Friday Seminar in Africa Studies

Can multispectral imaging and AI be an approach for early detection of malaria infection?

Ida-Maria Sintorn and Carolina Wählby, Department of Information Technology, Uppsala University

In 2023 more than 250 million Malaria cases were reported with close to 600 000 deaths. Approximately 95% of all deaths happen in Africa and mostly in small children. The gold standard to diagnose the disease is visual inspection by microscopy of stained blood smears. Staining is tedious and requires a hospital lab and skilled technicians. Together with researchers in Lund and the Ivory coast we are investigating an alternative approach based on multi-spectral imaging on unstained samples followed by AI based analysis for simpler and cheaper early diagnosis of Malaria infection.

Improving Livestock Health and Welfare Through Collaborative AI 

Salman Toor, Department of Information Technology, Uppsala University

Livestock health and welfare in sustainable agriculture, food security, and consumer satisfaction cannot be overstated. Precision livestock farming (PLF) has shown promise in monitoring animal health through biosensors and machine learning. By analysing images and videos of livestock, computer vision can re-identify individuals based on their unique markers and recognize signs of injury and illness. Machine learning for livestock health and production is primarily studied in developed countries. African livestock is exposed to different stressors, resulting in rich and nuanced datasets. A potential solution is combining datasets from South Africa with those from Sweden. However, transferring data from Africa to Sweden raises issues of agency and self-determination. The project aims to design a prototype that allows SASUF (South Africa Sweden University Forum) experts to train models while maintaining autonomy over their datasets.