Meta-Analysis of Computerized, Artificial Intelligence-Driven, and Physiological Cardiotocography versus Conventional Visual Guidelines for Intrapartum and Antenatal Fetal Surveillance
Main Article Content
Abstract
Cardiotocography has been the mainstay method of monitoring fetuses during intrapartum period for more than five decades, but visual interpretation techniques using traditional framework face the problem of having considerable observer variability and false-positive results in more than 98% of cases for neonatal metabolic acidemia. Three state-of-the-art technologies, which are computerized cardiotocography, artificial intelligence, and physiologic interpretation, have been developed to overcome this limitation; however, to date, no systematic evaluation through meta-analysis has been carried out on the effectiveness of all these three modalities in diagnosing and treating clinical cases from 2021 to 2026. Studies eligible for inclusion in the review were RCTs, prospective diagnostic cohorts, and retrospective validation cohorts, which compared any new method to traditional visual analysis in pregnant women undergoing electronic fetal monitoring. Bivariate mixed-effect modeling and hierarchical summary receiver operating characteristics methods were used to summarize diagnostic performance, while random-effect models were used to combine clinical outcomes. The review identified 39 studies, including 7 trials, 28 diagnostic cohorts, and 4 multi-center validation cohorts, totaling 75,000 monitoring episodes. The pooled sensitivity was 0.88 for artificial intelligence, 0.93 for physiological cardiotocography, 0.18 for the Dawes-Redman algorithm, and 0.61 for conventional visual guidelines, with pooled specificities ranging from 0.88 to 0.92. Only physiological cardiotocography showed a significant reduction in emergency cesarean rates (RR=0.44; 95% CI: 0.31 to 0.62; p<0.001), whereas conventional-rule-automating software demonstrated no clinical benefit. In conclusion, this meta-analysis establishes that automating conventional morphological rules, regardless of computational sophistication, provides no clinical benefit over visual interpretation, meaning that guidelines must transition to physiological frameworks prioritizing prospective validation with pathophysiology-anchored labeling.
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
References
[1] G. D. Jones, W. R. Cooke, M. Vatish, and C. W. G. Redman, “Computerized Analysis of Antepartum Cardiotocography: A Review.,” Maternal-fetal medicine (Wolters Kluwer Health, Inc.), vol. 4, no. 2, pp. 130–140, Apr. 2022, doi: 10.1097/FM9.0000000000000141.
[2] A. Lovers et al., “Advancements in Fetal Heart Rate Monitoring: A Report on Opportunities and Strategic Initiatives for Better Intrapartum Care,” BJOG: An International Journal of Obstetrics & Gynaecology, vol. 132, no. 7, pp. 853–866, Jun. 2025, doi: 10.1111/1471-0528.18097.
[3] C. Hernandez Engelhart et al., “Reliability and agreement in intrapartum fetal heart rate monitoring interpretation: A systematic review,” Acta Obstetricia et Gynecologica Scandinavica, vol. 102, no. 8, pp. 970–985, Aug. 2023, doi: 10.1111/aogs.14591.
[4] P. Brocklehurst et al., “Computerised interpretation of fetal heart rate during labour (INFANT): a randomised controlled trial,” The Lancet, vol. 389, no. 10080, pp. 1719–1729, Apr. 2017, doi: 10.1016/S0140-6736(17)30568-8.
[5] A. Tsipoura, P. Giaxi, A. Sarantaki, and K. Gourounti, “Conventional Cardiotocography versus Computerized CTG Analysis and Perinatal Outcomes: a Systematic Review.,” Maedica, vol. 18, no. 3, pp. 483–489, Sep. 2023, doi: 10.26574/maedica.2023.18.3.483.
[6] I. Ben M’Barek et al., “Large-scale analysis of interobserver agreement and reliability in cardiotocography interpretation during labor using an online tool,” BMC Pregnancy and Childbirth, vol. 24, no. 1, p. 136, Feb. 2024, doi: 10.1186/s12884-024-06322-4.
[7] N. Troha, K. Razem, U. Luzovec, and M. Lucovnik, “Comparison of Four Intrapartum Cardiotocography Classifications for Predicting Neonatal Acidemia at Birth,” Journal of Pregnancy, vol. 2023, pp. 1–10, Feb. 2023, doi: 10.1155/2023/5853889.
[8] M. Oikonomou, L. Velemir, and E. Chandraharan, “Intrapartum Fetal Heart Rate Monitoring: Quo Vadis?,” in Handbook of CTG Interpretation: From Patterns to Physiology, 2nd ed., E. Chandraharan, Ed., Cambridge: Cambridge University Press, 2026, pp. 307–316. doi: 10.1017/9781009322270.044.
[9] A. Dall’Asta, C. Melito, S. Fieni, and T. Ghi, “Ectopic Atrial Beats May Cause a ZigZag Pattern at Intrapartum Recording of the Fetal Heart Rate Using Fetal Scalp Electrode,” Fetal Diagnosis and Therapy, vol. 52, no. 3, pp. 314–319, Dec. 2024, doi: 10.1159/000542935.
[10] E. Chandraharan, I. Mappa, A. Gracia Perez-Bonfils, and S. Pereira, “Implementation of physiological interpretation of fetal heart rate changes: from scientific principles to frontline practice,” The Journal of Maternal-Fetal & Neonatal Medicine, vol. 39, no. 1, Dec. 2026, doi: 10.1080/14767058.2025.2612451.
[11] A. Bhide, A. Meroni, A. Frick, and B. Thilaganathan, “The significance of meeting Dawes–Redman criteria in computerised antenatal fetal heart rate assessment,” BJOG: An International Journal of Obstetrics & Gynaecology, vol. 131, no. 2, pp. 207–212, Jan. 2024, doi: 10.1111/1471-0528.17464.
[12] T. Stampalija, A. Bhide, A. E. P. Heazell, A. Sharp, and C. Lees, “Computerized cardiotocography and Dawes–Redman criteria: how should we interpret criteria not met?,” Ultrasound in Obstetrics & Gynecology, vol. 61, no. 6, pp. 661–666, Jun. 2023, doi: 10.1002/uog.26198.
[13] M. S. Scalia et al., “Use of computerized cardiotocography and Dawes–Redman criteria: results from a binational survey,” Ultrasound in Obstetrics & Gynecology, vol. 61, no. 6, pp. 773–775, Jun. 2023, doi: 10.1002/uog.26225.
[14] G. Davis Jones, B. Albert, W. Cooke, and M. Vatish, “Performance evaluation of computerized antepartum fetal heart rate monitoring: Dawes–Redman algorithm at term,” Ultrasound in Obstetrics & Gynecology, vol. 65, no. 2, pp. 191–197, Feb. 2025, doi: 10.1002/uog.29167.
[15] I. Ben M’Barek et al., “DeepCTG® 2.0: Development and validation of a deep learning model to detect neonatal acidemia from cardiotocography during labor,” Computers in Biology and Medicine, vol. 184, p. 109448, Jan. 2025, doi: 10.1016/j.compbiomed.2024.109448.
[16] J. A. McCoy, L. D. Levine, G. Wan, C. Chivers, J. Teel, and W. G. La Cava, “Intrapartum electronic fetal heart rate monitoring to predict acidemia at birth with the use of deep learning,” American Journal of Obstetrics and Gynecology, vol. 232, no. 1, pp. 116.e1-116.e9, Jan. 2025, doi: 10.1016/j.ajog.2024.04.022.
[17] Z. S. M. Hawrami, M. A. Cengiz, and E. Dünder, “Addressing Class Imbalance in Fetal Health Classification: Rigorous Benchmarking of Multi-Class Resampling Methods on Cardiotocography Data,” Diagnostics, vol. 16, no. 3, p. 485, Feb. 2026, doi: 10.3390/diagnostics16030485.
[18] I. Nazli, E. Korbeko, S. Dogru, E. Kugu, and O. K. Sahingoz, “Early Detection of Fetal Health Conditions Using Machine Learning for Classifying Imbalanced Cardiotocographic Data,” Diagnostics, vol. 15, no. 10, p. 1250, May 2025, doi: 10.3390/diagnostics15101250.
[19] S. Pereira et al., “Is the fetus fit for labor? Introducing fast‐and‐frugal trees (FFTrees) to simplify triage of women for STAN monitoring: An interobserver agreement comparison with traditional classification,” Acta Obstetricia et Gynecologica Scandinavica, vol. 103, no. 1, pp. 68–76, Jan. 2024, doi: 10.1111/aogs.14680.
[20] S. Andres et al., “Physiological cardiotocography interpretation and neonatal morbidity: A historical pre–post cohort study in a tertiary perinatal center,” Acta Obstetricia et Gynecologica Scandinavica, May 2026, doi: 10.1111/aogs.70245.
[21] I. Ben M’Barek, G. Jauvion, and P. Ceccaldi, “Computerized cardiotocography analysis during labor – A state‐of‐the‐art review,” Acta Obstetricia et Gynecologica Scandinavica, vol. 102, no. 2, pp. 130–137, Feb. 2023, doi: 10.1111/aogs.14498.
[22] J. B. Reitsma, A. S. Glas, A. W. S. Rutjes, R. J. P. M. Scholten, P. M. Bossuyt, and A. H. Zwinderman, “Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews,” Journal of Clinical Epidemiology, vol. 58, no. 10, pp. 982–990, Oct. 2005, doi: 10.1016/j.jclinepi.2005.02.022.
[23] P. F. Whiting et al., “QUADAS-2: A Revised Tool for the Quality Assessment of Diagnostic Accuracy Studies,” Annals of Internal Medicine, vol. 155, no. 8, pp. 529–536, Oct. 2011, doi: 10.7326/0003-4819-155-8-201110180-00009.
[24] J. A. C. Sterne et al., “RoB 2: a revised tool for assessing risk of bias in randomised trials,” BMJ, p. l4898, Aug. 2019, doi: 10.1136/bmj.l4898.
[25] M. J. Page et al., “PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews,” BMJ, p. n160, Mar. 2021, doi: 10.1136/bmj.n160.
[26] G. Guyatt et al., “GRADE guidelines: 1. Introduction—GRADE evidence profiles and summary of findings tables,” Journal of Clinical Epidemiology, vol. 64, no. 4, pp. 383–394, Apr. 2011, doi: 10.1016/j.jclinepi.2010.04.026.
[27] F. Francis, S. Luz, H. Wu, S. J. Stock, and R. Townsend, “Machine learning on cardiotocography data to classify fetal outcomes: A scoping review,” Computers in Biology and Medicine, vol. 172, p. 108220, Apr. 2024, doi: 10.1016/j.compbiomed.2024.108220.
[28] R. Pardasani, R. Vitullo, S. Harris, H. O. Yapici, and J. Beard, “Development of a novel artificial intelligence algorithm for interpreting fetal heart rate and uterine activity data in cardiotocography,” Frontiers in Digital Health, vol. 7, Sep. 2025, doi: 10.3389/fdgth.2025.1638424.
[29] S. Neri, R. Ramirez Zegarra, M. Dininno, E. Di Pasquo, S. Tagliaferri, and T. Ghi, “Interobserver agreement of intrapartum cardiotocography interpretation by midwives using current FIGO and physiology-based guidelines,” The Journal of Maternal-Fetal & Neonatal Medicine, vol. 37, no. 1, Jan. 2024, doi: 10.1080/14767058.2024.2425758.
[30] D. Farsetti et al., “The impact of maternal cardiovascular status prior to labor on birth outcomes: an observational study,” Journal of Perinatal Medicine, vol. 54, no. 1, pp. 61–71, Jan. 2026, doi: 10.1515/jpm-2025-0329.
[31] S. Vanbelle, C. H. Engelhart, and E. Blix, “Measuring Agreement in Diagnostics: A Practical Guide for Researchers,” Statistics in Medicine, vol. 44, no. 23–24, Oct. 2025, doi: 10.1002/sim.70299.
[32] M. Piergianni, R. Morelli, D. I. Pio Buca, F. Di Sebastiano, M. Liberati, and F. D’Antonio, “Prenatal predictor of composite adverse outcome in pregnancies undergoing induction of labour for fetal growth restriction,” Italian Journal of Gynaecology and Obstetrics, vol. 35, no. Supplement 01, p. 30, Jan. 2023, doi: 10.36129/jog.2022.S29.
[33] G. Davis Jones, W. R. Cooke, and M. Vatish, “Identifying High-Risk Pre-Term Pregnancies Using the Fetal Heart Rate and Machine Learning,” Bioengineering, vol. 13, no. 2, p. 203, Feb. 2026, doi: 10.3390/bioengineering13020203.