Identifying Birth Weight Cutoffs Based on Maternal Height and Apgar scores

Sumaiya Sultana Dola, Mir Md. Taosif Nur, Camilo Valderrama

Confirmation: I have read and agree with the IEEE BHI 2025 conference submission's policy on behalf of myself and my co-authors.

Keywords: Apgar Score, Low Birth Weight, Fuzzy Inference Model, Conditional Inference Tree, Machine Learning Models

Abstract: The birth weight cutoff suggested by the World Health Organization (under 2500g) fails to reflect health risk across populations of diverse ethnicities. Based on that, previous studies have suggested using other indicators, such as maternal height and infant sex, to derive more accurate low birth weight (LBW) cutoffs. However, such approaches have not considered fetal well-being when deriving the cutoffs. Therefore, this study addresses this limitation with a novel approach using Apgar scores to derive LBW cutoffs based on maternal height. We used the 2022 CDC birth dataset to implement a two-stage analytical approach. The first stage used a Conditional Inference Tree (CIT) and a Fuzzy Inference Model (FIM) to identify combinations of maternal height and birth weight values associated with low Apgar scores. The second stage employed an ensemble of five machine learning regressors to estimate birth weight thresholds associated with normal Apgar scores. Our experimental results indicate that adaptive cutoffs outperform the fixed 2500-gram WHO cutoff. Specifically, the WHO cutoff does not effectively scale; its ability to detect newborns with low Apgar scores diminishes as maternal height increases. Overall, this research contributes to perinatal assessment by offering a method for identifying at-risk newborns based on maternal height and infant sex.

Track: 5. Public Health Informatics

Registration Id: MPNNZ2Y7XX4

Submission Number: 53