Embracing cohort heterogeneity in clinical machine learning development: a step toward generalizable models

Michiel Schinkel; Frank C. Bennis; Anneroos W. Boerman; W. Joost Wiersinga; Prabath W. B. Nanayakkara

doi:10.1038/s41598-023-35557-y

Embracing cohort heterogeneity in clinical machine learning development: a step toward generalizable models

Michiel Schinkel, Frank C. Bennis, Anneroos W. Boerman, W. Joost Wiersinga, Prabath W. B. Nanayakkara

Research output: Contribution to Journal › Article › Academic › peer-review

Abstract

This study is a simple illustration of the benefit of averaging over cohorts, rather than developing a prediction model from a single cohort. We show that models trained on data from multiple cohorts can perform significantly better in new settings than models based on the same amount of training data but from just a single cohort. Although this concept seems simple and obvious, no current prediction model development guidelines recommend such an approach.

Original language	English
Article number	8363
Journal	Scientific Reports
Volume	13
Issue number	1
DOIs	https://doi.org/10.1038/s41598-023-35557-y
Publication status	Published - 1 Dec 2023

Access to Document

10.1038/s41598-023-35557-y

Persistent URL (handle)

Copy link address

Cite this

@article{51d132cad67e4c4bb4d085fb690cf73d,

title = "Embracing cohort heterogeneity in clinical machine learning development: a step toward generalizable models",

abstract = "This study is a simple illustration of the benefit of averaging over cohorts, rather than developing a prediction model from a single cohort. We show that models trained on data from multiple cohorts can perform significantly better in new settings than models based on the same amount of training data but from just a single cohort. Although this concept seems simple and obvious, no current prediction model development guidelines recommend such an approach.",

author = "Michiel Schinkel and Bennis, {Frank C.} and Boerman, {Anneroos W.} and Wiersinga, {W. Joost} and Nanayakkara, {Prabath W. B.}",

year = "2023",

month = dec,

day = "1",

doi = "10.1038/s41598-023-35557-y",

language = "English",

volume = "13",

journal = "Scientific Reports",

issn = "2045-2322",

publisher = "Nature Publishing Group",

number = "1",

}

TY - JOUR

T1 - Embracing cohort heterogeneity in clinical machine learning development

T2 - a step toward generalizable models

AU - Schinkel, Michiel

AU - Bennis, Frank C.

AU - Boerman, Anneroos W.

AU - Wiersinga, W. Joost

AU - Nanayakkara, Prabath W. B.

PY - 2023/12/1

Y1 - 2023/12/1

N2 - This study is a simple illustration of the benefit of averaging over cohorts, rather than developing a prediction model from a single cohort. We show that models trained on data from multiple cohorts can perform significantly better in new settings than models based on the same amount of training data but from just a single cohort. Although this concept seems simple and obvious, no current prediction model development guidelines recommend such an approach.

AB - This study is a simple illustration of the benefit of averaging over cohorts, rather than developing a prediction model from a single cohort. We show that models trained on data from multiple cohorts can perform significantly better in new settings than models based on the same amount of training data but from just a single cohort. Although this concept seems simple and obvious, no current prediction model development guidelines recommend such an approach.

UR - http://www.scopus.com/inward/record.url?scp=85160127763&partnerID=8YFLogxK

U2 - 10.1038/s41598-023-35557-y

DO - 10.1038/s41598-023-35557-y

M3 - Article

SN - 2045-2322

VL - 13

JO - Scientific Reports

JF - Scientific Reports

IS - 1

M1 - 8363

ER -

Embracing cohort heterogeneity in clinical machine learning development: a step toward generalizable models

Abstract

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this