ABSTRACT
Nigeria remains the global epicentre of malaria, contributing 27% of cases, with challenges in surveillance, data reliability, and intervention efficacy. This study integrates AI-driven modelling and spatial-statistical analysis to evaluate Plasmodium falciparum (Pf) malaria prevalence, mortality, and incidence trends in Nigeria (2010–2022) and proposes strategies for climate-resilient control. Leveraging Malaria Atlas Project data, we reveal disparities: Nigeria's 2022 Pf mortality (96.35 deaths/100,000), incidence (294.45 cases/1,000), and prevalence (26.30% in children aged 2–10 years) were 4.4×, 10.5×, and 8.8× higher than global comparators; these are relative magnitudes. Temporal analysis shows an annual prevalence decline of −1.1% versus −2.3% globally, with COVID-19 disruptions (2020 mortality rose to 106.83). Spatial autocorrelation at the state level using 2022 PfPR₂–₁₀ under a queen contiguity matrix yielded Moran's I=0.35 (p<0.01), with high-high clusters in Sokoto and Borno, contrasting with urban Lagos. To address low-confidence classification, we developed an AI-coupled SEI-SEIR-LSTM framework. The dynamic model achieved calibration RMSE 8.2; LSTM test RMSE 2.34, R²=0.91, outperforming ARIMA (12.5) on the same target. The model projects 2023 incidence of 298.2 cases/1,000 (95% CI: 294.6–301.8); a projection, not observed. Infection prevalence (23% mean absolute SHAP contribution) was a key predictor. We recommend AI-enhanced surveillance integrating climate data, bed-net distribution, larviciding in flood-prone regions, mobile clinics in conflict zones, and alignment with Nigeria's National Malaria Strategic Plan (2021–2025). This provides a blueprint for reducing Nigeria's burden and advancing global malaria elimination targets.
Keywords: Malaria burden, AI-driven modelling, Spatial clustering, Predictive analytics, Climate resilience, Nigeria
