This community innovation has been accepted at the 2026 DHIS2 Annual Conference as a physical poster.
AI fusion of DHIS2 and climate data for malaria GM
Malaria transmission in The Gambia is highly seasonal and strongly influenced by climatic variability; however, routinely collected daily surveillance data are underutilized for predictive analysis and decision-making. This study explores the integration of DHIS2 malaria surveillance data with key climatic indicators, employing AI-driven modelling approaches to predict malaria outbreaks and identify high-risk zones. The approach aims to enhance early warning systems and support timely, evidence-based public health interventions. Using longitudinal secondary surveillance data, individual- and regional-level variables extracted from DHIS2, including age, sex, reporting date, test results, symptoms, travel history, and bed-net usage, were analysed. Climate variables were incorporated to capture environmental influences on malaria transmission dynamics. Datasets were cleaned, harmonized, aggregated monthly, and analysed using Python within a Jupyter Notebook environment to facilitate predictive modelling and risk assessment. Classification techniques, including Decision Tree and Random Forest, were applied to predict malaria positivity, while K-Means and hierarchical clustering categorized regions into high-, medium-, and low-risk zones. The analysis revealed distinct seasonal and spatial patterns, with specific regions consistently clustering as higher risk during peak transmission periods. Classification models achieved robust predictive performance, demonstrating the feasibility of using routine DHIS2 data for early warning when combined with climatic variables. Interactive dashboards developed in Power BI enhanced interpretation and usability for program managers. The study underscores considerations of data quality, model interpretability, and the responsible application of machine learning in public health surveillance. Overall, the findings demonstrate the potential of DHIS2-based analytics to strengthen malaria surveillance, inform climate-sensitive decision-making, and improve outbreak preparedness in The Gambia and similar contexts.
Primary Author: Ousman Jallow
Keywords:
Topic: AI-Driven Fusion of DHIS2 Surveillance and Climate Data for Predictive Modelling of Malaria Outbreaks in The Gambia, Keywords: Malaria Surveillance, DHIS2, Predictive Modelling, Machine Learning, Climate Data
