Figure 1.

The network architecture for the clinical sepsis prediction model. Twenty-five maternal and neonatal variables at clinical diagnosis (Maternal age, maternal morbidity, Cervicovaginitis, Urinary tract infections, Premature rupture of membranes, Chorioamnionitis, Gender, Gestational age, neonatal weight, Fever, Hypothermia, Tachycardia, Bradycardia, Tachypnea, Bradypnea, Apneas, number of leukocytes and neutrophils, Band cells number and percentage, Platelets number, Platelet, fetal morbidity, Catheter, and Mechanical ventilation) were used to for the learning procedure with the Levenberg-Marquardt optimization algorithm and sepsis estimation as the output of the model.