Data-driven methods for dengue prediction and surveillance using real-world and Big Data: A systematic review

Area of Science:

• Epidemiology
• Public Health
• Data Science

Background:

• Traditional dengue surveillance relies on case reporting, which has inherent delays limiting responsiveness.
• Machine learning (ML) and Big Data offer novel approaches to reduce reporting delays and predict outbreaks using non-clinical data.
• This systematic review focuses on data-driven studies employing real-world data and ML for dengue monitoring and prediction.

Purpose of the Study:

• To systematically review studies utilizing real-world data, Big Data, and/or machine learning methods for monitoring and predicting dengue-related outcomes.
• To identify trends in the application of these methods in dengue surveillance.
• To assess the performance of different data sources and ML models in dengue prediction.

Main Methods:

• Systematic literature search conducted in PubMed, Scopus, Web of Science, and grey literature (January 2000 – August 2020).
• Inclusion criteria focused on data-driven studies; reviews, RCTs, and descriptive studies were excluded.
• Analysis of 119 included studies, categorizing them by aim, data streams used, and machine learning approaches.

Main Results:

• A significant increase in relevant publications from 2016-2020 (67%), with 39% incorporating novel data streams.
• The primary aims were dengue outcome prediction (55%), data source validation (23%), or both (22%).
• Machine learning was employed in 60% of studies. Rainfall, temperature, and humidity were key predictors. Neural Networks and Decision Trees showed the highest performance (52%).

Conclusions:

• Integrating real-world and Big Data with machine learning presents a promising strategy for enhancing dengue prediction and surveillance.
• Further research is needed to optimize the integration of diverse data sources and methods for improved dengue management and response.
• The findings highlight the potential of data-driven approaches to overcome limitations of traditional surveillance systems.