Data Management and Compute Platform for Data Science Training - Supplementary Grant for DATICAN

Supplementary Grant for DATICAN - Summary The Data Science and Medical Image Analysis Training for Improved Health Care Delivery in Nigeria (DATICAN) project is a UE5 DS-I Africa project funded by NIH. The main aim of DATICAN is to build capacity in Data Science and medical image analysis. DATICAN’s overarching objective is to produce a new generation of data scientists

possessing the requisite data science skills in medical image analysis and the potential to become the clinical research leaders needed to improve healthcare delivery in sub-Saharan Africa (SSA). DATICAN is a collaborative effort amongst 4 universities, 3 in Nigeria and 1 in the USA, these are Lagos State University, The

University of Ibadan, Redeemer’s University and The University of Chicago. The project is funded for 3-years from September 2023 to September 2026. Project implementation started immediately after receiving the award letter in September 2023 and we have met our target for year 1 recruitment. We have recruited 36 participants, comprising PhD and MSc students,

postdocs, and Faculty members. We are currently training and as contained in our year 1 plan, the USA trainers have visited Nigeria and the Nigerian trainers have visited the USA this year. DATICAN was structured to be skills acqusition-driven, hands-on and project-driven. To achieve these, each trainee was given a project work related to improved healthcare delivery in SSA using data science and AI

knowledge. This implies that we have 36 use cases already. There are 3 major requirements for implementing these use cases, these are the relevant skills set, local data peculiar to the SSA and computational resources. We have started training the students and the progress recorded so far shows that DATICAN has the potential

to give the students the relevant skills required to execute their use cases. On local data, we have collected medical images of 2000 individuals with different modalities (e.g. ultrasound, MRI, CT, etc) and different body parts (brain, liver, abdomen, etc) and stored them on our hard drives, and there are many more to collect. The

unfortunate thing is that these images are unstructured, un-curated and some of them are even unlabelled. This implies that the data are not in a useful form. On computational resources, imaging data are huge and processing them requires access to compute infrastructure such as High-Performance Computing (HPC) with

Graphic Processing Units (GPUs). UChicago has given our trainees access to its HPC, the downside of this is the lack of sustainability and scalability. DATICAN has a lifespan of 3-years and lack of access to computational resources afterwards will make the acquired skills useless. The aims of this supplementary project are (1) to extend DATICAN to include data curation and proper data

management, this is to make data sharing amongst the three institutions possible. (2) to procure an HPC with a GPU that will meet the computational needs of our trainees. We strongly believe that both the computational resources and repository will be useful not only for the member institutions of DATICAN, but for all data

scientists and clinicians interested in using AI and data science technologies to improve healthcare delivery in Africa.