With the help of HSRC-run workshops, researchers in South Africa and Portugal asked parents and healthcare professionals what a child health chatbot should and should not do. Their findings suggest that such tools could give caregivers timely, practical guidance and help them recognise when professional care is needed, but only if they are trustworthy, inclusive and carefully adapted to local realities.
Photo: Magnific
Findings from Stats SA showed that influenza and pneumonia remained the leading underlying natural cause of death among children aged 1–14 in South Africa. Additionally, many child deaths in South Africa between 2017 and 2020 were caused by ‘modifiable factors’ such as delays in seeking healthcare.
To support parents and mitigate preventable deaths, the Parent-Coach project aimed to leverage technological advancements in message-based systems (chatbots) to help first-time parents care for their children’s health and well-being.
The HSRC, working with the University of Cape Town (UCT) and Fraunhofer Research Center for Assistive Information and Communication Solutions (AICOS), ran workshops to inform the chatbot’s design. An earlier set of workshops in South Africa and Portugal looked at how parents assessed the child health advice they received. Those results were published as a journal article in the March edition of the HSRC Review.
In this latest article, researchers outlined the results of a further round of workshops exploring what healthcare professionals and recent parents thought about using chatbots to support child health. The workshops also asked how both groups wanted a chatbot to look and behave – its concept, interaction style, content design, personality and tone of voice. Participants of these workshops included 17 healthcare professionals, including paediatricians and therapists, and 36 first-time or recent parents from South Africa and Portugal.
What healthcare professionals wanted from chatbots
‘We found that healthcare professionals viewed chatbots as valuable tools to support parental decision-making, helping determine when medical attention is necessary, enhance health literacy, and complement – rather than replace – healthcare professionals,’ the researchers wrote.
Healthcare professionals welcomed the idea, believing chatbots could improve parents’ decision-making skills and health literacy. “In the emergency room, if there was health literacy, [most cases] would probably not even need medical assessment,” one paediatrician from Portugal said.
A South African nurse agreed: “[The chatbot] will give them information while they are just sitting at home – there won’t be a reason to come here. I am not saying they must not come, but only when there is a necessity. But people must get to be independent and be parents who are informed, fully informed.”
However, healthcare professionals did not want chatbots to provide medical diagnoses, prescribe medication, manage emergencies, or replace clinical judgement. “I don’t think it should do anything where a child needs to be seen by a doctor, or anything that needs a subjective opinion. If you’re unable to give an answer that is … yes or no, then you probably need a human eye to see it and to make that decision,” a paediatrician from South Africa said.
Finally, healthcare professionals wanted chatbots to offer evidence-based content that parents would find credible. That meant citing sources and having reputable institutions behind the tool. “Parents will trust it if we clinicians endorse it because people trust something that is sold to them by a well-known health department,” said a nurse from South Africa.
What parents wanted from chatbots
Parents agreed with healthcare professionals that chatbots should not replace professionals. ‘Parents imagined the chatbot as a first line of support, which could offer practical advice and help them assess the urgency of a situation they dealt with,’ the researchers wrote. ‘They saw the chatbot as a tool that could facilitate communication and provide clear, practical guidance whenever needed.’
Parents wanted clear steps and several options they could act on immediately, rather than long, theoretical explanations. A father in Portugal gave an example of how a chatbot might offer advice: ‘Hello! If you have already consulted your doctor and everything is fine with your child, try giving him [X]. If it doesn’t work, go back to your doctor.’ Parents wanted the chatbot to offer alternative approaches and explain what to do if the first suggestions did not work.
They preferred simple, everyday language that anyone could understand, whatever their education or health literacy. “It shouldn’t be complicated because everyone needs to understand it,” a South African father said.
At the same time, parents did not want information oversimplified. Some wanted useful medical terms included and clearly explained, to help them understand their child’s condition and communicate with healthcare professionals. Parents also wanted flexibility in how they found information, with most preferring the option to type their own questions alongside menus they could browse. “Some people can’t write, so it would be best if they just scroll down and select what they want to select,” a mother from South Africa said.
They also wanted the chatbot to communicate in an empathetic and inclusive way. Many supported a gender-neutral voice so that fathers and other caregivers would not feel excluded. “People think that it’s only women who take care of children, but there are fathers who can take care of the newborn,” a mother from Cape Town said. Others preferred a female voice, which they associated with greater warmth and reassurance.
Portugal versus South Africa
Participants in both countries agreed on most core principles, but the South African context raised additional design needs. These included difficulty in accessing healthcare professionals because of transport costs and long distances; fewer healthcare professionals and medical resources; mobile-data and connectivity limits; the need to communicate across several languages; and larger, more varied households.
For the South African setting, participants suggested that medical terms, images and videos could help overcome language and education barriers. Larger households mean the chatbot might also need to support grandparents and other caregivers. It could be built to work with low-data usage or offline, with careful local adaptation.
Three difficult design choices
From these findings, the authors identified three main design tensions that would need careful thought.
First, designers must decide whom the chatbot speaks to, honour preferences such as a woman’s voice, and not reinforce the idea that child health is mainly a mother’s responsibility but make sure fathers and other caregivers feel included.
Second, they must balance freedom with guidance. Parents want quick access to information they ask for, but menus and follow-up questions might steer them towards safer, more relevant advice.
Finally, the chatbot must balance plain and medical language, using everyday wording for accessibility while clearly explaining useful medical terms that can strengthen health literacy and help parents communicate with healthcare professionals.
What this means for South Africa
The work could improve child health in South Africa by guiding the design and development of digital tools that give parents and caregivers timely, trustworthy and locally relevant information. In communities with limited resources, where families may face transport costs, long travel distances, long clinic queues and few available clinicians, a well-designed child health chatbot could provide timely help with common childhood health concerns and guide parents on when to seek professional care if needed.
Any such tool should not replace or duplicate what healthcare professionals do, such as making diagnoses. Instead, it should act as a bridge between the home and the health system, supporting early care-seeking for serious symptoms while reducing unnecessary clinic visits for concerns that can be managed safely at home.
Research contacts and acknowledgements
This Review article was based on the article ‘Perspectives of parents and healthcare professionals on using chatbots to support parents in child health: A qualitative study’. It was written by Xolani Ntinga (senior research manager) in the HSRC’s Public Health, Societies and Belonging Division and HSRC science writer Jessie-Lee Smith.
For more information about this work, please contact Ntinga at xntinga@hsrc.ac.za.
The research team included: Joana Couto da Silva, Beatriz Félix, Dr Ricardo Melo and Dr Francisco Nunes from Fraunhofer Portugal AICOS; Leina Meoli, Dr Yaseen Joolay, Sarina Till and Prof. Melissa Densmore from the University of Cape Town; and Xolani Ntinga from the Human Sciences Research Council.
