﻿ 基于异构图嵌入学习的相似病案推荐
 计算机系统应用  2020, Vol. 29 Issue (10): 228-234 PDF

Similar Medical Records Recommendation Based on Heterogeneous Graph Embedding Learning
WANG Yi-Fan, LI Ji-Yun
School of Computer Science and Technology, Donghua University, Shanghai 201620, China
Foundation item: National Key Research and Development Program of China (2019YFE0190500); Science and Technology Development Fund of Shanghai Municipality (18511102703)
Abstract: Medical records of patients are basic to the clinical diagnoses and treatments. Accurate recommendation of similar medical records can assist doctors in clinical decision making. In this study, we propose a new embedding model of medical records in real diagnosis and treatment scenarios. To recommend better medical records, we model the medical entities and their relationships in the medical records by heterogeneous graph embeddings. We conduct experiments on medical records of patients diagnosed with breast diseases from a Grade III-A hospital. The accuracy of the proposed model is improved by 2% compared with the existing model.
Key words: representation learning     graph embedding     recommendation of similar medical records     auxiliary diagnosis and treatment     clinical text

1 相关工作 1.1 图嵌入学习

1.2 相似病案

2 问题描述

3 病案表示模型

 图 1 基于随机游走的病案表示学习

3.1 医疗实体表示

3.2 医疗实体与患者病案

 ${e_i} = ({e_{i,e}} + {e_{i,r}})$ (1)

 $e_{i,r,t}^k = \sigma \left( {{w^k} \cdot \frac{1}{N}\mathop \sum \limits_1^N e_{j,t}^{k - 1}} \right)$ (2)

 ${e_{i,r}} = \mathop \sum \limits_1^t a_{{{i}},r}^t{e_{i,r,t}}$ (3)

 $- \log {P_\theta }\left( {{e_{i - b}}, \cdots \left. {,{e_{i - 1}},{e_{i + 1}}, \cdots ,{e_{i + b}}} \right|{e_i}} \right)$ (4)

 $P_\theta ^{}\left( {\left. {{e_j}} \right|{e_i}} \right) = \frac{{\exp \left( {e{{_{j,r}^\prime }^{ T}} \cdot {e_{i,r}}} \right)}}{{\displaystyle\sum\nolimits_{1}^{e}{\exp \left( {e{{_i^\prime }^{ T}} \cdot {e_{i,r}}} \right)}}}$ (5)

 $E = - \log \sigma \left( {e{{_{j,r}^\prime }^{T}} \cdot {e_{i,r}}} \right) - \sum\limits_{i \in e_i^j}^{} {\log \sigma \left( {e{{_i^\prime }^{ T}} \cdot {e_{i,r}}} \right)}$ (6)

3.3 患者病案表示

 ${P_x} = g({e_1},{e_2}, \cdots ,{e_n})$ (7)

 $score({P_i},{P_j}) = \frac{{{P_i} * {P_j}}}{{\left| {{P_i}} \right| \cdot \left| {{P_j}} \right|}}$ (8)

4 实验

 图 2 病案表示的建模流程

4.1 实验数据集

4.2 评价方法

4.3 参数敏感性

 图 3 参数敏感性

5 总结

 [1] Sharafoddini A, Dubin JA, Lee J. Patient similarity in prediction models based on health data: A scoping review. JMIR Medical Informatics, 2017, 5(1): e7. DOI:10.2196/medinform.6730 [2] Chan LWC, Chan T, Cheng LF, et al. Machine learning of patient similarity: A case study on predicting survival in cancer patient after locoregional chemotherapy. Proceedings of 2010 IEEE International Conference on Bioinformatics and Biomedicine Workshops (BIBMW). Hong Kong, China. 2010. 467–470. [3] Jia Z, Lu XD, Duan HL, et al. Using the distance between sets of hierarchical taxonomic clinical concepts to measure patient similarity. BMC Medical Informatics and Decision Making, 2019, 19: 91. DOI:10.1186/s12911-019-0807-y [4] Zhao FY, Xu JL, Lin Y. Similarity measure for patients via A siamese CNN network. Proceedings of the 18th International Conference on Algorithms and Architectures for Parallel Processing. Guangzhou, China. 2018. 319–328. [5] Sherry-Ann B. Patient similarity: Emerging concepts in systems and precision medicine. Frontiers in Physiology, 2016, 7: 516. DOI:10.3389/fphys.2016.00561 [6] Parimbelli E, Marini S, Sacchi L, et al. Patient similarity for precision medicine: A systematic review. Journal of Biomedical Informatics, 2018, 83: 87-96. DOI:10.1016/j.jbi.2018.06.001 [7] Perozzi B, Al-Rfou R, Skiena S. DeepWalk: Online learning of social representations. Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York, NY, USA. 2014. 701–710. [8] Grover A, Leskovec J. Node2vec: Scalable feature learning for networks. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. San Francisco, CA, USA. 2016. 855–864. [9] Tang J, Qu M, Wang MZ, et al. LINE: Large-scale information network embedding. Proceedings of the 24th International Conference on World Wide Web. Florence, Italy. 2015. 1067–1077. [10] Ahmed A, Shervashidze N, Narayanamurthy S, et al. Distributed large-scale natural graph factorization. Proceedings of the 22nd International Conference on World Wide Web. Rio de Janeiro, Brazil. 2013. 37–48. [11] Ribeiro LFR, Saverese PHP, Figueiredo DR. Struc2vec: Learning node representations from structural identity. Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Halifax, Canada. 2017. 385–394. [12] Zhao Y, Liu ZY, Sun MS. Representation learning for measuring entity relatedness with rich information. Proceedings of the 24th International Joint Conference on Artificial Intelligence. Buenos Aires, Argentina. 2015. 1412–1418. [13] Wang HW, Zhao M, Xie X, et al. Knowledge graph convolutional networks for recommender systems. Proceedings of 2019 World Wide Web Conference. San Francisco, CA, USA. 2019. 3307–3313. [14] Shang JY, Xiao C, Ma TF, et al. GAMENet: Graph augmented Memory networks for recommending medication combination. Proceedings of 33rd AAAI Conference on Artificial Intelligence. Honolulu, HI, USA. 2019. 1126–1133. [15] Suo QL, Ma FL, Ye Y, et al. Deep patient similarity learning for personalized healthcare. IEEE Transactions on Nanobioscience, 2018, 17(3): 219-227. DOI:10.1109/TNB.2018.2837622 [16] Wang F, Sun JM. PSF: A unified patient similarity evaluation framework through metric learning with weak supervision. IEEE Journal of Biomedical and Health Informatics, 2015, 19(3): 1530-1060. DOI:10.1109/JBHI.2015.2425365 [17] Ni JZ, Liu J, Zhang CX, et al. Fine-grained patient similarity measuring using deep metric learning. Proceedings of 2017 ACM on Conference on Information and Knowledge Management. Singapore. 2017. 1189–1198. [18] Zhu ZH, Yin CC, Qian BY, et al. Measuring patient similarities via a deep architecture with medical concept embedding. Proceedings of 2016 IEEE 16th International Conference on Data Mining. Barcelona, Spain. 2016. 749–758. [19] Panahiazar M, Taslimitehrani V, Pereira NL, et al. Using EHRs for heart failure therapy recommendation using multidimensional patient similarity analytics. Studies in Health Technology and Informatics, 2015, 210: 369-373. DOI:10.3233/978-1-61499-512-8-369