Mode‐specific fetal heart rate signal dropout as a marker of perinatal asphyxia: A multicenter observational study using causal machine learning methods

Abstract Objective Cardiotocography (CTG) is the standard method for intrapartum fetal monitoring. Modern systems acquire fetal heart rate (FHR) using external Doppler ultrasound (US) or internal fetal scalp electrocardiography (ECG). We evaluated whether mode‐specific FHR signal dropout was associated with perinatal asphyxia and whether it should be retained as an informative, mode‐aware feature for downstream predictive models. Methods We conducted a retrospective multicenter cohort study using routinely collected clinical and CTG data from Mercy Hospital for Women, Melbourne, Australia, for 2010–2021; Werribee Mercy Hospital, Melbourne, Australia, for 2010–2021; and the Czech CTU‐UHB database from University Hospital Brno, for 2010–2012. A total of 46 662 labor episodes at ≥36 weeks' gestation with ≥15 min of interpretable CTG data in the final 60 min before birth were analyzed (32 242 in the model‐development cohort, 548 in the Czech validation cohort, and 13 872 in the Australian community validation cohort). The exposure was mode‐specific FHR dropout, classified as US low dropout, US high dropout (>30%), ECG low dropout, or ECG high dropout (>30%). The primary outcome was sustained perinatal asphyxia, defined using a Delphi‐consensus composite of clinical, biochemical, and neurologic criteria. Results In the model‐development cohort, 860 cases of perinatal asphyxia occurred (2.67%, 95% CI 2.49%–2.85%). Compared with US low dropout, ECG high dropout was associated with higher risk of asphyxia (inverse probability weighting IPW‐adjusted odds ratio [OR] 3.14, 95% CI 2.20–4.48; targeted maximum likelihood estimation‐adjusted OR 2.08, 95% CI 1.21–3.58), corresponding to an absolute risk increase of 2.27% (one additional case per 44 births). Survival analysis supported elevated hazards (HR 4.76, 95% CI 2.98–7.62). External validation supported the direction of association, although estimates varied across cohorts. Conclusion In retrospective multicenter analyses, high FHR dropout during ECG monitoring was associated with increased perinatal asphyxia risk. These findings support evaluating mode‐specific dropout as an informative signal‐quality feature in future predictive models. Prospective validation of such models is needed before clinical implementation.

Authors

Institutions

Publication Details

Journal
International Journal of Gynecology & Obstetrics
Published
2026-11-14
DOI
https://doi.org/10.1002/ijgo.71416
Primary Topic
Neonatal and fetal brain pathology
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Mode‐specific fetal heart rate signal dropout as a marker of perinatal asphyxia: A multicenter observational study using causal machine learning methods

Enes Makalic, Fiona Brownfoot, Debjyoti Karmakar, Marimuthu Swami Palaniswami et al.
1 links
International Journal of Gynecology & Obstetrics
Neonatal and fetal brain pathology
article

Mode‐specific fetal heart rate signal dropout as a marker of perinatal asphyxia: A multicenter observational study using causal machine learning methods

Enes Makalic, Fiona Brownfoot, Debjyoti Karmakar, Marimuthu Swami Palaniswami, Emerson Keenan, Lochana Mendis
article en

Abstract

Abstract Objective Cardiotocography (CTG) is the standard method for intrapartum fetal monitoring. Modern systems acquire fetal heart rate (FHR) using external Doppler ultrasound (US) or internal fetal scalp electrocardiography (ECG). We evaluated whether mode‐specific FHR signal dropout was associated with perinatal asphyxia and whether it should be retained as an informative, mode‐aware feature for downstream predictive models. Methods We conducted a retrospective multicenter cohort study using routinely collected clinical and CTG data from Mercy Hospital for Women, Melbourne, Australia, for 2010–2021; Werribee Mercy Hospital, Melbourne, Australia, for 2010–2021; and the Czech CTU‐UHB database from University Hospital Brno, for 2010–2012. A total of 46 662 labor episodes at ≥36 weeks' gestation with ≥15 min of interpretable CTG data in the final 60 min before birth were analyzed (32 242 in the model‐development cohort, 548 in the Czech validation cohort, and 13 872 in the Australian community validation cohort). The exposure was mode‐specific FHR dropout, classified as US low dropout, US high dropout (>30%), ECG low dropout, or ECG high dropout (>30%). The primary outcome was sustained perinatal asphyxia, defined using a Delphi‐consensus composite of clinical, biochemical, and neurologic criteria. Results In the model‐development cohort, 860 cases of perinatal asphyxia occurred (2.67%, 95% CI 2.49%–2.85%). Compared with US low dropout, ECG high dropout was associated with higher risk of asphyxia (inverse probability weighting IPW‐adjusted odds ratio [OR] 3.14, 95% CI 2.20–4.48; targeted maximum likelihood estimation‐adjusted OR 2.08, 95% CI 1.21–3.58), corresponding to an absolute risk increase of 2.27% (one additional case per 44 births). Survival analysis supported elevated hazards (HR 4.76, 95% CI 2.98–7.62). External validation supported the direction of association, although estimates varied across cohorts. Conclusion In retrospective multicenter analyses, high FHR dropout during ECG monitoring was associated with increased perinatal asphyxia risk. These findings support evaluating mode‐specific dropout as an informative signal‐quality feature in future predictive models. Prospective validation of such models is needed before clinical implementation.

International Journal of Gynecology & Obstetrics
The University of Melbourne (AU), Monash University (AU)
University of Melbourne, Medical Research Council, National Health and Medical Research Council
Good health and well-being
Neonatal and fetal brain pathology
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon