Utilizing Artificial Intelligence as a Dengue Surveillance and Prediction Tool
Objectives: A major challenge in passive surveillance is that an outbreak has often occurred before it is recognized. The main purpose of this study is to determine how well an artificial intelligence mediated system can improve the quality of Malaysia’s dengue surveillance system. In particular, the focus of the study was to evaluate the effectiveness of a real-time surveillance system in dengue case detection and prediction of future outbreaks.
Methods: A feasibility study was conducted in the state of Penang by incorporating artificial intelligence and machine learning capabilities to geo-locate and determine future dengue outbreaks. This decision-making tool supports data entry, retrieval, storage and analysis for dengue vector management and promotes the execution of dengue control programs that are designed, evaluated and refined based on locally gathered evidence.
Results: The system predicted 37 outbreaks up to 30 days in advance, geo-locating them up to 400 metres radius. This prediction was then cross-validated with the Penang State Health Department dengue reports in which 30 outbreaks occurred within the predicted period. The prediction accuracy of this console was 81.08%.
Conclusion: The Bayesian network system has the potential to report & predict the next dengue outbreaks in real-time. It incorporates user-friendly functionalities for data entry or input, data storage, data query, case and disease outbreak mapping, reporting, advance outbreak predictions and even suggested vector control management. This network system is anticipated to improve the current dengue surveillance, intervention monitoring and evaluation of the overall dengue vector control program performance.