Recent studies show that AI can help detect tumours during routine breast X-rays and colonoscopies, where AI software assists in scanning images. (Graphic representation)
Technologies such as artificial intelligence can enhance the quality of healthcare, but a recent study has found that cultural practices may hinder the effectiveness of maternal health AI solutions.
The research examined data from the DawaMom maternal healthcare app for expectant mothers in Zambia, South Africa and Zimbabwe.
“For AI to truly improve healthcare in underserved communities, it must reflect local realities. That means incorporating diverse datasets and traditional practices,” said researcher Min’enhle Ncube.
South Africa’s maternal mortality rate has improved in recent years, with a reduction from 105 deaths per 100,000 live births in 2019 to 88 in 2020, according to the most recent statistics in 2022.
But problems persist in countries like Zambia, where a newborn dies every 30 minutes and a stillbirth occurs every hour in certain regions, because of under-resourced medical facilities and a shortage of skilled healthcare workers.
Maternal AI apps could help reduce these mortality rates by providing personalised healthcare guidance to expectant mothers. This includes home visits to gather critical data such as demographics, medical history and key health indicators.
AI then processes this information to identify patient risks and enable early interventions for high-risk conditions.
Despite the rapid response of AI-driven applications, many new mothers share the sentiments voiced in Ncube’s study, expressing concerns that cultural practices may be overlooked.
“It’s great that there’s an app to help us track nutrition and medication, but it provides Western solutions for women who follow traditional routes,” a woman who had just given birth to her second child told the Mail & Guardian.
“Organic and ancestral guidance in traditional African homes is important for the wellness of both mother and child, which you can’t get from an app.”
The study also compared digital care, traditional care and clinical care, and found that the only commonalities between these factors were “care at home, face-to-face care and biomedical health education”.
In 2014, South Africa’s health department launched its flagship programme, MomConnect, which gives free maternal health through SMSs or WhatsApps and access to a text-based helpdesk for ratings or queries.
Although MomConnect has had five million users since its launch, Mandisa* said she had difficulties using the messaging service because of data connectivity problems.
“This is South Africa — there are electricity issues; not everyone has a smartphone and sometimes there’s no data. How will we, as lower-income people, be able to make use of AI apps to save our babies?” she asked.
President Cyril Ramaphosa recently promised to use South Africa’s presidency of the G20 to “highlight the barriers to AI readiness and opportunities facing developing countries, including unequal and disproportionate access to digital infrastructure”.
Midwife Lilo Mthembu, of the Ten n Ten Birthing Centre, said that although AI holds promise, birth cannot wait for technological interventions.
“The beauty of birth is that it brings human-to-human contact. If AI can assist in keeping the mother and baby healthy before the birth, it will work to our benefit as midwives. But how do you ensure that every expectant mother, in both cities and rural areas, gets the same guidance?” she asked.
The study identified a critical gap in the absence of indigenous and traditional maternal health practices in AI datasets, which risks sidelining the cultural knowledge essential to many people in rural and underserved areas.
AI has already demonstrated its potential in detecting medical conditions and providing personalised solutions. The National Institutes of Health has highlighted the effectiveness of AI in diagnosing clinical conditions in medical imaging and controlling the outbreak of Covid-19 through early detection.
Recent studies also show that AI can help detect tumours during routine breast X-rays and colonoscopies, where AI software assists in scanning images.
Although AI may not be an immediate solution for South Africa’s healthcare inequality, the study suggests there is potential for development, particularly as the technology continues to evolve.