Classification model of preterm prelabor rupture of membranes risk using an artificial neural network model based on hematological, dental, and periodontal markers.

BACKGROUND: Preterm prelabor rupture of membranes (PPROM) is a multifactorial obstetric complication that contributes significantly to neonatal morbidity and mortality. Recent evidence suggests that systemic inflammation and periodontal health are critical determinants in the etiology of PPROM. However, conventional predictive methods often fail to integrate multidomain data for early risk assessment. This study aimed to develop an ANN-based model to classify PPROM status by integrating hematological inflammatory markers, periodontal indices, and maternal behavioral factors. METHODS: A multilayer perceptron (MLP)-based artificial neural network (ANN) model was developed using data from 149 women (70 PPROM, 79 controls) and 24 input variables. Hematological parameters were derived from routine clinical blood tests obtained during late pregnancy or at hospital admission prior to delivery, while oral and periodontal indices were assessed within 24 h postpartum. Discriminative performance was evaluated using receiver operating characteristic (ROC) analysis (AUC, sensitivity, specificity), and mean squared error (MSE) was used as the training loss function. RESULTS: The ANN model demonstrated high predictive performance with an MSE of 0.049, an R2 value of 0.82, and an average deviation rate of -0.003%. These results indicate strong internal accuracy and suggest potential for generalizability. The model successfully captured complex, nonlinear relationships among biological and behavioral variables influencing PPROM risk. CONCLUSIONS: This study presents an interdisciplinary ANN framework integrating hematological, periodontal, and behavioral variables for delivery-period PPROM classification. The findings established in this study are characterized as associative given the postpartum collection of periodontal data, necessitating future prospective validation for clinical translation.