High-resolution mapping of onchocerciasis risk in Ghana using spatial machine learning models

Area of Science:

• Epidemiology
• Geospatial analysis
• Public health

Background:

• Onchocerciasis, caused by Onchocerca volvulus, is endemic in Ghana and transmitted by blackflies.
• Ivermectin (IVM) mass drug administration (MDA) has not eliminated transmission, especially in complex regions.
• Nationwide fine-scale risk maps are lacking for Ghana.

Purpose of the Study:

• To develop nationwide fine-scale geospatial risk maps for onchocerciasis in Ghana.
• To identify key environmental and climatic factors associated with onchocerciasis transmission risk.

Main Methods:

• Extracted and geocoded community infection data.
• Generated raster datasets for environmental, climatic, and healthcare accessibility variables.
• Employed logistic regression, random forest, and gradient boosting machine models for prediction.

Main Results:

• The random forest model achieved the highest balanced accuracy (70.30%) and AUC (76.40%).
• High-risk areas were identified in west-central, Volta Lake, southwest, and north-central Ghana.
• River density increased transmission risk, while high land surface temperatures and dense vegetation decreased it.

Conclusions:

• Fine-scale risk maps are crucial for adaptive onchocerciasis control strategies.
• Targeted vector control, improved healthcare access, and social interventions are necessary.
• These tools support the World Health Organization's 2030 onchocerciasis elimination targets.