KAME: Knowledge-based attention model for diagnosis prediction in healthcare

Fenglong Ma, Radha Chitta, Quanzeng You, Jing Zhou, Houping Xiao, Jing Gao

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

The goal of diagnosis prediction task is to predict the future health information of patients from their historical Electronic Healthcare Records (EHR). The most important and challenging problem of diagnosis prediction is to design an accurate, robust and interpretable predictive model. Existing work solves this problem by employing recurrent neural networks (RNNs) with attention mechanisms, but these approaches suffer from the data sufficiency problem. To obtain good performance with insufficient data, graph-based attention models are proposed. However, when the training data are sufficient, they do not offer any improvement in performance compared with ordinary attention-based models. To address these issues, we propose KAME, an end-to-end, accurate and robust model for predicting patients' future health information. KAME not only learns reasonable embeddings for nodes in the knowledge graph, but also exploits general knowledge to improve the prediction accuracy with the proposed knowledge attention mechanism. With the learned attention weights, KAME allows us to interpret the importance of each piece of knowledge in the graph. Experimental results on three real world datasets show that the proposed KAME significantly improves the prediction performance compared with the state-of-the-art approaches, guarantees the robustness with both sufficient and insufficient data, and learns interpretable disease representations.

Original languageEnglish (US)
Title of host publicationCIKM 2018 - Proceedings of the 27th ACM International Conference on Information and Knowledge Management
EditorsNorman Paton, Selcuk Candan, Haixun Wang, James Allan, Rakesh Agrawal, Alexandros Labrinidis, Alfredo Cuzzocrea, Mohammed Zaki, Divesh Srivastava, Andrei Broder, Assaf Schuster
PublisherAssociation for Computing Machinery
Pages743-752
Number of pages10
ISBN (Electronic)9781450360142
DOIs
StatePublished - Oct 17 2018
Event27th ACM International Conference on Information and Knowledge Management, CIKM 2018 - Torino, Italy
Duration: Oct 22 2018Oct 26 2018

Publication series

NameInternational Conference on Information and Knowledge Management, Proceedings

Other

Other27th ACM International Conference on Information and Knowledge Management, CIKM 2018
CountryItaly
CityTorino
Period10/22/1810/26/18

Fingerprint

Prediction
Healthcare
Graph
Knowledge-based
Health information
Robustness
Sufficiency
Prediction accuracy
Node
Recurrent neural networks
Guarantee

All Science Journal Classification (ASJC) codes

  • Business, Management and Accounting(all)
  • Decision Sciences(all)

Cite this

Ma, F., Chitta, R., You, Q., Zhou, J., Xiao, H., & Gao, J. (2018). KAME: Knowledge-based attention model for diagnosis prediction in healthcare. In N. Paton, S. Candan, H. Wang, J. Allan, R. Agrawal, A. Labrinidis, A. Cuzzocrea, M. Zaki, D. Srivastava, A. Broder, ... A. Schuster (Eds.), CIKM 2018 - Proceedings of the 27th ACM International Conference on Information and Knowledge Management (pp. 743-752). (International Conference on Information and Knowledge Management, Proceedings). Association for Computing Machinery. https://doi.org/10.1145/3269206.3271701
Ma, Fenglong ; Chitta, Radha ; You, Quanzeng ; Zhou, Jing ; Xiao, Houping ; Gao, Jing. / KAME : Knowledge-based attention model for diagnosis prediction in healthcare. CIKM 2018 - Proceedings of the 27th ACM International Conference on Information and Knowledge Management. editor / Norman Paton ; Selcuk Candan ; Haixun Wang ; James Allan ; Rakesh Agrawal ; Alexandros Labrinidis ; Alfredo Cuzzocrea ; Mohammed Zaki ; Divesh Srivastava ; Andrei Broder ; Assaf Schuster. Association for Computing Machinery, 2018. pp. 743-752 (International Conference on Information and Knowledge Management, Proceedings).
@inproceedings{a8ed1a6f2f354a0c9d07ddc8536ac739,
title = "KAME: Knowledge-based attention model for diagnosis prediction in healthcare",
abstract = "The goal of diagnosis prediction task is to predict the future health information of patients from their historical Electronic Healthcare Records (EHR). The most important and challenging problem of diagnosis prediction is to design an accurate, robust and interpretable predictive model. Existing work solves this problem by employing recurrent neural networks (RNNs) with attention mechanisms, but these approaches suffer from the data sufficiency problem. To obtain good performance with insufficient data, graph-based attention models are proposed. However, when the training data are sufficient, they do not offer any improvement in performance compared with ordinary attention-based models. To address these issues, we propose KAME, an end-to-end, accurate and robust model for predicting patients' future health information. KAME not only learns reasonable embeddings for nodes in the knowledge graph, but also exploits general knowledge to improve the prediction accuracy with the proposed knowledge attention mechanism. With the learned attention weights, KAME allows us to interpret the importance of each piece of knowledge in the graph. Experimental results on three real world datasets show that the proposed KAME significantly improves the prediction performance compared with the state-of-the-art approaches, guarantees the robustness with both sufficient and insufficient data, and learns interpretable disease representations.",
author = "Fenglong Ma and Radha Chitta and Quanzeng You and Jing Zhou and Houping Xiao and Jing Gao",
year = "2018",
month = "10",
day = "17",
doi = "10.1145/3269206.3271701",
language = "English (US)",
series = "International Conference on Information and Knowledge Management, Proceedings",
publisher = "Association for Computing Machinery",
pages = "743--752",
editor = "Norman Paton and Selcuk Candan and Haixun Wang and James Allan and Rakesh Agrawal and Alexandros Labrinidis and Alfredo Cuzzocrea and Mohammed Zaki and Divesh Srivastava and Andrei Broder and Assaf Schuster",
booktitle = "CIKM 2018 - Proceedings of the 27th ACM International Conference on Information and Knowledge Management",

}

Ma, F, Chitta, R, You, Q, Zhou, J, Xiao, H & Gao, J 2018, KAME: Knowledge-based attention model for diagnosis prediction in healthcare. in N Paton, S Candan, H Wang, J Allan, R Agrawal, A Labrinidis, A Cuzzocrea, M Zaki, D Srivastava, A Broder & A Schuster (eds), CIKM 2018 - Proceedings of the 27th ACM International Conference on Information and Knowledge Management. International Conference on Information and Knowledge Management, Proceedings, Association for Computing Machinery, pp. 743-752, 27th ACM International Conference on Information and Knowledge Management, CIKM 2018, Torino, Italy, 10/22/18. https://doi.org/10.1145/3269206.3271701

KAME : Knowledge-based attention model for diagnosis prediction in healthcare. / Ma, Fenglong; Chitta, Radha; You, Quanzeng; Zhou, Jing; Xiao, Houping; Gao, Jing.

CIKM 2018 - Proceedings of the 27th ACM International Conference on Information and Knowledge Management. ed. / Norman Paton; Selcuk Candan; Haixun Wang; James Allan; Rakesh Agrawal; Alexandros Labrinidis; Alfredo Cuzzocrea; Mohammed Zaki; Divesh Srivastava; Andrei Broder; Assaf Schuster. Association for Computing Machinery, 2018. p. 743-752 (International Conference on Information and Knowledge Management, Proceedings).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - KAME

T2 - Knowledge-based attention model for diagnosis prediction in healthcare

AU - Ma, Fenglong

AU - Chitta, Radha

AU - You, Quanzeng

AU - Zhou, Jing

AU - Xiao, Houping

AU - Gao, Jing

PY - 2018/10/17

Y1 - 2018/10/17

N2 - The goal of diagnosis prediction task is to predict the future health information of patients from their historical Electronic Healthcare Records (EHR). The most important and challenging problem of diagnosis prediction is to design an accurate, robust and interpretable predictive model. Existing work solves this problem by employing recurrent neural networks (RNNs) with attention mechanisms, but these approaches suffer from the data sufficiency problem. To obtain good performance with insufficient data, graph-based attention models are proposed. However, when the training data are sufficient, they do not offer any improvement in performance compared with ordinary attention-based models. To address these issues, we propose KAME, an end-to-end, accurate and robust model for predicting patients' future health information. KAME not only learns reasonable embeddings for nodes in the knowledge graph, but also exploits general knowledge to improve the prediction accuracy with the proposed knowledge attention mechanism. With the learned attention weights, KAME allows us to interpret the importance of each piece of knowledge in the graph. Experimental results on three real world datasets show that the proposed KAME significantly improves the prediction performance compared with the state-of-the-art approaches, guarantees the robustness with both sufficient and insufficient data, and learns interpretable disease representations.

AB - The goal of diagnosis prediction task is to predict the future health information of patients from their historical Electronic Healthcare Records (EHR). The most important and challenging problem of diagnosis prediction is to design an accurate, robust and interpretable predictive model. Existing work solves this problem by employing recurrent neural networks (RNNs) with attention mechanisms, but these approaches suffer from the data sufficiency problem. To obtain good performance with insufficient data, graph-based attention models are proposed. However, when the training data are sufficient, they do not offer any improvement in performance compared with ordinary attention-based models. To address these issues, we propose KAME, an end-to-end, accurate and robust model for predicting patients' future health information. KAME not only learns reasonable embeddings for nodes in the knowledge graph, but also exploits general knowledge to improve the prediction accuracy with the proposed knowledge attention mechanism. With the learned attention weights, KAME allows us to interpret the importance of each piece of knowledge in the graph. Experimental results on three real world datasets show that the proposed KAME significantly improves the prediction performance compared with the state-of-the-art approaches, guarantees the robustness with both sufficient and insufficient data, and learns interpretable disease representations.

UR - http://www.scopus.com/inward/record.url?scp=85058017391&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85058017391&partnerID=8YFLogxK

U2 - 10.1145/3269206.3271701

DO - 10.1145/3269206.3271701

M3 - Conference contribution

AN - SCOPUS:85058017391

T3 - International Conference on Information and Knowledge Management, Proceedings

SP - 743

EP - 752

BT - CIKM 2018 - Proceedings of the 27th ACM International Conference on Information and Knowledge Management

A2 - Paton, Norman

A2 - Candan, Selcuk

A2 - Wang, Haixun

A2 - Allan, James

A2 - Agrawal, Rakesh

A2 - Labrinidis, Alexandros

A2 - Cuzzocrea, Alfredo

A2 - Zaki, Mohammed

A2 - Srivastava, Divesh

A2 - Broder, Andrei

A2 - Schuster, Assaf

PB - Association for Computing Machinery

ER -

Ma F, Chitta R, You Q, Zhou J, Xiao H, Gao J. KAME: Knowledge-based attention model for diagnosis prediction in healthcare. In Paton N, Candan S, Wang H, Allan J, Agrawal R, Labrinidis A, Cuzzocrea A, Zaki M, Srivastava D, Broder A, Schuster A, editors, CIKM 2018 - Proceedings of the 27th ACM International Conference on Information and Knowledge Management. Association for Computing Machinery. 2018. p. 743-752. (International Conference on Information and Knowledge Management, Proceedings). https://doi.org/10.1145/3269206.3271701