Cover Image

Personalized federated learning for in-hospital mortality prediction of multi-center ICU

Federated learning (FL), as a paradigm for addressing challenges of machine learning (ML) to be applied in private distributed data provides a novel and promising scheme to promote ML in multiple independently distributed healthcare institutions. However, the non-IID and unbalanced nature of the dat...

Full description

Saved in:
Bibliographic Details
Main Authors: Deng, Ting, Hamdan, Hazlina, Yaakob, Razali, Kasmiran, Khairul Azhar
Format: Article
Published: Institute of Electrical and Electronics Engineers 2023
Online Access:http://psasir.upm.edu.my/id/eprint/109394/
https://ieeexplore.ieee.org/document/10034741
Tags: Add Tag
No Tags, Be the first to tag this record!
id my.upm.eprints.109394
record_format eprints
spelling my.upm.eprints.1093942024-08-05T02:40:30Z http://psasir.upm.edu.my/id/eprint/109394/ Personalized federated learning for in-hospital mortality prediction of multi-center ICU Deng, Ting Hamdan, Hazlina Yaakob, Razali Kasmiran, Khairul Azhar Federated learning (FL), as a paradigm for addressing challenges of machine learning (ML) to be applied in private distributed data provides a novel and promising scheme to promote ML in multiple independently distributed healthcare institutions. However, the non-IID and unbalanced nature of the data distribution can decrease its performance, even resulting in the institutions losing motivation to participate in its training. This paper explored the problem with an in-hospital mortality prediction task under an actual multi-center ICU electronic health record database that preserves the original non-IID and unbalanced data distribution. It first analyzed the reason for the performance degradation of baseline FL under this data scenario, and then proposed a personalized FL (PFL) approach named POLA to tackle the problem. POLA is a personalized one-shot and two-step FL method capable of generating high-performance personalized models for each independent participant. The proposed method, POLA was compared with two other PFL methods in experiments, and the results indicate that it not only effectively improves the prediction performance of FL but also significantly reduces the communication rounds. Moreover, its generality and extensibility also make it potential to be extended to other similar cross-silo FL application scenarios. Institute of Electrical and Electronics Engineers 2023-02-01 Article PeerReviewed Deng, Ting and Hamdan, Hazlina and Yaakob, Razali and Kasmiran, Khairul Azhar (2023) Personalized federated learning for in-hospital mortality prediction of multi-center ICU. IEEE Access, 11. 11652- 11663. ISSN 2169-3536 https://ieeexplore.ieee.org/document/10034741 10.1109/access.2023.3241488
institution Universiti Putra Malaysia
building UPM Library
collection Institutional Repository
continent Asia
country Malaysia
content_provider Universiti Putra Malaysia
content_source UPM Institutional Repository
url_provider http://psasir.upm.edu.my/
description Federated learning (FL), as a paradigm for addressing challenges of machine learning (ML) to be applied in private distributed data provides a novel and promising scheme to promote ML in multiple independently distributed healthcare institutions. However, the non-IID and unbalanced nature of the data distribution can decrease its performance, even resulting in the institutions losing motivation to participate in its training. This paper explored the problem with an in-hospital mortality prediction task under an actual multi-center ICU electronic health record database that preserves the original non-IID and unbalanced data distribution. It first analyzed the reason for the performance degradation of baseline FL under this data scenario, and then proposed a personalized FL (PFL) approach named POLA to tackle the problem. POLA is a personalized one-shot and two-step FL method capable of generating high-performance personalized models for each independent participant. The proposed method, POLA was compared with two other PFL methods in experiments, and the results indicate that it not only effectively improves the prediction performance of FL but also significantly reduces the communication rounds. Moreover, its generality and extensibility also make it potential to be extended to other similar cross-silo FL application scenarios.
format Article
author Deng, Ting
Hamdan, Hazlina
Yaakob, Razali
Kasmiran, Khairul Azhar
spellingShingle Deng, Ting
Hamdan, Hazlina
Yaakob, Razali
Kasmiran, Khairul Azhar
Personalized federated learning for in-hospital mortality prediction of multi-center ICU
author_facet Deng, Ting
Hamdan, Hazlina
Yaakob, Razali
Kasmiran, Khairul Azhar
author_sort Deng, Ting
title Personalized federated learning for in-hospital mortality prediction of multi-center ICU
title_short Personalized federated learning for in-hospital mortality prediction of multi-center ICU
title_full Personalized federated learning for in-hospital mortality prediction of multi-center ICU
title_fullStr Personalized federated learning for in-hospital mortality prediction of multi-center ICU
title_full_unstemmed Personalized federated learning for in-hospital mortality prediction of multi-center ICU
title_sort personalized federated learning for in-hospital mortality prediction of multi-center icu
publisher Institute of Electrical and Electronics Engineers
publishDate 2023
url http://psasir.upm.edu.my/id/eprint/109394/
https://ieeexplore.ieee.org/document/10034741
_version_ 1806690500349001728
score 13.211869

Similar Items