A One-Size-Fits-Three Representation Learning Framework for Patient Similarity Search

Data Science and Engineering

Research Paper

Feng Luo¹  Zhu Di²  Bohan Li³  Yefan Huang⁴  Xiaoli Wang⁴  Bin Luo⁵ 

[1] ByteDance Inc., Beijing, China;College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China;College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China;Key Laboratory of Safety-Critical Software, Ministry of Industry and Information Technology, Nanjing, China;National Engineering Laboratory for Integrated Aero-Space-Ground Ocean Big Data Application Technology, Xi’an, China;School of Informatics, Xiamen University, Xiamen, China;Sichuan Huhui Software Co. Ltd., Mianyang, China;

关键词: Patient similarity search;    Multi-modal EHRs;    Medical concepts;    External knowledge;    Graph representation learning;   

DOI  :  10.1007/s41019-023-00216-9

 received in 2023-04-01, accepted in 2023-06-12,  发布年份 2023

来源: Springer

【 摘 要 】

Patient similarity search is an essential task in healthcare. Recent studies adopted electronic health records (EHRs) to learn patient representations for measuring the clinical similarities. These methods outperformed traditional methods, by capturing more information from various sources consisting of multi-modal EHRs, external knowledge and correlations among medical concepts. They often concerned certain type of data without taking full advantage of various information. We propose a graph representation learning framework, denoted by One-Size-Fits-Three (OSFT), that takes into account fusion-attention, neighbor-attention and global-attention from three types of information. Extensive experiments are conducted on two real datasets of MIMIC-III and MIMIC-IV, and the results verified the effectiveness and generality of our framework. When compared with baselines on patient similarity search, our framework achieved good effectiveness and comparative efficiency. The results provide new insights about whether the use of various information can better measure the patient similarity. The source codes are available at https://github.com/emmali808/ADDS/tree/master/EHRDeepHelper.

【 授权许可】

CC BY   
© The Author(s) 2023

【 图 表 】

Fig. 5

12888_2023_5172_Article_IEq27.gif

13690_2023_1170_Article_IEq115.gif

12888_2023_5172_Article_IEq26.gif

13690_2023_1170_Article_IEq2.gif

Fig. 1

13690_2023_1170_Article_IEq1.gif

Fig. 7

Fig. 1

Fig. 5

Fig. 1

Fig. 11

13063_2023_7595_Article_IEq4.gif

Fig. 5

Fig. 1

Fig. 6

Fig. 4

12888_2023_5172_Article_IEq25.gif

13690_2023_1170_Article_IEq113.gif

13690_2023_1170_Article_IEq111.gif

12888_2023_5172_Article_IEq21.gif

Fig. 2

12888_2023_5172_Article_IEq20.gif

13690_2023_1170_Article_IEq107.gif

12888_2023_5168_Article_IEq4.gif

12888_2023_5168_Article_IEq3.gif

12888_2023_5168_Article_IEq2.gif

Fig. 9

Fig. 1

Fig. 5

Fig. 5

Fig. 1

Fig. 1

13015_2023_241_Article_IEq346.gif

Fig. 4

Fig. 5

Fig. 1

Fig. 1

12888_2023_5172_Article_IEq19.gif

Fig. 6

12888_2023_5172_Article_IEq18.gif

12888_2023_5172_Article_IEq17.gif

12888_2023_5172_Article_IEq15.gif

Fig. 1

12888_2023_5172_Article_IEq13.gif

12888_2023_5172_Article_IEq12.gif

13690_2023_1170_Article_IEq100.gif

13690_2023_1170_Article_IEq99.gif

13690_2023_1170_Article_IEq98.gif

13690_2023_1170_Article_IEq97.gif

13690_2023_1170_Article_IEq96.gif

Fig. 3

Fig. 9

Fig. 4

Fig. 4

Fig. 3

13690_2023_1170_Article_IEq219.gif

Fig. 2

12936_2023_4724_Article_IEq33.gif

13690_2023_1170_Article_IEq205.gif

Fig. 7

12888_2023_5142_Article_IEq13.gif

13690_2023_1170_Article_IEq74.gif

Fig. 5

Fig.3

13690_2023_1170_Article_IEq182.gif

Fig. 6

12951_2023_2095_Article_IEq11.gif

Fig. 1

Fig. 2

Fig. 2

Fig. 3

Fig. 4

Fig. 2

Fig. 2

Fig. 2

Fig. 3

Fig. 3

13690_2023_1170_Article_IEq181.gif

13690_2023_1170_Article_IEq72.gif

13690_2023_1170_Article_IEq180.gif

Fig. 4

Fig. 6

Fig. 4

Fig. 2

Fig. 6

Fig. 3

Fig. 2

Fig. 1

40798_2023_628_Figh_HTML.png

40798_2023_628_Figg_HTML.png

Fig. 4

12951_2023_2095_Article_IEq8.gif

Fig. 1

Fig. 3

13690_2023_1170_Article_IEq64.gif

Fig. 4

Fig. 2

13690_2023_1170_Article_IEq169.gif

Fig. 6

Fig. 1

Fig. 1

42004_2023_990_Article_IEq88.gif

42004_2023_990_Article_IEq87.gif

Fig. 6

12951_2023_2095_Article_IEq6.gif

Fig. 6

12951_2023_2095_Article_IEq5.gif

42004_2023_990_Article_IEq81.gif

12951_2023_2095_Article_IEq4.gif

12951_2023_2095_Article_IEq3.gif

Fig. 8

40798_2023_628_Figd_HTML.png

Fig. 7

12936_2020_3271_Article_IEq8.gif

Fig. 4

Scheme 1

40249_2023_1135_Article_IEq5.gif

13690_2023_1170_Article_IEq60.gif

13690_2023_1170_Article_IEq58.gif

Fig. 1

13690_2023_1170_Article_IEq56.gif

Fig. 5

Fig. 7

Fig. 4

Fig. 6

13690_2023_1170_Article_IEq50.gif

Fig. 1

13690_2023_1170_Article_IEq49.gif

13395_2023_325_Article_IEq1.gif

13690_2023_1170_Article_IEq157.gif

Fig. 2

Fig. 1

13690_2023_1170_Article_IEq152.gif

Fig. 1

Fig. 1