Data Science for Health in Africa Seminars
I have been attending a five-part seminar series leading up to the Data Science for Health in Africa Workshop that will be taking place at the Deep Learning Indaba 2025 conference in Kigali next month. Hosted by SisonkeBiotik, DS-I Africa, and Ro’ya, this series features trailblazers whose work sits at the intersection of artificial intelligence, health and innovation in Africa.
Up to now, I have attended the first two seminars. The first seminar titled “Accelerating Medical Insights with RAG: From Patient Data to Precision Care” was presented by Elisha Bere. Bere explored how optimising Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG) systems can revolutionise healthcare. LLMs can expedite the retrieval of academic studies and diagnostic information, enabling doctors to spend less time searching for information and more time helping patients. Optimising LLMs with RAG systems helps minimise hallucinations by introducing a database from which the model generates more accurate and relevant responses.
The second seminar titled “Bridging Neuroscience and Artificial Intelligence: Deep Learning Models for Understanding Vision” was presented by Mai Gamal. Gamal illustrated how deep neural networks (DNN) can be used to increase our understanding of visual perception. In particular, she showed how these networks can model activity in the Lateral Geniculate Nucleus, a subcortical structure in the brain’s visual pathway, for which limited research exists regarding what this nucleus learns. This was followed by a demonstration of a large-scale study comparing how different DNN architectures model the brain’s response to video stimuli, using fMRI recordings (i.e., recordings of brain activity) as a reference. Finally, she introduced a neural encoding scheme that enables the understanding of both inter- and intra-regional connectivity across visual areas of the brain.
While I am relatively new to the application of data science in healthcare, I can confidently say that I have enjoyed attending these seminars and have learned a great deal. These models demonstrate remarkable capabilities; however, they are not without limitations. It has been interesting to see how both Bere and Gamal navigate the challenges that these models present. I look forward to the remaining seminars in the series!