A scoping review of artificial intelligence applications for mpox preparedness, prediction, prevention, and surveillance in the era of emerging epidemics

The 2022–2024 global resurgence of mpox (formerly monkeypox) and the successive declarations of a Public Health Emergency of International Concern have exposed persistent weaknesses in conventional disease preparedness, prediction, prevention, and surveillance. Artificial intelligence (AI) and machine learning (ML) offer scalable tools to strengthen each of these pillars, yet the evidence remains fragmented across diagnostic imaging, epidemiological forecasting, genomic analytics, digital infodemiology, and therapeutic discovery. To systematically map the scope, methods, performance, and translational maturity of AI applications across the mpox epidemic continuum, and to identify research gaps that should guide future preparedness for emerging epidemics. We conducted a scoping review following the PRISMA Extension for Scoping Reviews (PRISMA-ScR) framework. PubMed, Scopus, Web of Science, IEEE Xplore, and Embase were searched from database inception to the final search date of 30 September 2025. Eligible records were concentrated between 2018 and 2025. After de-duplication and two-stage screening performed by a single reviewer, 48 studies meeting predefined eligibility criteria were charted and synthesized under four thematic domains: diagnosis and detection, prediction and forecasting, surveillance and digital epidemiology, and drug and vaccine discovery. Deep-learning image classifiers dominated the diagnostic literature, with reported accuracies and area-under-the-curve values ranging from approximately 0.83 to 0.99 for distinguishing mpox skin lesions from clinical mimics. Time-series and neural-network models (ARIMA, LSTM, GRU, and ensemble sub-epidemic frameworks) produced credible short-term case forecasts, while natural-language-processing pipelines tracked public sentiment and misinformation in near real time. AI-assisted genomic surveillance, wastewater monitoring, and structure-based drug and epitope discovery represented rapidly emerging but less mature domains. Reported performance metrics were author-reported values obtained under heterogeneous datasets and validation protocols, and were not pooled. Across studies, small and imbalanced datasets, limited external validation, and weak clinical integration were recurrently reported by the included primary studies. AI shows promise in augmenting mpox preparedness, prediction, prevention, and surveillance, but most tools remain at a proof-of-concept stage. Prospective validation, equitable data governance, explainability, and integration into public-health workflows are prerequisites for real-world impact in future epidemics.