Application of artificial intelligence for epidemic outbreak prediction in low and middle income countries: A Scoping Review

EXECUTIVE SUMMARY: Introduction: Low- and middle-income countries face multiple challenges in surveillance and timely response to epidemic outbreaks. Diseases such as dengue, malaria, Zika, and COVID-19 have highlighted the urgent need for more effective predictive tools. In this context, artificial intelligence (AI), including techniques such as machine learning and deep learning, has shown high potential to anticipate outbreaks by analyzing epidemiological, climatic, and behavioral data. This scoping review aims to map and characterize existing AI applications for epidemic outbreak prediction in low- and middle-income countries, identifying the techniques used, the diseases addressed, the types of data employed, and the barriers and facilitators for their implementation. Methodology: A scoping review will be conducted of observational studies, AI-based predictive modeling articles, methodological studies, systematic or scoping reviews, and grey literature published between 2010 and 2025, in English or Spanish, reporting the use of AI techniques for epidemic outbreak prediction in low- and middle-income countries. Searches will be performed in scientific databases such as PubMed, Scopus, Web of Science, SciELO, and arXiv using MeSH terms related to AI, machine learning, prediction, infectious diseases, and developing countries. Variables extracted will include study country, disease addressed, AI technique used, data sources, performance metrics (e.g., accuracy, sensitivity, specificity), and reported barriers and facilitators. Expected Results: It is expected to identify and categorize the most commonly used AI models for predicting epidemic outbreaks in low- and middle-income countries, the priority diseases addressed, the most used data sources, and the reported performance metrics (accuracy, sensitivity, specificity, among others). Additionally, the review aims to describe factors that have facilitated or limited the implementation of these approaches in low-resource settings. The findings will consolidate existing evidence and guide the design of future research, surveillance strategies, and predictive tools adapted to resource-limited environments.