Unraveling global malaria incidence and mortality using machine learning and artificial intelligence-driven spatial...

Area of Science:

• Global Health Epidemiology
• Computational Biology
• Spatial Analysis

Background:

• Malaria remains a critical global health issue, causing significant morbidity and mortality.
• Effective control strategies require precise identification of high-risk areas and contributing factors.

Purpose of the Study:

• To identify malaria high-risk areas globally using spatial analysis, machine learning (ML), and explainable AI (XAI).
• To uncover key determinants of malaria incidence and mortality and establish causal relationships.
• To predict disease outcomes and inform targeted public health interventions.

Main Methods:

• Analyzed data from 106 countries (2000-2022) from WHO, World Bank, and UNICEF.
• Employed XGBoost (ML classifier) with XAI and causal AI (CAI) for incidence and mortality evaluation.
• Utilized spatial autocorrelation (Getis-Ord Gi*, Moran's I) to detect geographical malaria clusters and hotspots.

Main Results:

• Identified high-incidence clusters in Benin, Burkina Faso, and Ghana; mortality clusters in Benin, Central African Republic, and Liberia.
• XGBoost model showed high predictive accuracy for malaria incidence and mortality (r² = 0.93).
• Key determinants included access to basic sanitation, electricity, population growth, and under-5 mortality rate.

Conclusions:

• An integrated AI-driven spatial framework effectively identifies malaria determinants and hotspots.
• Advocates for incorporating AI-powered spatial models into national malaria surveillance systems.
• Supports evidence-based, targeted interventions for global malaria burden reduction.