Letter to Editor

Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction

Volume: 6 Number: 1 April 27, 2026
EN TR

Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction

Abstract

Dear Editor, We read with great interest the article entitled “Prediction of Total Bilirubin Levels in Newborns Using Machine Learning”. Machine learning -based approaches for the early identification of neonatal hyperbilirubinemia represent an important and evolving field within clinical decision support systems. In the study, multiple ML algorithms were compared, and the Gradient Boosting model demonstrated the highest accuracy and AUC values in the test dataset . However, the dataset exhibited a marked class imbalance (<12.5 mg/dL: 658; ≥12.5 mg/dL: 40) . It is well established that, in imbalanced datasets, accuracy may overestimate model performance due to majority-class bias (1,2). Indeed, the study reports error rates of 85–100% in the high-risk (≥12.5 mg/dL) group . Given that the primary clinical concern in neonatal hyperbilirubinemia is the timely identification of high-risk infants, model performance within this subgroup warrants particular attention. In clinical prediction modeling, evaluation metrics should extend beyond accuracy and AUC to include sensitivity, positive predictive value, and decision curve analysis, which better reflect clinical utility (3). Missing high-risk cases may have serious neurological consequences, including bilirubin-induced neurotoxicity (5). Although the use of a single bilirubin cut-off value is practical, neonatal bilirubin assessment is typically performed using hour-specific nomograms and gestational age–adjusted risk stratification (4,7,8). Incorporating these dynamic clinical parameters into future machine learning models may enhance predictive accuracy and clinical applicability. Furthermore, the single-center design and absence of external validation limit the generalizability of the findings. Validation using independent multicenter datasets would be essential before clinical implementation. In conclusion, the study provides a valuable contribution to the application of machine learning in neonatal hyperbilirubinemia prediction. Nevertheless, improving model performance in high-risk infants, employing clinically meaningful evaluation metrics, and conducting external validation studies appear essential for translation into routine clinical practice. Sincerely.

Keywords

References

  1. 1. He H, Garcia EA. Learning from imbalanced data. IEEE Trans Knowl Data Eng. 2009;21(9):1263-1284. doi:10.1109/TKDE.2008.239
  2. 2. Saito T, Rehmsmeier M. The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLoS One. 2015;10(3):e0118432. doi:10.1371/journal.pone.0118432
  3. 3. Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Making. 2006;26(6):565-574. doi:10.1177/0272989X06295361
  4. 4. Bhutani VK, Vilms RJ, Hamerman-Johnson L. Universal bilirubin screening for severe neonatal hyperbilirubinaemia. J Perinatol. 2010;30(Suppl):S6-S15. doi:10.1038/jp.2010.98
  5. 5. Stevenson DK, Vreman HJ, Wong RJ. Bilirubin production and the risk of bilirubin neurotoxicity. Semin Perinatol. 2011;35(3):121-126. doi:10.1053/j.semperi.2011.02.005
  6. 6. Carbonell X, Botet F, Figueras J, Riu-Godó A. Prediction of hyperbilirubinaemia in the healthy term newborn. Acta Paediatr. 2001;90(2):166-170. doi:10.1080/080352501300049343
  7. 7. Newman TB, Liljestrand P, Escobar GJ. Combining clinical risk factors with serum bilirubin levels to predict hyperbilirubinemia in newborns. Arch Pediatr Adolesc Med. 2005;159(2):113-119. doi:10.1001/archpedi.159.2.113
  8. 8. Kuzniewicz MW, Escobar GJ, Wi S, Newman TB. Risk factors for severe hyperbilirubinemia among infants with borderline bilirubin levels: a nested case-control study. J Pediatr. 2008;153(2):234-240. doi:10.1016/j.jpeds.2008.01.028

Details

Primary Language

English

Subjects

Clinical Sciences (Other)

Journal Section

Letter to Editor

Publication Date

April 27, 2026

Submission Date

February 24, 2026

Acceptance Date

April 2, 2026

Published in Issue

Year 2026 Volume: 6 Number: 1

APA
Coşkun, S. M., & Şengül, M. (2026). Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction. Sağlık Bilimlerinde Yapay Zeka Dergisi, 6(1), 4-6. https://doi.org/10.52309/jaihs.1896978
AMA
1.Coşkun SM, Şengül M. Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction. JAIHS. 2026;6(1):4-6. doi:10.52309/jaihs.1896978
Chicago
Coşkun, Semih Musa, and Melih Şengül. 2026. “Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction”. Sağlık Bilimlerinde Yapay Zeka Dergisi 6 (1): 4-6. https://doi.org/10.52309/jaihs.1896978.
EndNote
Coşkun SM, Şengül M (April 1, 2026) Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction. Sağlık Bilimlerinde Yapay Zeka Dergisi 6 1 4–6.
IEEE
[1]S. M. Coşkun and M. Şengül, “Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction”, JAIHS, vol. 6, no. 1, pp. 4–6, Apr. 2026, doi: 10.52309/jaihs.1896978.
ISNAD
Coşkun, Semih Musa - Şengül, Melih. “Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction”. Sağlık Bilimlerinde Yapay Zeka Dergisi 6/1 (April 1, 2026): 4-6. https://doi.org/10.52309/jaihs.1896978.
JAMA
1.Coşkun SM, Şengül M. Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction. JAIHS. 2026;6:4–6.
MLA
Coşkun, Semih Musa, and Melih Şengül. “Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction”. Sağlık Bilimlerinde Yapay Zeka Dergisi, vol. 6, no. 1, Apr. 2026, pp. 4-6, doi:10.52309/jaihs.1896978.
Vancouver
1.Semih Musa Coşkun, Melih Şengül. Clinical Applicability of Machine Learning Models for Neonatal Bilirubin Prediction. JAIHS. 2026 Apr. 1;6(1):4-6. doi:10.52309/jaihs.1896978