In a Nigerian study now appearing in the Journal of Clinical and Translational Science, cancer education materials drafted by ChatGPT were set beside those written by local clinicians, and readers judged them largely equal in usefulness, understandability and clarity. It is a modest but meaningful step in a long human effort to put plain, trustworthy knowledge about cancer into the hands of people who have too little of it, particularly in regions where such material is scarce. The finding offers reassurance, yet it also quietly reminds us that a machine's fluency is not the same as local wisd
ChatGPT-Generated Cancer Patient Education Materials Match Clinician-Written Versions in Nigeria Study
Only 14% use AI for health information, yet 58% say they trust it.
Mark
So what's the headline here? AI wrote cancer education materials as well as doctors did?
Mimi
Roughly, yes. GPT-4o materials were comparable to clinician-written ones on usefulness, understandability and clarity, with no significant overall differences.
Luke
Worth saying that this comes from the lead author's own post on X. The numbers are the team's, and the post doesn't give us sample size or who the raters were.
Mark
Were there places where the AI fell short?
Mimi
Two. Clinician statements on healthy diet and limiting alcohol were rated clearer, and both differences were statistically significant.
Luke
And that's a useful caution. An average with no difference can hide specific messages where the AI does worse.
Mark
The usage numbers seem odd. Only 14% use AI for health information, but 58% trust it.
Mimi
That gap is striking. Doctors lead at 73%, then Google at 52%, with a physician influencer, Aproko Doctor, at 22%.
Luke
I'd be careful about reading too much into the trust figure. We don't know how the question was asked, or whether people trusting it have actually used it.
Mark
Why does the call for African evidence matter?
Mimi
Because 90% said they'd want cancer apps, and AI use is growing. If the tools lack culturally relevant content, misinformation could follow.
Luke
That's the authors' argument, not something this study measured. It's a reasonable concern, but still a forward-looking claim.
The Pulse
- Clinician-written lines on healthy diet and limiting alcohol outshone their AI counterparts in clarity, and the gap was statistically significant.
- Only 14% of participants had turned to AI for health information, yet 58% said they trusted it, a dangerous distance between use and belief.
- Doctors (73%) and Google (52%) remain the first stops for health questions, with the Nigerian physician influencer Aproko Doctor drawing 22%.
- Smartphones carry nearly every participant online at 95%, but only 4% use cancer-related apps, even though 90% say they would be interested.
- Researchers are pressing to weave African, evidence-based and culturally relevant content into AI outputs before misinformation takes root.
In a Nigerian study now appearing in the Journal of Clinical and Translational Science, cancer education materials drafted by ChatGPT were set beside those written by local clinicians, and readers judged them largely equal in usefulness, understandability and clarity. It is a modest but meaningful step in a long human effort to put plain, trustworthy knowledge about cancer into the hands of people who have too little of it, particularly in regions where such material is scarce. The finding offers reassurance, yet it also quietly reminds us that a machine's fluency is not the same as local wisdom, and that the question of whose knowledge shapes what people read is still very much open.
Runcie Chidebe, an interprofessional learning intern at Cleveland Clinic, shared on X the results of a study that tested a simple question: if ChatGPT writes cancer prevention materials for patients in Nigeria, do readers find them as useful and clear as those written by local clinicians? The answer, for the most part, was yes. The paper, a cross-sectional assessment co-written with colleagues including Krishna Prasad Sapkota and Fabio Ynoe de Moraes, appears in the Journal of Clinical and Translational Science, and its authors describe it as the first of its kind in Nigeria and across Africa.
The work grew from a gap that is easy to overlook. Digital tools have helped tailor cancer information to patients in underserved settings, and large language models can now produce answers that many find understandable. Yet research from low- and middle-income countries, and from Africa in particular, has been thin. The team hypothesized that the AI materials would match the clinicians' work on usefulness, understandability and clarity, and overall they did, with no significant differences. The exceptions were specific: statements on healthy diet and limiting alcohol were judged clearer when clinicians wrote them.
The survey also showed how people actually look for health answers. Doctors led at 73%, followed by Google, social media and the influencer Aproko Doctor. About one in seven had used AI tools, yet more than half said they trusted them. Most reached the internet by phone, and interest in cancer apps ran high even though few used them.
The authors conclude that as AI spreads across Africa, its outputs must carry African evidence and cultural context, so that misinformation does not take hold. Chidebe's post is a first-person account of his team's work, and the full paper holds the methods and sample details it leaves out.
Runcie Chidebe, an Education Administrative College Intern in Interprofessional Learning at Cleveland Clinic, took to X to share the results of a new study: when cancer prevention materials written by ChatGPT were put next to materials written by clinicians in Nigeria, readers rated them about the same.
The paper is titled "Comparing AI-ChatGPT and clinician-authored cancer patient education materials: A cross-sectional assessment," and it appears in the Journal of Clinical and Translational Science. Chidebe wrote it with Krishna Prasad Sapkota, Sampson C. Ipiankama, Deborah N. Ejemole, Leigh Leibel, Faridah M. Banwo and Fabio Ynoe de Moraes. By the authors' account, it is the first study to evaluate AI-generated cancer prevention education materials in Nigeria and across Africa.
The question grew out of a gap. Digital tools have helped tailor cancer information to patients' needs, particularly in underserved settings. Large language models such as ChatGPT can generate dynamic answers to patient questions, and earlier research suggests they can produce cancer education materials that are understandable and acceptable. But work from low- and middle-income countries, and from Africa especially, is scarce and only beginning to emerge. The team set out to compare the clarity, usefulness and understandability of ChatGPT-generated materials with those written by clinicians in Nigeria. Their hypothesis was that the AI content would match the clinicians' on all three measures.
Largely, it did. Materials produced with GPT-4o were comparable to the clinician-developed ones in usefulness, understandability and clarity, with no significant overall differences. There were exceptions at the level of individual statements. Clinician-written lines on "healthy diet" and "limit alcohol" were rated clearer than the AI's equivalents, and the differences were statistically significant for both.
The survey of participants also sketched how people in the study find health information. Doctors came first, at 73%. Google followed at 52%, other social media at 36%, and "Aproko Doctor," a Nigerian physician influencer, at 22%. AI tools such as ChatGPT were used by 14%. Yet 58% said they trust AI, a notable gap between how many use it and how many are willing to believe it.
The technology picture was mobile-first. Some 95% reached the internet by smartphone, and 49% used laptops. Thirty percent used fitness wearables and 36% used health-related mobile apps, but only 4% used cancer-related apps. Even so, 90% said they would be interested in using cancer apps for prevention and care.
The authors draw a cautionary conclusion from this. AI adoption across Africa is rising, they argue, which makes it urgent to build more African, evidence-based and culturally relevant information into AI outputs, so that misinformation does not spread among African users. Chidebe's account is a first-person summary of his own team's work rather than independent reporting, and the full paper is the place to look for the methods and sample details the post does not give.
Notable Quotes
AI adoption across Africa is increasing, underscoring the urgent need to incorporate more African evidence-based and culturally relevant information into AI outputs to prevent misinformation.— Runcie Chidebe, summarizing the study authors' argument on X
Want the full story?
Read the original at Oncodaily ↗
Technology