目的 探讨多模态深度学习融合心脏超声与心电图特征对冠心病患者心源性猝死的预测价值。方法 选取2024年1月—2025年6月收治的60例冠心病患者，所有患者均接受心脏超声、心电图检查及多模态深度学习模型预测，随访6个月记录心源性猝死事件。根据随访结果分为事件组（n=15）和非事件组（n=45），比较两组临床资料及各项指标差异。结果 多模态深度学习模型预测敏感度为86.67%，特异度为91.11%，准确率为90.00%，AUC为0.923，显著优于单独心脏超声（AUC=0.761）和单独心电图（AUC=0.788）。结论 多模态深度学习融合心脏超声与心电图特征能够有效预测冠心病患者心源性猝死风险，预测性能优于传统单一模态评估方法。

      Objective To investigate the predictive value of multimodal deep learning integrating echocardiography and electrocardiogram features for sudden cardiac death in patients with coronary heart disease．Methods A total of 60 patients with coronary heart disease admitted from January 2024 to June 2025 were enrolled．All patients underwent echocardiography，electrocardiogram examination，and multimodal deep learning model prediction，with a 6-month follow-up to record sudden cardiac death events．According to the follow-up results，patients were divided into the event group（n=15） and non-event group（n=45），and clinical data and various indicators were compared between the two groups．Results The multimodal deep learning model achieved a sensitivity of 86.67%，specificity of 91.11%，accuracy of 90.00%，and AUC of 0.923，which were significantly superior to echocardiography alone（AUC=0.761） and electrocardiogram alone（AUC=0.788）．Conclusions Multimodal deep learning integrating echocardiography and electrocardiogram features can effectively predict the risk of sudden cardiac death in patients with coronary heart disease，with predictive performance superior to traditional single-modality assessment methods．

       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。

      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．