Prediction of ICU in-hospital mortality using a deep Boltzmann machine and dropout neural net

Daniel P. Ryan, Brian Daley, Kwai Wong, Xiaopeng Zhao

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

1 Citation (Scopus)

Abstract

The capability to predict in-hospital mortality of patients in intensive care units will be of paramount importance. We explore state-of-the-art machine learning techniques to estimate the in-hospital mortality probability of a patient using various physiological measurements taken within the first forty-eight hours of patient admission. A generative model, a deep Boltzmann machine, is trained using a set of recently developed techniques to automatically extract features from the patient data, and then used to initialize a feed-forward neural network. The neural network is then discriminatively fine-tuned using an efficient approximation to an ensemble of neural networks, dropout, to prevent overfitting on the limited number of labeled training examples.

Original languageEnglish (US)
Title of host publicationProceedings of the 2013 4th Annual ORNL Biomedical Sciences and Engineering Conference
Subtitle of host publicationCollaborative Biomedical Innovations, BSEC 2013
DOIs
StatePublished - Nov 21 2013
Event2013 4th Annual ORNL Biomedical Sciences and Engineering Conference: Collaborative Biomedical Innovations, BSEC 2013 - Oak Ridge, TN, United States
Duration: May 21 2013May 23 2013

Other

Other2013 4th Annual ORNL Biomedical Sciences and Engineering Conference: Collaborative Biomedical Innovations, BSEC 2013
CountryUnited States
CityOak Ridge, TN
Period5/21/135/23/13

Fingerprint

Intensive care units
Neural networks
Feedforward neural networks
Learning systems

All Science Journal Classification (ASJC) codes

  • Biomedical Engineering

Cite this

Ryan, D. P., Daley, B., Wong, K., & Zhao, X. (2013). Prediction of ICU in-hospital mortality using a deep Boltzmann machine and dropout neural net. In Proceedings of the 2013 4th Annual ORNL Biomedical Sciences and Engineering Conference: Collaborative Biomedical Innovations, BSEC 2013 [6618491] https://doi.org/10.1109/BSEC.2013.6618491

Prediction of ICU in-hospital mortality using a deep Boltzmann machine and dropout neural net. / Ryan, Daniel P.; Daley, Brian; Wong, Kwai; Zhao, Xiaopeng.

Proceedings of the 2013 4th Annual ORNL Biomedical Sciences and Engineering Conference: Collaborative Biomedical Innovations, BSEC 2013. 2013. 6618491.

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

Ryan, DP, Daley, B, Wong, K & Zhao, X 2013, Prediction of ICU in-hospital mortality using a deep Boltzmann machine and dropout neural net. in Proceedings of the 2013 4th Annual ORNL Biomedical Sciences and Engineering Conference: Collaborative Biomedical Innovations, BSEC 2013., 6618491, 2013 4th Annual ORNL Biomedical Sciences and Engineering Conference: Collaborative Biomedical Innovations, BSEC 2013, Oak Ridge, TN, United States, 5/21/13. https://doi.org/10.1109/BSEC.2013.6618491
Ryan DP, Daley B, Wong K, Zhao X. Prediction of ICU in-hospital mortality using a deep Boltzmann machine and dropout neural net. In Proceedings of the 2013 4th Annual ORNL Biomedical Sciences and Engineering Conference: Collaborative Biomedical Innovations, BSEC 2013. 2013. 6618491 https://doi.org/10.1109/BSEC.2013.6618491
Ryan, Daniel P. ; Daley, Brian ; Wong, Kwai ; Zhao, Xiaopeng. / Prediction of ICU in-hospital mortality using a deep Boltzmann machine and dropout neural net. Proceedings of the 2013 4th Annual ORNL Biomedical Sciences and Engineering Conference: Collaborative Biomedical Innovations, BSEC 2013. 2013.
@inproceedings{2ac0de244a9f42b9b76effb540b178fd,
title = "Prediction of ICU in-hospital mortality using a deep Boltzmann machine and dropout neural net",
abstract = "The capability to predict in-hospital mortality of patients in intensive care units will be of paramount importance. We explore state-of-the-art machine learning techniques to estimate the in-hospital mortality probability of a patient using various physiological measurements taken within the first forty-eight hours of patient admission. A generative model, a deep Boltzmann machine, is trained using a set of recently developed techniques to automatically extract features from the patient data, and then used to initialize a feed-forward neural network. The neural network is then discriminatively fine-tuned using an efficient approximation to an ensemble of neural networks, dropout, to prevent overfitting on the limited number of labeled training examples.",
author = "Ryan, {Daniel P.} and Brian Daley and Kwai Wong and Xiaopeng Zhao",
year = "2013",
month = "11",
day = "21",
doi = "10.1109/BSEC.2013.6618491",
language = "English (US)",
isbn = "9781479921188",
booktitle = "Proceedings of the 2013 4th Annual ORNL Biomedical Sciences and Engineering Conference",

}

TY - GEN

T1 - Prediction of ICU in-hospital mortality using a deep Boltzmann machine and dropout neural net

AU - Ryan, Daniel P.

AU - Daley, Brian

AU - Wong, Kwai

AU - Zhao, Xiaopeng

PY - 2013/11/21

Y1 - 2013/11/21

N2 - The capability to predict in-hospital mortality of patients in intensive care units will be of paramount importance. We explore state-of-the-art machine learning techniques to estimate the in-hospital mortality probability of a patient using various physiological measurements taken within the first forty-eight hours of patient admission. A generative model, a deep Boltzmann machine, is trained using a set of recently developed techniques to automatically extract features from the patient data, and then used to initialize a feed-forward neural network. The neural network is then discriminatively fine-tuned using an efficient approximation to an ensemble of neural networks, dropout, to prevent overfitting on the limited number of labeled training examples.

AB - The capability to predict in-hospital mortality of patients in intensive care units will be of paramount importance. We explore state-of-the-art machine learning techniques to estimate the in-hospital mortality probability of a patient using various physiological measurements taken within the first forty-eight hours of patient admission. A generative model, a deep Boltzmann machine, is trained using a set of recently developed techniques to automatically extract features from the patient data, and then used to initialize a feed-forward neural network. The neural network is then discriminatively fine-tuned using an efficient approximation to an ensemble of neural networks, dropout, to prevent overfitting on the limited number of labeled training examples.

UR - http://www.scopus.com/inward/record.url?scp=84887725298&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84887725298&partnerID=8YFLogxK

U2 - 10.1109/BSEC.2013.6618491

DO - 10.1109/BSEC.2013.6618491

M3 - Conference contribution

AN - SCOPUS:84887725298

SN - 9781479921188

BT - Proceedings of the 2013 4th Annual ORNL Biomedical Sciences and Engineering Conference

ER -