Predicting malaria epidemics in Burkina Faso with machine learning
Problem
Malaria outbreaks in Burkina Faso previously required complicated numerical transmission simulations, preventing simple data-driven early warning.
Why it matters
Accurate case-rate forecasts would let decision makers intervene months before an outbreak, potentially saving lives.
What the AI did
Gaussian Processes and Random Forest Regressors trained on high-fidelity IeDA infant consultation data estimate weekly malaria cases over a 13-week horizon, tested on historical epidemics.
Results
At lowest alert threshold: 30% precision with >99% recall; rises to >99% precision and 5% rec… (abstract truncated at final figure)
Impact
First data-driven early warning system validated on historical epidemics at one facility; operational deployment not stated in source.
Record
| Venue | PLoS ONE |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | David Harvey, Wessel Valkenburg, Amara Amara |
| Citations | 34 (OpenAlex, at import time) |
| References | 26 |
| Model or method | Gaussian Processes combined with Random Forest Regressors |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1371/journal.pone.0253302 |
| Open access | gold |
| Keywords | malaria, epidemic forecasting, machine learning, early warning, Burkina Faso |
Importance rating
significant — thousands to millions affected
Malaria epidemic early warning could let decision makers intervene months ahead, potentially saving lives; first data-driven system, but validated at one facility with no stated scale.