The healthcare landscape in Africa is on the brink of transformation, driven by advancements in data analytics, artificial intelligence, and computer vision. Despite these strides, unique challenges hinder the full integration of these innovations in the African healthcare sector. Recognizing the urgent need to bridge this gap and empower stakeholders in Africa's healthcare ecosystem, this year's DS4Health Africa workshop takes an interdisciplinary approach built around current deep learning methods applied to healthcare and computer vision. It will be a joint workout between Ro'ya and Sisonke Biotik aiming to bring together the machine learning for healthcare and computer vision communities. This interdisciplinary workshop will serve as a platform for experts and communities engaged in healthcare-related research to interact with computer vision researchers and explore opportunities for collaboration. It will include work on both topics and their intersection with panel discussions, interactive sessions, and tutorials on medical imaging and image segmentation. By fostering knowledge exchange and collaboration, the workshop endeavors to deepen participants' understanding of the capabilities and constraints of deep learning in healthcare, paving the way for collaborative initiatives and future research endeavors with the potential to drive meaningful impact in African healthcare delivery.

Medical Technologies for Equitable AI

Dr. Mercy Asiedu

Applications of artificial intelligence (AI) in healthcare have accelerated exponentially. While AI has the potential to improve healthcare for the better, like other advancements in medical technology, it also has the potential to increase health disparities. Historically, medical technology has not been created with the needs of underserved settings in mind and this has led to limited access in low/middle income countries. Reduced access to these healthcare tools leads to downstream gaps in data used to train machine learning models, and by extension, the potential for machine learning biases, should these models be used on individuals from historically excluded settings.

In this talk I will discuss research projects on developing AI models to improve health equity outcomes, particularly in African countries. This work ranges from development of AI-enabled cervical cancer screening tools, machine learning models for maternal and fetal health, development of African-focussed datasets for LLM evaluation and tuning, and our research on understanding African perspectives on AI for health application including algorithmic fairness.

Overall, I hope to make a case for considering geo-contextual AI development and evaluation for health equity and to demonstrate methods I have employed to ensure inclusive data representation for the future of healthcare.

New insight on Machine Learning applied to Neuroimaging

Dr. Alessandro Crimi

Machine learning is revolutionizing the way we tackle complex neurological diseases like glioma and Alzheimer's. These conditions are challenging to diagnose and treat due to their intricate nature, which is where machine learning (ML) comes in. By harnessing the power of ML, we can dive deep into neuroimaging data—one of the most complex and rich sources of medical information. For glioma, ML not only helps in detecting tumors earlier but also in classifying them with remarkable precision, guiding personalized treatment strategies. In Alzheimer's, ML is crucial in identifying subtle changes in brain scans long before symptoms appear, offering a chance for early intervention. The blend of advanced neuroimaging and machine learning is paving the way for smarter, more accurate healthcare, bringing us closer to understanding and combating these devastating diseases.

Weather forecasting in Africa

MohammedElFatih MohamedKair

Natural disasters often leave widespread devastation in their wake, making timely damage assessment critical for effective response and recovery efforts. This presentation explores SKAI, a Google-developed framework that leverages AI and machine learning to accelerate and enhance the damage assessment process. By combining computer vision models and large vision language models, SKAI aims to improve both the speed and coverage of post-disaster evaluations. This presentation will delve into the technical aspects of SKAI, illustrating how it can be a valuable tool for decision-makers in the crucial aftermath of disaster. The potential implications of such a tool for disaster response and recovery will be discussed, aiming to engage a mixed audience of technical experts and those interested in the broader applications of AI for social good.

Computer Vision for Medical Imaging Tutorial

This tutorial aims to offer a well-curated background on the basics of medical imaging, particularly highlighting their fundamental differences from natural images. This will aid the audience, especially new researchers, in understanding the challenges of applying out-of-the-box computer vision approaches in medical imaging and in generating new research ideas. Furthermore, the section incorporates case studies, allowing participants to bridge theoretical concepts with real-world examples, thereby enhancing the learning experience.

Image Segmentation Tutorial

We will delve deeply into one case study, focusing on medical image segmentation, which is arguably the most common task in medical image analysis. Our medical image segmentation tutorial provides a comprehensive overview of cutting-edge techniques for delineating anatomical structures and pathological regions from medical images. Participants will gain hands-on experience with popular Python libraries such as NumPy, OpenCV, scikit-image, and TensorFlow/Keras for image loading, preprocessing, and deep learning-based segmentation. The tutorial covers fundamental segmentation methodologies, including thresholding, region-based methods, and advanced deep learning approaches like U-Net and DeepLab. Participants will learn how to evaluate segmentation performance using quantitative metrics like Dice similarity coefficient (DSC) and Hausdorff distance, and they will explore real-world applications in neuroimaging, oncology, and radiology. By the end of the tutorial, participants will be equipped with the knowledge and practical skills needed to tackle medical image segmentation tasks and contribute to advancements in healthcare and medical imaging research.