Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction

Roghaiyeh Gachpaz Hamed; Gaye Stephens; Mark Little; Keith Doyle; Dearbhail Ni Cathain; Adam Bowden; Pete Struthers; Sinead Impey; Jonathan Turner; Yiagmour Dogaya; Lucy Hederman; Donal Sexton

doi:10.3233/SHTI250706

Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction

Roghaiyeh Gachpaz Hamed
, Gaye Stephens
, Mark Little
, Keith Doyle
, Dearbhail Ni Cathain
, Adam Bowden
, Pete Struthers
, Sinead Impey
, Jonathan Turner
, Yiagmour Dogaya
, Lucy Hederman
, Donal Sexton

School of Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Reproducibility is essential in machine learning for healthcare (ML4H) research, particularly for its generalisability and external validation. This study investigates data mapping challenges in reproducing an acute kidney injury (AKI) prediction model within the local Electronic Health Record (EHR) system at St. James Hospital (SJH), Ireland. Key challenges include structural, syntactic, and semantic heterogeneity in EHR data and regulatory constraints. We employed a combination of expert-driven mapping, natural language processing (NLP) techniques, and standardised terminologies to align predictor variables. Despite these efforts, missing data and unit discrepancies required adaptations in feature selection and conversion methods. Our findings highlight the complexities of reproducibility in ML4H and underscore the necessity of domain expertise and standardised frameworks for cross-institutional model validation. Addressing these challenges is essential for improving generalisability and clinical impact.

Original language	English
Title of host publication	Global Healthcare Transformation in the Era of Artificial Intelligence and Informatics
Editors	John Mantas, Arie Hasman, Parisis Gallos, Emmanouil Zoulias, Konstantinos Karitis
Publisher	IOS Press BV
Pages	220-224
Number of pages	5
ISBN (Electronic)	9781643686004
DOIs	https://doi.org/10.3233/SHTI250706
Publication status	Published - 26 Jun 2025
Event	23rd Annual International Conference on Informatics, Management, and Technology in Healthcare, ICIMTH 2025 - Athens, Greece Duration: 4 Jul 2025 → 6 Jul 2025

Publication series

Name	Studies in Health Technology and Informatics
Volume	328
ISSN (Print)	0926-9630
ISSN (Electronic)	1879-8365

Conference

Conference	23rd Annual International Conference on Informatics, Management, and Technology in Healthcare, ICIMTH 2025
Country/Territory	Greece
City	Athens
Period	4/07/25 → 6/07/25

Keywords

Acute Kidney Injury Prediction
Data heterogeneity in Electronic Health Records
Machine Learning for Healthcare
Reproducibility

Access to Document

10.3233/SHTI250706

Cite this

Hamed, R. G., Stephens, G., Little, M., Doyle, K., Cathain, D. N., Bowden, A., Struthers, P., Impey, S., Turner, J., Dogaya, Y., Hederman, L., & Sexton, D. (2025). Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction. In J. Mantas, A. Hasman, P. Gallos, E. Zoulias, & K. Karitis (Eds.), Global Healthcare Transformation in the Era of Artificial Intelligence and Informatics (pp. 220-224). (Studies in Health Technology and Informatics; Vol. 328). IOS Press BV. https://doi.org/10.3233/SHTI250706

Hamed, Roghaiyeh Gachpaz ; Stephens, Gaye ; Little, Mark et al. / Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction. Global Healthcare Transformation in the Era of Artificial Intelligence and Informatics. editor / John Mantas ; Arie Hasman ; Parisis Gallos ; Emmanouil Zoulias ; Konstantinos Karitis. IOS Press BV, 2025. pp. 220-224 (Studies in Health Technology and Informatics).

@inproceedings{95d00c7a4a2741dcaf85350efb3108ec,

title = "Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction",

abstract = "Reproducibility is essential in machine learning for healthcare (ML4H) research, particularly for its generalisability and external validation. This study investigates data mapping challenges in reproducing an acute kidney injury (AKI) prediction model within the local Electronic Health Record (EHR) system at St. James Hospital (SJH), Ireland. Key challenges include structural, syntactic, and semantic heterogeneity in EHR data and regulatory constraints. We employed a combination of expert-driven mapping, natural language processing (NLP) techniques, and standardised terminologies to align predictor variables. Despite these efforts, missing data and unit discrepancies required adaptations in feature selection and conversion methods. Our findings highlight the complexities of reproducibility in ML4H and underscore the necessity of domain expertise and standardised frameworks for cross-institutional model validation. Addressing these challenges is essential for improving generalisability and clinical impact.",

keywords = "Acute Kidney Injury Prediction, Data heterogeneity in Electronic Health Records, Machine Learning for Healthcare, Reproducibility",

author = "Hamed, \{Roghaiyeh Gachpaz\} and Gaye Stephens and Mark Little and Keith Doyle and Cathain, \{Dearbhail Ni\} and Adam Bowden and Pete Struthers and Sinead Impey and Jonathan Turner and Yiagmour Dogaya and Lucy Hederman and Donal Sexton",

note = "Publisher Copyright: {\textcopyright} 2025 The Authors.; 23rd Annual International Conference on Informatics, Management, and Technology in Healthcare, ICIMTH 2025 ; Conference date: 04-07-2025 Through 06-07-2025",

year = "2025",

month = jun,

day = "26",

doi = "10.3233/SHTI250706",

language = "English",

series = "Studies in Health Technology and Informatics",

publisher = "IOS Press BV",

pages = "220--224",

editor = "John Mantas and Arie Hasman and Parisis Gallos and Emmanouil Zoulias and Konstantinos Karitis",

booktitle = "Global Healthcare Transformation in the Era of Artificial Intelligence and Informatics",

address = "Netherlands",

}

Hamed, RG, Stephens, G, Little, M, Doyle, K, Cathain, DN, Bowden, A, Struthers, P, Impey, S, Turner, J, Dogaya, Y, Hederman, L & Sexton, D 2025, Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction. in J Mantas, A Hasman, P Gallos, E Zoulias & K Karitis (eds), Global Healthcare Transformation in the Era of Artificial Intelligence and Informatics. Studies in Health Technology and Informatics, vol. 328, IOS Press BV, pp. 220-224, 23rd Annual International Conference on Informatics, Management, and Technology in Healthcare, ICIMTH 2025, Athens, Greece, 4/07/25. https://doi.org/10.3233/SHTI250706

Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction. / Hamed, Roghaiyeh Gachpaz; Stephens, Gaye; Little, Mark et al.
Global Healthcare Transformation in the Era of Artificial Intelligence and Informatics. ed. / John Mantas; Arie Hasman; Parisis Gallos; Emmanouil Zoulias; Konstantinos Karitis. IOS Press BV, 2025. p. 220-224 (Studies in Health Technology and Informatics; Vol. 328).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction

AU - Hamed, Roghaiyeh Gachpaz

AU - Stephens, Gaye

AU - Little, Mark

AU - Doyle, Keith

AU - Cathain, Dearbhail Ni

AU - Bowden, Adam

AU - Struthers, Pete

AU - Impey, Sinead

AU - Turner, Jonathan

AU - Dogaya, Yiagmour

AU - Hederman, Lucy

AU - Sexton, Donal

PY - 2025/6/26

Y1 - 2025/6/26

N2 - Reproducibility is essential in machine learning for healthcare (ML4H) research, particularly for its generalisability and external validation. This study investigates data mapping challenges in reproducing an acute kidney injury (AKI) prediction model within the local Electronic Health Record (EHR) system at St. James Hospital (SJH), Ireland. Key challenges include structural, syntactic, and semantic heterogeneity in EHR data and regulatory constraints. We employed a combination of expert-driven mapping, natural language processing (NLP) techniques, and standardised terminologies to align predictor variables. Despite these efforts, missing data and unit discrepancies required adaptations in feature selection and conversion methods. Our findings highlight the complexities of reproducibility in ML4H and underscore the necessity of domain expertise and standardised frameworks for cross-institutional model validation. Addressing these challenges is essential for improving generalisability and clinical impact.

AB - Reproducibility is essential in machine learning for healthcare (ML4H) research, particularly for its generalisability and external validation. This study investigates data mapping challenges in reproducing an acute kidney injury (AKI) prediction model within the local Electronic Health Record (EHR) system at St. James Hospital (SJH), Ireland. Key challenges include structural, syntactic, and semantic heterogeneity in EHR data and regulatory constraints. We employed a combination of expert-driven mapping, natural language processing (NLP) techniques, and standardised terminologies to align predictor variables. Despite these efforts, missing data and unit discrepancies required adaptations in feature selection and conversion methods. Our findings highlight the complexities of reproducibility in ML4H and underscore the necessity of domain expertise and standardised frameworks for cross-institutional model validation. Addressing these challenges is essential for improving generalisability and clinical impact.

KW - Acute Kidney Injury Prediction

KW - Data heterogeneity in Electronic Health Records

KW - Machine Learning for Healthcare

KW - Reproducibility

UR - https://www.scopus.com/pages/publications/105010177094

U2 - 10.3233/SHTI250706

DO - 10.3233/SHTI250706

M3 - Conference contribution

C2 - 40588914

AN - SCOPUS:105010177094

T3 - Studies in Health Technology and Informatics

SP - 220

EP - 224

BT - Global Healthcare Transformation in the Era of Artificial Intelligence and Informatics

A2 - Mantas, John

A2 - Hasman, Arie

A2 - Gallos, Parisis

A2 - Zoulias, Emmanouil

A2 - Karitis, Konstantinos

PB - IOS Press BV

T2 - 23rd Annual International Conference on Informatics, Management, and Technology in Healthcare, ICIMTH 2025

Y2 - 4 July 2025 through 6 July 2025

ER -

Hamed RG, Stephens G, Little M, Doyle K, Cathain DN, Bowden A et al. Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction. In Mantas J, Hasman A, Gallos P, Zoulias E, Karitis K, editors, Global Healthcare Transformation in the Era of Artificial Intelligence and Informatics. IOS Press BV. 2025. p. 220-224. (Studies in Health Technology and Informatics). doi: 10.3233/SHTI250706

Data Mapping Challenges in Reproducibility of Machine Learning for Acute Kidney Injury Prediction

Abstract

Publication series

Conference

Keywords

Access to Document

Fingerprint

Cite this