A heterogeneous multi-task learning for predicting RBC transfusion and perioperative outcomes

Che Ngufor, Sudhindra Upadhyaya, Dennis Murphree, Nageswar Madde, Daryl J Kor, Jyotishman Pathak

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

4 Citations (Scopus)

Abstract

It would be desirable before a surgical procedure to have a prediction rule that could accurately estimate the probability of a patient bleeding, need for blood transfusion, and other important outcomes. Such a prediction rule would allow optimal planning, more efficient use of blood bank resources, and identification of high-risk patient cohort for specific perioperative interventions. The goal of this study is to develop an efficient and accurate algorithm that could estimate the risk of multiple outcomes simultaneously. Specifically, a heterogeneous multi-task learning method is proposed for learning outcomes such as perioperative bleeding, intraoperative RBC transfusion, ICU care, and ICU length of stay. Additional outcomes not normally predicted are incorporated in the model for transfer learning and help improve the performance of relevant outcomes. Results for predicting perioperative bleeding and need for blood transfusion for patients undergoing non-cardiac operations from an institutional transfusion datamart show that the proposed method significantly increases AUC and G-Mean by more than 6% and 5% respectively over standard single-task learning methods.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages287-297
Number of pages11
Volume9105
ISBN (Print)9783319195506
DOIs
StatePublished - 2015
Event15th Conference on Artificial Intelligence in Medicine, AIME 2015 - Pavia, Italy
Duration: Jun 17 2015Jun 20 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9105
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other15th Conference on Artificial Intelligence in Medicine, AIME 2015
CountryItaly
CityPavia
Period6/17/156/20/15

Fingerprint

Multi-task Learning
Red Blood Cells
Intensive care units
Blood
Multiple Outcomes
Transfer Learning
Prediction
Estimate
Planning
Resources
Learning

Keywords

  • Blood transfusion
  • Classification
  • Health care
  • Machine learning
  • Multi-task Learning
  • Regression

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Ngufor, C., Upadhyaya, S., Murphree, D., Madde, N., Kor, D. J., & Pathak, J. (2015). A heterogeneous multi-task learning for predicting RBC transfusion and perioperative outcomes. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9105, pp. 287-297). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9105). Springer Verlag. https://doi.org/10.1007/978-3-319-19551-3_37

A heterogeneous multi-task learning for predicting RBC transfusion and perioperative outcomes. / Ngufor, Che; Upadhyaya, Sudhindra; Murphree, Dennis; Madde, Nageswar; Kor, Daryl J; Pathak, Jyotishman.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9105 Springer Verlag, 2015. p. 287-297 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9105).

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

Ngufor, C, Upadhyaya, S, Murphree, D, Madde, N, Kor, DJ & Pathak, J 2015, A heterogeneous multi-task learning for predicting RBC transfusion and perioperative outcomes. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 9105, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9105, Springer Verlag, pp. 287-297, 15th Conference on Artificial Intelligence in Medicine, AIME 2015, Pavia, Italy, 6/17/15. https://doi.org/10.1007/978-3-319-19551-3_37
Ngufor C, Upadhyaya S, Murphree D, Madde N, Kor DJ, Pathak J. A heterogeneous multi-task learning for predicting RBC transfusion and perioperative outcomes. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9105. Springer Verlag. 2015. p. 287-297. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-19551-3_37
Ngufor, Che ; Upadhyaya, Sudhindra ; Murphree, Dennis ; Madde, Nageswar ; Kor, Daryl J ; Pathak, Jyotishman. / A heterogeneous multi-task learning for predicting RBC transfusion and perioperative outcomes. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9105 Springer Verlag, 2015. pp. 287-297 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{8b15bfb5f25c4d69b484c050ceca1a09,
title = "A heterogeneous multi-task learning for predicting RBC transfusion and perioperative outcomes",
abstract = "It would be desirable before a surgical procedure to have a prediction rule that could accurately estimate the probability of a patient bleeding, need for blood transfusion, and other important outcomes. Such a prediction rule would allow optimal planning, more efficient use of blood bank resources, and identification of high-risk patient cohort for specific perioperative interventions. The goal of this study is to develop an efficient and accurate algorithm that could estimate the risk of multiple outcomes simultaneously. Specifically, a heterogeneous multi-task learning method is proposed for learning outcomes such as perioperative bleeding, intraoperative RBC transfusion, ICU care, and ICU length of stay. Additional outcomes not normally predicted are incorporated in the model for transfer learning and help improve the performance of relevant outcomes. Results for predicting perioperative bleeding and need for blood transfusion for patients undergoing non-cardiac operations from an institutional transfusion datamart show that the proposed method significantly increases AUC and G-Mean by more than 6{\%} and 5{\%} respectively over standard single-task learning methods.",
keywords = "Blood transfusion, Classification, Health care, Machine learning, Multi-task Learning, Regression",
author = "Che Ngufor and Sudhindra Upadhyaya and Dennis Murphree and Nageswar Madde and Kor, {Daryl J} and Jyotishman Pathak",
year = "2015",
doi = "10.1007/978-3-319-19551-3_37",
language = "English (US)",
isbn = "9783319195506",
volume = "9105",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "287--297",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - A heterogeneous multi-task learning for predicting RBC transfusion and perioperative outcomes

AU - Ngufor, Che

AU - Upadhyaya, Sudhindra

AU - Murphree, Dennis

AU - Madde, Nageswar

AU - Kor, Daryl J

AU - Pathak, Jyotishman

PY - 2015

Y1 - 2015

N2 - It would be desirable before a surgical procedure to have a prediction rule that could accurately estimate the probability of a patient bleeding, need for blood transfusion, and other important outcomes. Such a prediction rule would allow optimal planning, more efficient use of blood bank resources, and identification of high-risk patient cohort for specific perioperative interventions. The goal of this study is to develop an efficient and accurate algorithm that could estimate the risk of multiple outcomes simultaneously. Specifically, a heterogeneous multi-task learning method is proposed for learning outcomes such as perioperative bleeding, intraoperative RBC transfusion, ICU care, and ICU length of stay. Additional outcomes not normally predicted are incorporated in the model for transfer learning and help improve the performance of relevant outcomes. Results for predicting perioperative bleeding and need for blood transfusion for patients undergoing non-cardiac operations from an institutional transfusion datamart show that the proposed method significantly increases AUC and G-Mean by more than 6% and 5% respectively over standard single-task learning methods.

AB - It would be desirable before a surgical procedure to have a prediction rule that could accurately estimate the probability of a patient bleeding, need for blood transfusion, and other important outcomes. Such a prediction rule would allow optimal planning, more efficient use of blood bank resources, and identification of high-risk patient cohort for specific perioperative interventions. The goal of this study is to develop an efficient and accurate algorithm that could estimate the risk of multiple outcomes simultaneously. Specifically, a heterogeneous multi-task learning method is proposed for learning outcomes such as perioperative bleeding, intraoperative RBC transfusion, ICU care, and ICU length of stay. Additional outcomes not normally predicted are incorporated in the model for transfer learning and help improve the performance of relevant outcomes. Results for predicting perioperative bleeding and need for blood transfusion for patients undergoing non-cardiac operations from an institutional transfusion datamart show that the proposed method significantly increases AUC and G-Mean by more than 6% and 5% respectively over standard single-task learning methods.

KW - Blood transfusion

KW - Classification

KW - Health care

KW - Machine learning

KW - Multi-task Learning

KW - Regression

UR - http://www.scopus.com/inward/record.url?scp=84947968546&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84947968546&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-19551-3_37

DO - 10.1007/978-3-319-19551-3_37

M3 - Conference contribution

SN - 9783319195506

VL - 9105

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 287

EP - 297

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

PB - Springer Verlag

ER -