Due to the characteristics of electronic medical records (EMRs), such as the diversity of data types and temporal irregularity inherent, most existing deep learning-based methods cannot simultaneously capture static correlations between different types of clinical data and dynamic temporal dependencies between visits during the feature learning process. To address this issue, this paper proposes a disease prediction model based on multi-domain graph neural network. In this model, a temporal feature learning module that combines code level attention and time aware LSTM is first utilized to obtain the initial feature representation of patient visits. Then, based on the correlation and time interval information between different visits, a visit affinity graph and a visit sequence graph are constructed, and a graph convolutional neural network is used to mine the static and dynamic semantic associations between visit records from these graphs. Finally, a multi-domain feature fusion module based on self-attention mechanism is utilized to combine temporal features and semantic association features to obtain the final patient fusion representation for future disease prediction. The experimental results on two real clinical datasets show that our method outperforms other existing methods and achieves higher prediction accuracy.
疾病风险预测因其在疾病防控和治疗等方面突出的临床意义一直是医疗卫生领域关注的重要研究课题.传统的基于病例队列研究的疾病预测方法耗时冗长,人力、物力投入巨大,难以满足服务临床应用的需求.随着大数据、人工智能技术的发展以及医院信息化程度的提高,各个医疗机构积累了丰富的临床电子病历(electronic medical records, EMR)数据,为研究人员提供了大量类型多样、易于获取的患者诊疗全流程临床信息.因此,如何利用深度学习技术对电子病历数据进行分析,挖掘患者疾病发展的潜在模式以支持疾病风险预测等相关临床应用的辅助决策,一直是该领域的研究热点之一.
本文研究中使用了MIMIC-Ⅲ[34]和MIMIC-Ⅳ[35]两个数据集来进行模型的性能评估.其中,MIMIC-Ⅲ是一个大型的免费医疗数据集,该数据集收集了2001年至2012年之间波士顿贝斯以色列医院(Beth Israel Deaconess Medical Center)重症监护病房收治的4万多患者的诊疗记录.MIMIC-Ⅳ是对MIMIC-Ⅲ数据集的改进和扩展,记录了共256 878名患者的就诊信息,其中包括超过50 000名有重症监护单位经历的患者.
MAF L, CHITTAR, ZHOUJ,et al .Dipole:diagnosis prediction in healthcare via attention-based bidirectional recurrent neural networks[C]//Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.Halifax NS, Canada.ACM,2017:13-17.
[2]
KIMY J, LEEY G, KIMJ W, et al. Highrisk prediction from electronic medical records via deep attention networks[C]//Proceedings of the 31st Conference on Neural Information Processing Systems. Long Beach, CA, USA. ACM, 2017:1-10.
[3]
ANY, HUANGN J, CHENX L,et al. High-risk prediction of cardiovascular diseases via attention-based deep neural networks[J]. IEEE/ACM Transactions on Computational Biology and Bioinformatics,2021,18(3):1093-1105.
[4]
BAYTASI M, XIAOC, ZHANGX,et al .Patient subtyping via time-aware LSTM networks[C]//Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Halifax,NS,Canada. ACM, 2017: 65-74.
[5]
LUOJ Y, YEM C, XIAOC,et al .HiTANet:hierarchical time-aware attention networks for risk prediction on electronic health records[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. Virtual Event,CA,USA. ACM,2020:647-656.
[6]
LIR, GAOJ .Multi-modal contrastive learning for healthcare data analytics[C]//2022 IEEE 10th International Conference on Healthcare Informatics (ICHI). Rochester,MN,USA. IEEE,2022:120-127.
[7]
NIUS, YINQ, SONGY Y,et al .Label dependent attention model for disease risk prediction using multimodal electronic health records[C]//2021 IEEE International Conference on Data Mining (ICDM). Auckland,New Zealand. IEEE, 2021: 449-458.
[8]
MENGY W, SPEIERW, ONGM K,et al. Bidirectional representation learning from transformers using multimodal electronic health record data to predict depression[J]. IEEE Journal of Biomedical and Health Informatics,2021,25(8):3121-3129.
[9]
CHOIE, BAHADORIM T, SONGL,et al .GRAM[C]//Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Halifax,NS,Canada. ACM,2017:787-795.
[10]
MAF L, YOUQ Z, XIAOH P,et al .KAME[C]//Proceedings of the 27th ACM International Conference on Information and Knowledge Management. Torino,Italy. ACM,2018:743-752.
[11]
LIY, QIANB Y, ZHANGX L,et al .Knowledge guided diagnosis prediction via graph spatial-temporal network[C]//Proceedings of the 2020 SIAM International Conference on Data Mining.Philadelphia,PA:Society for Industrial and Applied Mathematics,2020:19-27.
[12]
CHOIE, XIAOC, SUNJ, et al. MiME: multilevel medical embedding of electronic health records for predictive healthcare[C] //Proceedings of the 32st Conference on Neural Information Processing Systems. Montreal, QC, Canada. ACM, 2018:4547-4557.
[13]
MENGQ W, QIANH W, LIUY,et al .MHCCL: Masked hierarchical cluster-wise contrastive learning for multivariate time series[C]//Proceedings of the AAAI Conference on Artificial Intelligence. Washington, DC, USA. ACM, 2023:9153-9161.
[14]
ANSARIA F, HENGA,LIM A,et al .Neural continuous-discrete state space models for irregularly-sampled time series[C]//Proceedings of the 40th International Conference on Machine Learning. Honolulu,Hawaii,USA. ACM,2023:926-951.
[15]
RAGHUA, CHANDAKP, ALAMR,et al .Sequential multi-dimensional self-supervised learning for clinical time series[C]//Proceedings of the 40th International Conference on Machine Learning. Honolulu,Hawaii,USA. ACM,2023:28531-28548.
[16]
NGUYENP, TRANT, WICKRAMASINGHEN,et al .Deepr: a convolutional net for medical records[J].IEEE Journal of Biomedical and Health Informatics,2017,21(1):22-30.
[17]
LEEW, PARKS,JOO W,et al .Diagnosis prediction via medical context attention networks using deep generative modeling[C]//2018 IEEE International Conference on Data Mining (ICDM). Singapore.IEEE,2018:1104-1109.
[18]
YEM C, LUOJ Y, XIAOC,et al .LSAN:modeling long-term dependencies and short-term correlations with hierarchical attention for risk prediction[C]//Proceedings of the 29th ACM International Conference on Information & Knowledge Management. Virtual Event,Ireland. ACM,2020:1753-1762.
[19]
TANQ X, YEM, YANGB Y,et al .DATA-GRU:dual-attention time-aware gated recurrent unit for irregular multivariate time series[C]//Proceedings of the AAAI Conference on Artificial Intelligence. New York, USA. ACM, 2020:930-937.
[20]
ANY, TANGK, WANGJ X. Time-aware multi-type data fusion representation learning framework for risk prediction of cardio vascular diseases[J]. IEEE/ACM Transactions on Computational Biology and Bioinformatics,2022, 19(6): 3725-3734.
[21]
ANY, CAIG L, CHENX L,et al. PARSE:a personalized clinical time-series representation learning framework via abnormal offsets analysis[J]. Computer Methods and Programs in Biomedicine,2023, 242: 107838.
[22]
ANY, ZHANGH J, SHENGY, et al. MAIN:multimodal attention-based fusion networks for diagnosis prediction[C]//2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). Houston,TX,USA. IEEE,2021:809-816.
[23]
ZHANGC H, CHUX, MAL T,et al .M3Care:learning with missing modalities in multimodal healthcare data[C]//Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. Washington DC, USA. ACM,2022:2418-2428.
[24]
LIUS C, WANGX L, XIANGY,et al .Multi-channel fusion LSTM for medical event prediction using EHRs[J].Journal of Biomedical Informatics,2022,127:104011.
[25]
SCARSELLIF, GORIM, TSOIA C,et al .The graph neural network model[J].IEEE Transactions on Neural Networks,2009,20(1):61-80.
[26]
KANGY, ZHENGJ Y, YANGM J,et al. Inter-structure and intra-semantics graph contrastive learning for disease prediction[J]. Knowledge-Based Systems, 2024, 300: 112059.
[27]
WANGY C, XUY C, YANGJ F,et al .Fully-connected spatial-temporal graph for multivariate time-series data[J].Proceedings of the AAAI Conference on Artificial Intelligence,2024,38(14):15715-15724.
[28]
CHANT H, YING S,BAE K,et al .Multi-task heterogeneous graph learning on electronic health records[J].Neural Networks,2024,180:106644.
[29]
YEM C, CUIS H, WANGY Q,et al .MedPath:augmenting health risk prediction via medical knowledge paths[C]//Proceedings of the Web Conference 2021. Ljubljana,Slovenia. ACM,2021: 1397-1409.
[30]
CHOIE, XUZ, LIY J,et al .Learning the graphical structure of electronic health records with graph convolutional transformer[J].Proceedings of the AAAI Conference on Artificial Intelligence,2020,34(1):606-613.
[31]
ANY, SHIY H, GUOL,et al .Knowledge-enhanced difference-aware clinical time series representation learning for diagnosis prediction[C]//2023 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). Istanbul,Turkiye. IEEE,2023:1014-1021.
[32]
WUT, WANGY L, WANGY,et al. Leveraging graph- based hierarchical medical entity embedding for healthcare applications[J]. Scientific Reports, 2021, 11(1): 5858.
[33]
ANY, LIR Z, CHENX L. MERGE: a multi-graph attentive representation learning framework integrating group information from similar patients[J]. Computers in Biology and Medicine,2022,151:106245.
[34]
JOHNSONA E W, POLLARDT J, SHENL,et al. MIMIC-Ⅲ,a freely accessible critical care database[J]. Scientific Data, 2016,3: 160035.
[35]
JOHNSONA E W, BULGARELLIL, SHENL,et al. MIMIC-Ⅳ,a freely accessible electronic health record dataset[J]. Scientific Data, 2023, 10(1): 1.
基金资助
湖南省教育厅科学研究重点项目(23A0702)
Key Scientific Research Project of Hunan Provincial Department of Education(23A0702)