Identifying risk of stillbirth using machine learning.
boosted trees
clinical decision-making
factor analysis
maternal serum alpha-fetoprotein)
prenatal care
previability
random forests
second-trimester prenatal screen (Down syndrome risk
structural racism
ultrasound
unconjugated estriol
Journal
American journal of obstetrics and gynecology
ISSN: 1097-6868
Titre abrégé: Am J Obstet Gynecol
Pays: United States
ID NLM: 0370476
Informations de publication
Date de publication:
09 2023
09 2023
Historique:
received:
06
03
2023
revised:
08
06
2023
accepted:
09
06
2023
pmc-release:
01
09
2024
medline:
28
8
2023
pubmed:
15
6
2023
entrez:
14
6
2023
Statut:
ppublish
Résumé
Previous predictive models using logistic regression for stillbirth do not leverage the advanced and nuanced techniques involved in sophisticated machine learning methods, such as modeling nonlinear relationships between outcomes. This study aimed to create and refine machine learning models for predicting stillbirth using data available before viability (22-24 weeks) and throughout pregnancy, as well as demographic, medical, and prenatal visit data, including ultrasound and fetal genetics. This is a secondary analysis of the Stillbirth Collaborative Research Network, which included data from pregnancies resulting in stillborn and live-born infants delivered at 59 hospitals in 5 diverse regions across the United States from 2006 to 2009. The primary aim was the creation of a model for predicting stillbirth using data available before viability. Secondary aims included refining models with variables available throughout pregnancy and determining variable importance. Among 3000 live births and 982 stillbirths, 101 variables of interest were identified. Of the models incorporating data available before viability, the random forests model had 85.1% accuracy (area under the curve) and high sensitivity (88.6%), specificity (85.3%), positive predictive value (85.3%), and negative predictive value (84.8%). A random forests model using data collected throughout pregnancy resulted in accuracy of 85.0%; this model had 92.2% sensitivity, 77.9% specificity, 84.7% positive predictive value, and 88.3% negative predictive value. Important variables in the previability model included previous stillbirth, minority race, gestational age at the earliest prenatal visit and ultrasound, and second-trimester serum screening. Applying advanced machine learning techniques to a comprehensive database of stillbirths and live births with unique and clinically relevant variables resulted in an algorithm that could accurately identify 85% of pregnancies that would result in stillbirth, before they reached viability. Once validated in representative databases reflective of the US birthing population and then prospectively, these models may provide effective risk stratification and clinical decision-making support to better identify and monitor those at risk of stillbirth.
Sections du résumé
BACKGROUND
Previous predictive models using logistic regression for stillbirth do not leverage the advanced and nuanced techniques involved in sophisticated machine learning methods, such as modeling nonlinear relationships between outcomes.
OBJECTIVE
This study aimed to create and refine machine learning models for predicting stillbirth using data available before viability (22-24 weeks) and throughout pregnancy, as well as demographic, medical, and prenatal visit data, including ultrasound and fetal genetics.
STUDY DESIGN
This is a secondary analysis of the Stillbirth Collaborative Research Network, which included data from pregnancies resulting in stillborn and live-born infants delivered at 59 hospitals in 5 diverse regions across the United States from 2006 to 2009. The primary aim was the creation of a model for predicting stillbirth using data available before viability. Secondary aims included refining models with variables available throughout pregnancy and determining variable importance.
RESULTS
Among 3000 live births and 982 stillbirths, 101 variables of interest were identified. Of the models incorporating data available before viability, the random forests model had 85.1% accuracy (area under the curve) and high sensitivity (88.6%), specificity (85.3%), positive predictive value (85.3%), and negative predictive value (84.8%). A random forests model using data collected throughout pregnancy resulted in accuracy of 85.0%; this model had 92.2% sensitivity, 77.9% specificity, 84.7% positive predictive value, and 88.3% negative predictive value. Important variables in the previability model included previous stillbirth, minority race, gestational age at the earliest prenatal visit and ultrasound, and second-trimester serum screening.
CONCLUSION
Applying advanced machine learning techniques to a comprehensive database of stillbirths and live births with unique and clinically relevant variables resulted in an algorithm that could accurately identify 85% of pregnancies that would result in stillbirth, before they reached viability. Once validated in representative databases reflective of the US birthing population and then prospectively, these models may provide effective risk stratification and clinical decision-making support to better identify and monitor those at risk of stillbirth.
Identifiants
pubmed: 37315754
pii: S0002-9378(23)00409-X
doi: 10.1016/j.ajog.2023.06.017
pmc: PMC10527568
mid: NIHMS1909202
pii:
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Langues
eng
Sous-ensembles de citation
IM
Pagination
327.e1-327.e16Subventions
Organisme : NICHD NIH HHS
ID : U10 HD045953
Pays : United States
Organisme : NICHD NIH HHS
ID : U01 HD045954
Pays : United States
Organisme : NICHD NIH HHS
ID : U10 HD045925
Pays : United States
Organisme : NICHD NIH HHS
ID : U10 HD045952
Pays : United States
Organisme : NICHD NIH HHS
ID : U10 HD045955
Pays : United States
Organisme : NICHD NIH HHS
ID : U10 HD045944
Pays : United States
Informations de copyright
Copyright © 2023 Elsevier Inc. All rights reserved.
Références
Am J Obstet Gynecol. 2009 Nov;201(5):469.e1-8
pubmed: 19762004
BMC Pregnancy Childbirth. 2021 Mar 12;21(1):202
pubmed: 33706701
BMC Med Res Methodol. 2019 Mar 19;19(1):64
pubmed: 30890124
Paediatr Perinat Epidemiol. 2018 Jan;32(1):40-51
pubmed: 28984372
Matern Child Health J. 2017 Jun;21(6):1336-1348
pubmed: 28120290
Curr Opin Obstet Gynecol. 2006 Dec;18(6):625-30
pubmed: 17099333
N Engl J Med. 2016 Jan 7;374(1):13-22
pubmed: 26735990
PLoS One. 2017 Mar 7;12(3):e0173461
pubmed: 28267756
Obstet Gynecol. 2020 Oct;136(4):e48-e69
pubmed: 32804883
J Midwifery Womens Health. 2020 Sep;65(5):621-626
pubmed: 32749063
Health Place. 2018 May;51:208-216
pubmed: 29715639
BJOG. 2021 Jan;128(2):214-224
pubmed: 32894620
Nat Mach Intell. 2019 May;1(5):206-215
pubmed: 35603010
Matern Health Neonatol Perinatol. 2016 Sep 15;2:10
pubmed: 27660718
Informatics (MDPI). 2020 Sep;7(3):
pubmed: 33274178
Clin Obstet Gynecol. 2010 Sep;53(3):700-9
pubmed: 20661054
Am J Perinatol. 2012 Nov;29(10):787-93
pubmed: 22836820
IEEE Sens J. 2022 Nov 15;22(22):21362-21390
pubmed: 36439060
BMJ Open. 2020 Oct 19;10(10):e040132
pubmed: 33077570
Paediatr Perinat Epidemiol. 2008 Jul;22(4):314-20
pubmed: 18578744
Obstet Gynecol. 2020 Mar;135(3):e110-e132
pubmed: 32080052
Paediatr Perinat Epidemiol. 2011 Sep;25(5):425-35
pubmed: 21819424
Ultrasound Obstet Gynecol. 2016 Jan;47(1):22-7
pubmed: 26031231
Int J Med Inform. 2017 Jul;103:1-6
pubmed: 28550994
Cochrane Database Syst Rev. 2020 Dec 18;12:CD009599
pubmed: 33336827
Acta Obstet Gynecol Scand. 2009;88(11):1276-83
pubmed: 19832550
BJOG. 2007 Jun;114(6):705-14
pubmed: 17516962
Acta Obstet Gynecol Scand. 2016 Nov;95(11):1313-1318
pubmed: 27588413
BJOG. 2021 Jan;128(2):238-250
pubmed: 32931648
Ultrasound Obstet Gynecol. 2016 Nov;48(5):624-630
pubmed: 27854387
Sci Rep. 2020 Mar 24;10(1):5354
pubmed: 32210300
Health Inf Sci Syst. 2020 Mar 25;8(1):14
pubmed: 32226625
Curr Opin Obstet Gynecol. 2021 Oct 1;33(5):405-411
pubmed: 34459791
Ultrasound Obstet Gynecol. 2022 Feb;59(2):209-219
pubmed: 34405928
Minerva Ginecol. 2005 Aug;57(4):397-410
pubmed: 16170285
Clin Obstet Gynecol. 2010 Sep;53(3):597-606
pubmed: 20661044
Ultrasound Obstet Gynecol. 2019 Oct;54(4):492-499
pubmed: 30549126
BJOG. 2015 Jan;122(1):41-55
pubmed: 25236870
J Clin Med. 2022 Jun 15;11(12):
pubmed: 35743521
JAMA. 2000 Nov 22-29;284(20):2611-7
pubmed: 11086368
PLoS One. 2022 Jan 20;17(1):e0260964
pubmed: 35051188
Am J Obstet Gynecol. 2020 Jun;222(6):613.e1-613.e12
pubmed: 32007491
J Obstet Gynaecol. 2020 Aug;40(6):797-802
pubmed: 31769709
EGEMS (Wash DC). 2019 Jan 24;7(1):1
pubmed: 30705919
BMC Pregnancy Childbirth. 2016 Sep 20;16:274
pubmed: 27649795
Ultrasound Obstet Gynecol. 2016 Jan;47(1):74-80
pubmed: 26327300
Am J Obstet Gynecol MFM. 2023 Jun;5(6):100968
pubmed: 37061041
J Matern Fetal Neonatal Med. 2012 Jun;25(6):699-705
pubmed: 22339200
Schizophr Res. 2021 Jan;227:52-60
pubmed: 32571619