Personalized disease prediction using a CNN-based similarity learning method

Qiuling Suo, Fenglong Ma, Ye Yuan, Mengdi Huai, Weida Zhong, Aidong Zhang, Jing Gao

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

17 Citations (Scopus)

Abstract

Predicting patients' risk of developing certain diseases is an important research topic in healthcare. Personalized predictive modeling, which focuses on building specific models for individual patients, has shown its advantages on utilizing heterogeneous health data compared to global models trained on the entire population. Personalized predictive models use information from similar patient cohorts, in order to capture the specific characteristics. Accurately identifying and ranking the similarity among patients based on their historical records is a key step in personalized modeling. The electric health records (EHRs), which are irregular sampled and have varied patient visit lengths, cannot be directly used to measure patient similarity due to lack of an appropriate vector representation. In this paper, we build a novel time fusion CNN framework to simultaneously learn patient representations and measure pairwise similarity. Compared to a traditional CNN, our time fusion CNN can learn not only the local temporal relationships but also the contributions from each time interval. Along with the similarity learning process, the output information which is the probability distribution is used to rank similar patients. Utilizing the similarity scores, we perform personalized disease predictions, and compare the effect of different vector representations and similarity learning metrics.

Original languageEnglish (US)
Title of host publicationProceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017
EditorsIllhoi Yoo, Jane Huiru Zheng, Yang Gong, Xiaohua Tony Hu, Chi-Ren Shyu, Yana Bromberg, Jean Gao, Dmitry Korkin
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages811-816
Number of pages6
ISBN (Electronic)9781509030491
DOIs
StatePublished - Dec 15 2017
Event2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017 - Kansas City, United States
Duration: Nov 13 2017Nov 16 2017

Publication series

NameProceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017
Volume2017-January

Other

Other2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017
CountryUnited States
CityKansas City
Period11/13/1711/16/17

Fingerprint

Learning
Fusion reactions
Health
Information use
Probability distributions
Delivery of Health Care
Research
Population

All Science Journal Classification (ASJC) codes

  • Biomedical Engineering
  • Health Informatics

Cite this

Suo, Q., Ma, F., Yuan, Y., Huai, M., Zhong, W., Zhang, A., & Gao, J. (2017). Personalized disease prediction using a CNN-based similarity learning method. In I. Yoo, J. H. Zheng, Y. Gong, X. T. Hu, C-R. Shyu, Y. Bromberg, J. Gao, ... D. Korkin (Eds.), Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017 (pp. 811-816). (Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017; Vol. 2017-January). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/BIBM.2017.8217759
Suo, Qiuling ; Ma, Fenglong ; Yuan, Ye ; Huai, Mengdi ; Zhong, Weida ; Zhang, Aidong ; Gao, Jing. / Personalized disease prediction using a CNN-based similarity learning method. Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017. editor / Illhoi Yoo ; Jane Huiru Zheng ; Yang Gong ; Xiaohua Tony Hu ; Chi-Ren Shyu ; Yana Bromberg ; Jean Gao ; Dmitry Korkin. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 811-816 (Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017).
@inproceedings{4c3abe9a032343b8b23b985c9c3b612e,
title = "Personalized disease prediction using a CNN-based similarity learning method",
abstract = "Predicting patients' risk of developing certain diseases is an important research topic in healthcare. Personalized predictive modeling, which focuses on building specific models for individual patients, has shown its advantages on utilizing heterogeneous health data compared to global models trained on the entire population. Personalized predictive models use information from similar patient cohorts, in order to capture the specific characteristics. Accurately identifying and ranking the similarity among patients based on their historical records is a key step in personalized modeling. The electric health records (EHRs), which are irregular sampled and have varied patient visit lengths, cannot be directly used to measure patient similarity due to lack of an appropriate vector representation. In this paper, we build a novel time fusion CNN framework to simultaneously learn patient representations and measure pairwise similarity. Compared to a traditional CNN, our time fusion CNN can learn not only the local temporal relationships but also the contributions from each time interval. Along with the similarity learning process, the output information which is the probability distribution is used to rank similar patients. Utilizing the similarity scores, we perform personalized disease predictions, and compare the effect of different vector representations and similarity learning metrics.",
author = "Qiuling Suo and Fenglong Ma and Ye Yuan and Mengdi Huai and Weida Zhong and Aidong Zhang and Jing Gao",
year = "2017",
month = "12",
day = "15",
doi = "10.1109/BIBM.2017.8217759",
language = "English (US)",
series = "Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "811--816",
editor = "Illhoi Yoo and Zheng, {Jane Huiru} and Yang Gong and Hu, {Xiaohua Tony} and Chi-Ren Shyu and Yana Bromberg and Jean Gao and Dmitry Korkin",
booktitle = "Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017",
address = "United States",

}

Suo, Q, Ma, F, Yuan, Y, Huai, M, Zhong, W, Zhang, A & Gao, J 2017, Personalized disease prediction using a CNN-based similarity learning method. in I Yoo, JH Zheng, Y Gong, XT Hu, C-R Shyu, Y Bromberg, J Gao & D Korkin (eds), Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017. Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017, vol. 2017-January, Institute of Electrical and Electronics Engineers Inc., pp. 811-816, 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017, Kansas City, United States, 11/13/17. https://doi.org/10.1109/BIBM.2017.8217759

Personalized disease prediction using a CNN-based similarity learning method. / Suo, Qiuling; Ma, Fenglong; Yuan, Ye; Huai, Mengdi; Zhong, Weida; Zhang, Aidong; Gao, Jing.

Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017. ed. / Illhoi Yoo; Jane Huiru Zheng; Yang Gong; Xiaohua Tony Hu; Chi-Ren Shyu; Yana Bromberg; Jean Gao; Dmitry Korkin. Institute of Electrical and Electronics Engineers Inc., 2017. p. 811-816 (Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017; Vol. 2017-January).

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

TY - GEN

T1 - Personalized disease prediction using a CNN-based similarity learning method

AU - Suo, Qiuling

AU - Ma, Fenglong

AU - Yuan, Ye

AU - Huai, Mengdi

AU - Zhong, Weida

AU - Zhang, Aidong

AU - Gao, Jing

PY - 2017/12/15

Y1 - 2017/12/15

N2 - Predicting patients' risk of developing certain diseases is an important research topic in healthcare. Personalized predictive modeling, which focuses on building specific models for individual patients, has shown its advantages on utilizing heterogeneous health data compared to global models trained on the entire population. Personalized predictive models use information from similar patient cohorts, in order to capture the specific characteristics. Accurately identifying and ranking the similarity among patients based on their historical records is a key step in personalized modeling. The electric health records (EHRs), which are irregular sampled and have varied patient visit lengths, cannot be directly used to measure patient similarity due to lack of an appropriate vector representation. In this paper, we build a novel time fusion CNN framework to simultaneously learn patient representations and measure pairwise similarity. Compared to a traditional CNN, our time fusion CNN can learn not only the local temporal relationships but also the contributions from each time interval. Along with the similarity learning process, the output information which is the probability distribution is used to rank similar patients. Utilizing the similarity scores, we perform personalized disease predictions, and compare the effect of different vector representations and similarity learning metrics.

AB - Predicting patients' risk of developing certain diseases is an important research topic in healthcare. Personalized predictive modeling, which focuses on building specific models for individual patients, has shown its advantages on utilizing heterogeneous health data compared to global models trained on the entire population. Personalized predictive models use information from similar patient cohorts, in order to capture the specific characteristics. Accurately identifying and ranking the similarity among patients based on their historical records is a key step in personalized modeling. The electric health records (EHRs), which are irregular sampled and have varied patient visit lengths, cannot be directly used to measure patient similarity due to lack of an appropriate vector representation. In this paper, we build a novel time fusion CNN framework to simultaneously learn patient representations and measure pairwise similarity. Compared to a traditional CNN, our time fusion CNN can learn not only the local temporal relationships but also the contributions from each time interval. Along with the similarity learning process, the output information which is the probability distribution is used to rank similar patients. Utilizing the similarity scores, we perform personalized disease predictions, and compare the effect of different vector representations and similarity learning metrics.

UR - http://www.scopus.com/inward/record.url?scp=85045976120&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85045976120&partnerID=8YFLogxK

U2 - 10.1109/BIBM.2017.8217759

DO - 10.1109/BIBM.2017.8217759

M3 - Conference contribution

AN - SCOPUS:85045976120

T3 - Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017

SP - 811

EP - 816

BT - Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017

A2 - Yoo, Illhoi

A2 - Zheng, Jane Huiru

A2 - Gong, Yang

A2 - Hu, Xiaohua Tony

A2 - Shyu, Chi-Ren

A2 - Bromberg, Yana

A2 - Gao, Jean

A2 - Korkin, Dmitry

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Suo Q, Ma F, Yuan Y, Huai M, Zhong W, Zhang A et al. Personalized disease prediction using a CNN-based similarity learning method. In Yoo I, Zheng JH, Gong Y, Hu XT, Shyu C-R, Bromberg Y, Gao J, Korkin D, editors, Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017. Institute of Electrical and Electronics Engineers Inc. 2017. p. 811-816. (Proceedings - 2017 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017). https://doi.org/10.1109/BIBM.2017.8217759