Detecting Dengue Infection’s Initial Stages Using Neural Networks

Vega-Huerta H.; Vilca-Velasques J.; Vilca B.Z.; Maquen-Niño G.L.E.; Pacheco O.B.; Puelles-Bulnes M. · Smart Innovation, Systems and Technologies (2025) · DOI: 10.1007/978-3-031-92651-8_47

Resumen

The Aedes aegypti mosquito is the vector responsible for transmitting the dengue infection, which has a significant impact on the health public in Peru. The objective of this study is to detect the presence of the infection in the initial stages using the most common symptoms presented by patients as input. Previous studies related to dengue diagnosis were reviewed, which showed that the models based on Artificial Neural Network (ANN) achieved better results in terms of accuracy, sensitivity, and specificity. Therefore, the ANN model was chosen as the methodology of this study proposal. First, data was obtained from the Figshare platform. Then, the neural network architecture was designed, followed by data preprocessing, model compilation and training, as well as performance evaluation. Finally, forecasts were made using the trained model for new patients under suspicion. The presented neural network model achieved an accuracy of 95.70% in the last training epoch. Additionally, high values of sensitivity (100%) and specificity (98.50%) were obtained. These results indicate that the implementation of the neural network was effective for diagnosis dengue infection in its initial stages. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Instituciones: Universidad Nacional Mayor de San Marcos; Universidad Nacional Pedro Ruiz Gallo