Research & Publications
tech
Machine Learning for Malaria Early Warning: Feasibility in Resource-Constrained Labs
Akol Bol Akol 2 views
Abstract
We evaluate gradient-boosted models on routine surveillance data to predict county-level malaria surges four weeks ahead.
Methodology
Mixed methods: structured surveys, record audits and key-informant interviews across representative sites.
Findings
Mean precision of 0.81 at four-week horizon using only existing DHIS2 features.
References
Ministry of Health SS (2024); WHO Digital Health Atlas (2025); SSHIA Annual Report (2025).