June 7, 2022

U-M, AKU collaborators: Africa needs Artificial Intelligence for better health

Using colorectal cancer as a test case, research collaborators from U-M and the Aga Khan University (AKU) are proposing that AI and machine learning could be effective in addressing some of the continent’s health challenges. 

Is Africa ready for Artificial Intelligence?

Collaborators from the University of Michigan and the Aga Khan University (AKU) in Kenya believe so. In a new paper in Gut, they propose that AI and machine learning could be deployed to address the continent’s emerging colorectal cancer (CRC) problem.

“Recent technological advances and developments in AI have the potential to transform global health, particularly for early detection and diagnosis of CRC,” writes the research team led by Mansoor Saleh, MD, Chair of Haematology-Oncology at AKU Hospital, Nairobi, and Akbar Waljee, MD, UMMS Professor of Gastroenterology.

The paper is the first produced through a new data science-focused partnership between U-M and AKU. Appearing Gut’s April issue, it offers CRC screening as a test case for AI-driven solutions to some of the continent’s health issues.

“While data science applications are largely underdeveloped in Africa, many enabling factors are already in place,” the researchers note.

Dr. Monsoor Saleh, Chair of Haematology-Oncology at AKU Hospital and lead author on the paper, tours U-M delegates through his institution in Nairobi earlier this year.

The prevalence of CRC is growing in Africa. Limited colonoscopy screening capacity means most cases go undetected for too long; more than 60% of patients in sub-Saharan Africa present with stage-4 CRC. When detected this late, their five-year survival rate is less than 1%.

Artificial Intelligence and Machine Learning might play a role in ways that alleviate—rather than exacerbate—the screening scarcity. Computer algorithms could examine population-level data to determine which patients are at the highest risk and should be prioritized for screening. Once screened, other algorithms designed for pattern recognition can quickly scan the images to identify abnormalities that warrant a closer inspection by trained pathologists.

Such tools are already being tested and have shown promise in the US and other high-income countries. But advancements in cloud computing, decreasing costs, the penetration of mobile phones, and other factors have primed parts of Africa for AI health applications – assuming, the authors caution, that a host of other challenges can be overcome.

“If we do not directly address the challenges of dissemination and adoption of these prediction models in a way that supports social justice and health equity, data science approaches will have minimal impact on the health of individuals and populations,” the team writes.

“Given the growing investments in data science infrastructure, the demonstrated openness to embracing technological change (ie, mobile banking proliferation), and the urgent need to develop more efficient approaches to cancer screening … sub-Saharan Africa is well poised to drive innovative AI-based solutions,” the team writes.