKKOS I,MILOULIS S-T,et al.A machine learning fMRI approach in the diagnosis of autism[C]//2020 IEEE International Conference on Big Data(Big Data),December 10-13,2020,Atlanta,GA,USA.IEEE,2020:3628-3631.
[9] LIU M,LI B,HU D.Autism spectrum disorder studies using fMRI data and machine learning:A review[J].Front Neurosci,2021(15):697870.
[10] SQUIRES M,TAO X,ELANGOVAN S,et al.Deep learning and machine learning in psychiatry:A survey of current progress in depression detection,diagnosis and treatment[J].Brain Inform,2023,10(1):10.
[11] 李昕,罗子恒,欧阳樱君,等.自闭症谱系障碍人工智能早期诊断研究进展与挑战[J/OL].广州医药,1-11[2025-10-24].https://link.cnki.net/urlid/44.1199.r.20250519.0856.004.
[12] OSAMA A,ANWAR A M,EZZELDIN S,et al.Integrative multi-omics analysis of autism spectrum disorder reveals unique microbial macromolecules interactions[J].J Adv Res,2025(77):265-279.
[13] 顾剑,钱育蓉,王兰兰,等.人工智能在功能磁共振成像数据中的自闭症研究综述[J].计算机工程与应用,2023,59(22):57-68.
[14] TIAN A,ZHANG C,RANG M,et al.RA-GCN:Relational aggregation graph convolutional network for knowledge graph completion[C]//Proceedings of the 2020 12th International Conference on Machine Learning and Computing,Shenzhen,China,ACM,2020:580-586.
[15] DI MARTINO A,YAN C G,LI Q,et al.The autism brain imaging data exchange:Towards a large-scale evaluation of the intrinsic brain architecture in autism[J].Mol Psychiatry,2014,19(6):659-667.
[16] YAN C G,ZANG Y F.DPARSF:A MATLAB toolbox for “pipeline” data analysis of resting-state fMRI[J].Front Syst Neurosci,2010(4):13.
[17] SCHAEFER A,KONG R,GORDON E M,et al.Local-global parcellation of the human cerebral cortex from intrinsic functional connectivity MRI[J].Cereb Cortex,2018,28(9):3095-3114.
[18] MOTTRON L,BZDOK D.Autism spectrum heterogeneity:Fact or artifact[J].Mol Psychiatry,2020,25(12):3178-3185.
[19] ZHANG J,FANG S,YAO Y,et al.Parsing the heterogeneity of brain-symptom associations in autism spectrum disorder via random forest with homogeneous canonical correlation[J].J Affect Disord,2023(335):36-43.
[20] JIANG H,CAO P,XU M,et al.Hi-GCN:A hierarchical graph convolution network for graph embedding learning of brain network and brain disorders prediction[J].Comput Biol Med,2020(127):104096.
[21] FITZGERALD J,JOHNSON K,KEHOE E,et al.Disrupted functional connectivity in dorsal and ventral attention networks during attention orienting in autism spectrum disorders[J].Autism Res,2015,8(2):136-152.
[22] ZAHN R,MOLL J,KRUEGER F,et al.Social concepts are represented in the superior anterior temporal cortex[J].Proc Natl Acad Sci USA,2007,104(15):6430-6435.
[23] OLSON I R,PLOTZKER A,EZZYAT Y.The Enigmatic temporal pole:A review of findings on social and emotional processing[J].Brain,2007,130(Pt 7):1718-1731.
[24] SHELINE Y I,BARCH D M,PRICE J L,et al.The default mode network and self-referential processes in depression[J].Proc Natl Acad Sci USA,2009,106(6):1942-1947.
[25] POSAR A,VISCONTI P.Sensory abnormalities in children with autism spectrum disorder[J].J Pediatr(Rio J),2018,94(4):342-350.

基于图卷积神经网络的孤独症谱系障碍多模态数据融合与诊断模型研究

Development of an interpretable graph convolutional neural network for multimodal evidence integration and quantitative diagnosis of autism spectrum disorder

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 3

编辑推荐

Metrics

本文评价

[1]	冉欣, 左晓晓, 于楠湘, 张玲. 运动干预联合治疗性聆听对孤独症谱系障碍儿童社会交往功能的影响[J]. 广州医药, 2025, 56(3): 403-409.
[2]	黄庆, 成艳玲, 凡伟, 吴文华, 郭碧华, 温慧玲. 干预因素对孤独症谱系障碍患儿社交行为影响的观察性研究[J]. 广州医药, 2024, 55(10): 1159-1165.
[3]	吴文华, 温慧玲, 郭碧华, 黄肖霞, 沈锦红. 家庭睡眠习惯调查在孤独症患儿睡眠评估中的信效度研究[J]. 广州医药, 2022, 53(3): 70-73.