When her grandfather lost his fight against prostate cancer, Dr Mpho Mokoatle’s world was shaken. Raised by him in Orkney, a small town in North West, she not only lost the man who had been her father figure but found herself asking a question that would later define her life’s work: why did he not respond to treatment?
That question became the seed of a PhD in computer science from the University of Pretoria, research that has now positioned her at the cutting edge of global efforts to detect cancer earlier, more accurately, and potentially at lower cost.
Her grandfather was diagnosed with prostate cancer in 2016 and treated at a public hospital. He was placed on Zoladex, a hormonal therapy designed to cut off the supply of testosterone that fuels prostate tumours. But instead of improving, his condition worsened, spreading to other organs. By 2019 he had passed away.
“It was devastating. He raised me, he was the only father figure I knew,” said Mokoatle. “Other patients go into remission or live better lives after treatment. I wanted to understand what was so different about him.”
Her grief turned into curiosity, then into purpose.
For her PhD, Mokoatle developed a model that uses natural language processing (NLP), a branch of artificial intelligence (A) that powers tools such as chatbots and translation apps, to treat DNA as a language.
“DNA has its own alphabet,” she said. “Just like English or isiZulu, it has sequences and patterns. If we can train machines to understand those patterns, we can detect mutations or abnormalities that lead to diseases such as cancer.”
Her model did not just work, it outperformed DNABERT-6, a state-of-the-art genomics model developed on a far larger scale with far more computing power. “I was shocked. DNABERT-6 was trained on massive data, but my model, with fewer resources, performed better in certain DNA tasks,” she said.
The implications are far-reaching. By spotting genetic signals before symptoms appear, AI could allow for much earlier diagnoses. “Early detection saves lives. It means patients have better chances of survival and treatment becomes cheaper,” said Mokoatle.
Right now, cancer detection in South Africa is a lengthy and invasive process. For prostate cancer, men undergo rectal exams, PSA blood tests, MRIs, and biopsies before doctors can confirm a diagnosis and decide on treatment.
“Imagine, instead of going through all of that, you could run a simple DNA sequencing test, feed the data into an algorithm, and get an accurate picture of what’s happening in the body,” said Mokoatle. “That’s the future I’m working towards.”
Though her grandfather’s illness sparked her focus on prostate cancer, Mokoatle’s model has potential far beyond oncology. It can be applied to a wide range of genomic problems, from identifying gene regulation patterns to improving precision medicine by tailoring treatments to a patient’s unique genetic profile.
I want students from small towns to know that your background doesn’t define you. Public schools can still equip you to succeed. I am proof of that
— Dr Mpho Mokoatle
Still, she admits there is a long way to go. Hospitals would need DNA sequencing facilities, budgets for equipment and teams of doctors, bioinformaticians and AI experts working together. “It requires a multidisciplinary approach. But it is possible,” she said.
Mokoatle’s journey makes her achievements even more remarkable. She grew up in a household where money was tight, attending public schools with limited resources. She only touched a computer for the first time when she arrived at university.
“I want students from small towns to know that your background doesn’t define you,” she said. “Public schools can still equip you to succeed. I am proof of that.”
Her studies were made possible by support from the South African Medical Research Council, the CSIR, IBM and Google. “That funding changed my life. I didn’t have to worry about food or equipment. I could just focus on my work. I even sent some money home to help my grandmother, who is a pensioner,” she said.
Her achievement is also historic on a personal level; she is the first PhD holder in her entire clan. “I stand on the shoulders of my ancestors. This degree is not just mine, it belongs to my family and community. I know my grandfather would have been proud,” she said.
She hopes her story inspires more African women to pursue careers in fields such as AI and health sciences. “We have limited resources in Africa, but that shouldn’t stop us from innovating. My work shows it can be done.”
Mokoatle is determined to take her research further, expanding her model to cover entire DNA sequences and not just parts. That will require more powerful computing, significant resources and collaboration with medical specialists.
She believes AI will eventually transform health care across Africa. “Automation will reduce human error, speed up diagnosis, and free up overburdened health-care workers. Imagine shorter queues, faster test results, and more precise treatments. That’s the future AI can bring.”
For now, she is focused on building partnerships that can take her PhD from theory to practice. “This is just the beginning,” she said. “My dream is for my research to save lives.”
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.