Development and external validation of machine learning algorithms for postnatal gestational age estimation using clinical data and metabolomic markers.
Journal
PloS one
ISSN: 1932-6203
Titre abrégé: PLoS One
Pays: United States
ID NLM: 101285081
Informations de publication
Date de publication:
2023
2023
Historique:
received:
14
01
2022
accepted:
14
01
2023
entrez:
6
3
2023
pubmed:
7
3
2023
medline:
9
3
2023
Statut:
epublish
Résumé
Accurate estimates of gestational age (GA) at birth are important for preterm birth surveillance but can be challenging to obtain in low income countries. Our objective was to develop machine learning models to accurately estimate GA shortly after birth using clinical and metabolomic data. We derived three GA estimation models using ELASTIC NET multivariable linear regression using metabolomic markers from heel-prick blood samples and clinical data from a retrospective cohort of newborns from Ontario, Canada. We conducted internal model validation in an independent cohort of Ontario newborns, and external validation in heel prick and cord blood sample data collected from newborns from prospective birth cohorts in Lusaka, Zambia and Matlab, Bangladesh. Model performance was measured by comparing model-derived estimates of GA to reference estimates from early pregnancy ultrasound. Samples were collected from 311 newborns from Zambia and 1176 from Bangladesh. The best-performing model accurately estimated GA within about 6 days of ultrasound estimates in both cohorts when applied to heel prick data (MAE 0.79 weeks (95% CI 0.69, 0.90) for Zambia; 0.81 weeks (0.75, 0.86) for Bangladesh), and within about 7 days when applied to cord blood data (1.02 weeks (0.90, 1.15) for Zambia; 0.95 weeks (0.90, 0.99) for Bangladesh). Algorithms developed in Canada provided accurate estimates of GA when applied to external cohorts from Zambia and Bangladesh. Model performance was superior in heel prick data as compared to cord blood data.
Sections du résumé
BACKGROUND
Accurate estimates of gestational age (GA) at birth are important for preterm birth surveillance but can be challenging to obtain in low income countries. Our objective was to develop machine learning models to accurately estimate GA shortly after birth using clinical and metabolomic data.
METHODS
We derived three GA estimation models using ELASTIC NET multivariable linear regression using metabolomic markers from heel-prick blood samples and clinical data from a retrospective cohort of newborns from Ontario, Canada. We conducted internal model validation in an independent cohort of Ontario newborns, and external validation in heel prick and cord blood sample data collected from newborns from prospective birth cohorts in Lusaka, Zambia and Matlab, Bangladesh. Model performance was measured by comparing model-derived estimates of GA to reference estimates from early pregnancy ultrasound.
RESULTS
Samples were collected from 311 newborns from Zambia and 1176 from Bangladesh. The best-performing model accurately estimated GA within about 6 days of ultrasound estimates in both cohorts when applied to heel prick data (MAE 0.79 weeks (95% CI 0.69, 0.90) for Zambia; 0.81 weeks (0.75, 0.86) for Bangladesh), and within about 7 days when applied to cord blood data (1.02 weeks (0.90, 1.15) for Zambia; 0.95 weeks (0.90, 0.99) for Bangladesh).
CONCLUSIONS
Algorithms developed in Canada provided accurate estimates of GA when applied to external cohorts from Zambia and Bangladesh. Model performance was superior in heel prick data as compared to cord blood data.
Identifiants
pubmed: 36877673
doi: 10.1371/journal.pone.0281074
pii: PONE-D-22-01354
pmc: PMC9987787
doi:
Banques de données
Dryad
['10.5061/dryad.m37pvmd6b']
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
e0281074Subventions
Organisme : FIC NIH HHS
ID : D43 TW009340
Pays : United States
Informations de copyright
Copyright: © 2023 Hawken et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Déclaration de conflit d'intérêts
The authors have declared that no competing interests exist.
Références
BJOG. 2005 Feb;112(2):145-52
pubmed: 15663577
Am J Obstet Gynecol. 2016 Apr;214(4):511.e1-511.e13
pubmed: 26688490
Am J Obstet Gynecol. 2016 Apr;214(4):515.e1-515.e13
pubmed: 26645954
PLoS One. 2019 Feb 27;14(2):e0198919
pubmed: 30811399
Lancet. 2016 Feb 27;387(10021):844-5
pubmed: 26898853
J Perinatol. 1992 Jun;12(2):115-9
pubmed: 1522427
Lancet. 2018 May 12;391(10133):1927-1938
pubmed: 29550029
Orphanet J Rare Dis. 2019 Jan 30;14(1):25
pubmed: 30700313
Am J Obstet Gynecol. 1992 Mar;166(3):891-5
pubmed: 1550159
Gates Open Res. 2018 Dec 4;2:25
pubmed: 30706053
BMJ Open. 2017 Sep 03;7(9):e015615
pubmed: 28871012
Lancet Glob Health. 2019 Apr;7(4):e415
pubmed: 30879504
BMJ Glob Health. 2017 Jul 27;2(2):e000365
pubmed: 29104765
EBioMedicine. 2017 Feb;15:203-209
pubmed: 27939425
Elife. 2019 Mar 19;8:
pubmed: 30887951
Pediatr Res. 2014 Feb;75(2):367-73
pubmed: 24216540
J Health Popul Nutr. 2009 Jun;27(3):332-8
pubmed: 19507748
Intern Med J. 2007 May;37(5):308-14
pubmed: 17504278
Am J Obstet Gynecol. 2016 Apr;214(4):513.e1-513.e9
pubmed: 26519781
Lancet. 2015 Dec 5;386(10010):2275-86
pubmed: 26361942
Obstet Gynecol. 1984 Apr;63(4):491-5
pubmed: 6700894