Predictive Analytics and Machine Learning for Disease Outbreak Detection: A Systematic Review of Applications and Implications for Nigeria's Public Health Sector

Abstract

Infectious disease outbreaks remain a leading driver of morbidity and mortality in Nigeria, where cholera, Lassa fever, meningitis, and vaccine-preventable diseases recur against a surveillance backdrop of fragmented data systems and infrastructural constraints. Machine learning (ML) has been proposed as a means of shortening the interval between disease emergence and detection, but the evidence base specific to Nigeria has not previously been assembled using a systematic method. A systematic review was conducted following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 principles Structured searches were run in the bibliographic databases PubMed, Scopus, Web of Science, and the IEEE Xplore Digital Library, supplemented by Google Scholar and the Nigeria Centre for Disease Control and Prevention research repository, with the last search conducted on August 15, 2026. Searches used combinations of the terms "machine learning," "artificial intelligence," "disease surveillance," "outbreak detection," "predictive analytics," and "Nigeria," restricted to sources published between 2015 and 2026. Records were screened against pre-specified eligibility criteria, and data were extracted on study setting, disease focus, ML method, and reported outcomes. Because of heterogeneity in study design and outcome reporting, findings were synthesised narratively rather than pooled statistically. Thirty-nine records were identified, of which 16 met eligibility criteria and were included in the final synthesis (PRISMA 2020 flow diagram, Figure 1). Included studies addressed malaria, cholera, Lassa fever, COVID-19, tuberculosis, and hepatitis B, alongside cross-cutting methodological and ethics literature. Supervised learning methods, particularly decision trees, random forests, and support vector machines, were the most frequently applied and generally achieved the strongest reported predictive performance in resource-limited settings. The included evidence indicates that interpretable, computationally lightweight ML models represent the most realistic near-term pathway for strengthening outbreak detection in Nigeria. Progress will depend on interoperable digital infrastructure, Nigerian-specific training data, workforce development, and alignment with the World Health Organization's ethical principles for artificial intelligence in health.