Soutenance publique de thèse de doctorat en Sciences géographiques - Camille Morlighem

Multi-source data and methods for malaria risk mapping in Senegal: building on the Hazard-Vulnerability framework

Jury

• Prof. Sabine HENRY (UNamur), Présidente
• Prof. Catherine LINARD (UNamur), Secrétaire
• Dr Ibrahima DIA (Institut Pasteur de Dakar)
• Dr Annelise TRAN (CIRAD, Montpellier)
• Prof. Christel FAES (Université de Hasselt)

Abstract

 

The distribution of malaria in Sub-Saharan Africa is becoming increasingly heterogeneous, with emerging hotspots and a growing urban burden rather than following traditional vector suitability gradients. This pattern has been observed in Senegal, where malaria transmission ranges from very low to high. In this context, disease risk maps can help to identify hotspots and improve the targeting of interventions. However, previous studies have often relied on freely available, low-resolution remote sensing data and focused primarily on environmental factors related to vector presence—the hazard—while overlooking population vulnerability (e.g. influenced by access to healthcare). This thesis presents an open-source malaria risk mapping framework incorporating high-resolution, open-access data on both hazard and vulnerability, and identifies the key factors sustaining transmission in Senegal.

Interpolated surfaces of vulnerability indicators were integrated with Sentinel-based hazard variables to model survey-based malaria prevalence, used as an indicator of risk. The framework was then extended to model malaria incidence from routine health facility data, using dasymetric disaggregation to create fine-scale incidence maps. The findings reveal that vulnerability is a key determinant of malaria and that both hazard and vulnerability risk factors vary with urbanisation level, transmission intensity, seasonality and spatial scale. Discrepancies emerged between malaria prevalence and incidence, such as low incidence but high prevalence in remote areas, suggesting potential underdiagnosis. This thesis also offers a comparative overview of various modelling approaches, ranging from machine learning to Bayesian geostatistics, with implementation code and guidance for future malaria research. Anticipated improvements in malaria epidemiological data will further enable elimination strategies to leverage the full potential of geospatial methods for fine-scale risk mapping.

 

Télecharger : vCal