Prediction of Emergency Room Arrivals of Patients with Preeclampsia Disease Using Artificial Neural Network Model

Abstract

The administration of a country's public health is an essential issue. Conversely, preeclampsia ranks among the most frequent causes of maternal fatalities during pregnancy. Moreover, as in many nations, emergency rooms are severely overcrowded. Thus, we enhanced the management of Emergency Care for Preeclampsia at the “IESS Hospital del Día Sur Valdivia” so as not to risk the patients' lives. Techniques such as Extreme Learning Machine (ELM), Multilayer Perceptron (MLP), and Dynamic Bayesian Networks (DBN) were used to train the forecasting models. Mean Absolute Percentage Error (MAPE), Root-mean-square error (RMSE), and Mean Absolute Error (MAE) metrics were used to assess the model's performance. The model with the lowest MAPE value was chosen among the trained models. The model trained with MLP achieved the lowest MAPE, whose value was 17.21%. This model also achieved an RMSE of 4.21 and a MAE of 2.74. According to the MLP model, 11 to 13 emergency arrivals due to preeclampsia were predicted for September to November of 2024 in the “IESS Hospital del Día Sur Valdivia”, i.e., AI models foresee the hospital's medical needs for timely management of patients with this disease