<![CDATA[Preterm birth is a maternal-neonatal health problem that contributes to morbidity, mortality, and long-term healthcare needs. This study aimed to develop and conduct preliminary internal validation of a preterm birth prediction model based on chronic energy deficiency (KEK) status and anaemia among pregnant women in Bima City. This quantitative analytical study developed a prediction model for preterm birth using logistic regression. A total of 44 pregnant women were recruited through purposive sampling, based on predetermined inclusion criteria and the availability of complete data on birth outcomes, chronic energy deficiency (CED/KEK) status, anaemia, maternal age, parity, and antenatal care (ANC) visits. The analyses comprised descriptive and bivariate analyses, multivariable logistic regression, discrimination assessment using the receiver operating characteristic-area under the curve (ROC-AUC), calibration assessment using the Brier Score, threshold determination using the Youden Index, and internal validation through bootstrap resampling. Five of 44 respondents (11,36%) experienced preterm birth. The main model yielded the equation Logit(P) = -2,9517 + 0,2041(KEK) + 3,3171(Anemia). KEK yielded an adjusted odds ratio (AOR) of 1,226 (95% CI 0,069-21,715; p=0,889), while anaemia yielded an AOR of 27,579 (95% CI 2,800-271,677; p=0,004). The apparent ROC-AUC of 0,7487 decreased to 0,6725 after optimism correction using bootstrap. The apparent Brier Score was 0,0697. At a threshold of 0,5903, model sensitivity was 60,0% and specificity was 94,9%. The combination of KEK and anaemia can be formulated into a probabilistic model for predicting preterm birth, with anaemia providing the strongest predictive signal in the study sample. However, the limited number of preterm birth events resulted in unstable estimates and performance optimism. The model should be positioned as a pilot model and requires further development in a larger sample and external validation before being used for risk stratification in ANC services.]]>
