GHM & HSG Guiding Research Special Series | Responsible AI for Stronger Health Systems

Artificial intelligence is rapidly transforming health systems, creating new opportunities to improve disease surveillance, diagnostics and decision-making. Yet as AI becomes more integrated into healthcare, important questions remain. How do we ensure these technologies are designed responsibly? How can they strengthen health systems without reinforcing existing inequalities? And what role should communities play in shaping the AI solutions intended to serve them?

In this episode of Guiding Research, Chaitali Sinha, Senior Programme Specialist in the Global Health Division at Canada's International Development Research Centre (IDRC), speaks with Dr. Rose-Mary Owusuaa Mensah Gyening, Senior Lecturer at Kwame Nkrumah University of Science and Technology (KNUST), Ghana.

Drawing on her work developing responsible AI for vector-borne disease surveillance, Dr. Gyening discusses how artificial intelligence can support stronger health systems when communities are involved from the outset. The conversation explores why trust, transparency and inclusion are essential to responsible AI, how citizen science can improve health interventions, and why technology should strengthen-not replace-the people and systems at the heart of healthcare.

In this episode

• Responsible AI and health systems 
• Community participation and citizen science 
• Trust, ethics and data governance 
• AI for disease surveillance and preparedness 
• Building inclusive and equitable digital health solutions 

This episode forms part of the Guiding Research podcast series, produced by Global Health Matters in collaboration with Health Systems Global (HSG). The series explores the themes of the Health Systems Research (HSR) 2026 Symposium and brings together leading voices in health policy and systems research.

* HSG’s involvement in the production of these podcast episodes was supported by IDRC.

Subscribe to the Global Health Matters podcast newsletter. 
 
Follow us for updates:

• @TDRnews on X
• TDR on LinkedIn
• @ghm_podcast on Instagram
• @ghm-podcast.bsky.social on Bluesky

 
Disclaimer: The views, information, or opinions expressed during the Global Health Matters podcast series are solely those of the individuals involved and do not necessarily represent those of TDR or the World Health Organization.  

All content © 2026 Global Health Matters.  

Transcript

Garry Aslanyan [00:00:00] Hi, everyone. Garry Aslanyan, your host. We have something special for you. Global Health Matters teamed up with Health Systems Global (HSG) to bring you a special series called Guiding Research. In this series, there will be guest hosts who will meet with experts in global health policy and systems research for one-on-one conversation that will delve into the themes of the upcoming Health Systems Research (HSR) 2026 Symposium and ultimately discuss how we can build sustainable health systems. Join us as we gain their insights into the impacts of conflict and climate change, the I hope you enjoy each of these episodes. Let's begin! 

Chaitali Sinha [00:01:04] Hi everyone. My name is Chaitali Sinha, Senior Programme Specialist from the Global Health Division at Canada's International Development Research Centre, IDRC. It is my pleasure to host this podcast episode, which is part of a series between Health Systems Global and Global Health Matters. The topic of today's discussion is platforms and participation in health systems. This is a broad topic, but today we are going to go into a specific aspect of it, which is the increasing use of artificial intelligence, AI. As the use of AI increases, important questions arise about how these technologies are designed, governed, and integrated into decision-making processes. While AI holds considerable promise for strengthening disease surveillance, diagnostics, improving the timeliness of quality health information, and supporting more effective clinical and public health responses, achieving these benefits requires careful attention to issues of equity, trust, accountability, and community participation. Today we are joined by Dr. Rose-Mary Owusuaa Mensah Gyening, Senior Lecturer at Kwame Nkrumah University of Science and Technology, KNUST in Ghana, to explore some of these questions and to listen to her insights on how she is addressing them in her own work. So welcome, Dr. Gyening. To begin, Can you please introduce yourself a little bit more and provide a very brief overview of your work on using responsible AI as a platform for participation? 

Rose-Mary Owusuaa Mensah Gyening [00:03:03] Hi Chaitali and everyone. I am Dr. Rose-Mary Owusuaa Gyening, a senior lecturer at the Department of Computer Science at Kwame Nkrumah University of Science and Technology in Ghana. I am part of a research group called RAPiD-VBP. Which is about developing a robust public health surveillance system for early detection and prediction of vector-borne viral zoonotic pathogens. That's quite a huge name, but in short, what we are doing is to find out how we can use artificial intelligence to support mosquito and vector-borne surveillance. In this project, we are looking at how we can make community members central to whatever we are doing. We believe that we can develop artificial intelligence solutions which serve the needs of people only when we involve them at their initial stages. In our project, we are listening to the people, we're listening to the environmental signals, and we are listening to the data. When we have combined all these three, we can come up with responsible solutions that respect our local customs, respect the local knowledge, and as well as solve the needs of the people for whom we have developed the solution. 

Chaitali Sinha [00:04:29] Thank you very much, that's wonderful. I can't wait to hear even more about everything that you're going to share with us. What would you suggest is the most significant thing people can do to allow us to build sustainable health systems in a changing world? 

Rose-Mary Owusuaa Mensah Gyening [00:04:47] That's a very interesting question. For me, it's all about first getting to the people, engaging in citizen science, letting the people contribute to whatever interventions that we are coming up with. Let me tell you this. In the communities, as the parents or the caregivers who know when people are getting sick, even before they report to the hospital, and it's also the people who know where the stagnant waters are, where mosquitoes are breeding. So, if we go to them to collect the data, these data that we get can be very useful to whatever we want to do. But if we just leave our sensors in the community to pick data and those data is used to train the models, we might miss out on very important information that would have gotten through the citizen science approach, where we could have got people to talk to us about what is happening in the communities. And all these can influence the interventions that we come up with. So for me, listening to the people observing the environment using sensors are very, very important. So not just observation, but listening to the people, letting them contribute to the solutions that we are coming up with, and that can really be helpful and can make a huge influence in the solutions that come up with. 

Chaitali Sinha [00:06:07] These are such very important points, thank you, and really putting people at the centre, the communities at the centre, and making them active agents in any solutions that they benefit from. So important for us to keep in mind. Now, you've already given us a lot of context and introduced us to the work that you're doing and why you're are doing it. I'm going to delve a little bit more into the specifics of the use of AI. We know it's becoming more and more visible across health systems and health systems research. As climate change is altering ecosystems, affecting humans, animals, insects and plants, disease surveillance systems are under pressure to respond more quickly and more effectively. With this in mind, Dr. Gyening. Where do you see AI having the biggest impact? And I'm going to ask you to choose just one. So what would you think is the biggest impact you see AI having? 

Rose-Mary Owusuaa Mensah Gyening [00:07:10] For me, the biggest impact is to see AI putting all the pieces together and helping us to see what we currently cannot see with our eyes. We have so many things going on in the world, climate change, we have mosquitoes breeding, we have air quality issues and all these can be very very very difficult to deal with within a very short period of time. So it's like trying to complete a jigsaw puzzle but if you have an AI agent helping you in that it can be really easy to do that. So the AI agent can pick weather data, it can pick acoustic signals from mosquitoes. We know in our projects, we know that mosquitoes carry different wingbeats. When I talk about wingbeat, I'm referring to the sounds that they make. The wingbeats can help us to detect the influences and even the species of mosquitoes that we are dealing with. So if we have an AI agent picking mosquito acoustic signals, we have AI agents taking weather data, taking hospital records, taking information from the communities themselves. Then this AI agent can help us to see the things that we cannot see and I think that would be a very huge impact. 

Chaitali Sinha [00:08:30] Thank you so much, that's excellent, and thank you for walking us through the specific solution that you and your team are working on right now. Before we get into more about that tool and what it is doing and the network that it's part of, I want you to take a step back, if you will, and give me your reflexion on what you think is the most significant risk for us to bear in mind. I asked you about the biggest impact that AI can have. Now, can you reflect a little bit on the most significant risks? 

Rose-Mary Owusuaa Mensah Gyening [00:09:06] That's a very, very interesting question. Most of the time, we are so excited to talk about what AI can do, but we are very reluctant or we forget to talk about what could go wrong or what could be the biggest risk that we are ignoring. For me, we have to understand that AI is smart, but it cannot be smarter than it really is currently. And also, AI will be smart based on the data we feed to it and then the assumptions that we make. So if we are working with systems that are currently isolating certain groups of people or making them invisible, then we may be developing solutions that are not inclusive enough. So the biggest risk I will say will be in the area of inclusivity. If you are designing AI solutions and the grandmother in a rural area cannot use it or cannot contribute to it, it can be a very, very huge challenge. And the solutions that we have, unfortunately, we cannot say are representative of the people for whom we are developing these interventions. 

Chaitali Sinha [00:10:16] What a great response. And I'm just thinking back to your response on the biggest impact. And it seems they're almost two sides of the same coin, because you had mentioned the biggest impact of AI can help us see more and get pieces to come together like a jigsaw puzzle. I believe that's the analogy you used. But then if you're not seeing everything that you should, if you are not including everyone in that view that you have. Then you're missing out on some very important pieces of that jigsaw puzzle. 

Rose-Mary Owusuaa Mensah Gyening [00:10:50] I agree with that. I definitely agree with that and I think that researchers have to put in a lot of effort to get community members to participate. In certain parts of our world there are superstitions and people feel that when they are contributing information in the form of let's say their voice or they are even giving images or they are coming up with an intervention they may feel that perhaps somebody can use the information for other superstitious things like witchcraft activities or things that you as a researcher may not even imagine. So if we want to deal with this risk that we are talking about, which is about inclusivity, then we have to go the extra mile to educate the people on what we want do, tell them about how they are going to benefit from the solutions that we are coming up with and keep them actively involved. One thing that comes up every time when we go for community outreach is you people keep coming for data, but we never see results. So the people want to see how the data they have given you is leading to the results that we envision. Even if it's going to take a very long time, they want to see how far we have gone, what the challenges are, and how they can help us to deal with them. So if we are able to get the people involved, then I think we can mitigate the risk that we are talking about in terms of data bias. If we are able to minimise data bias, then we are definitely going to get better AI solutions that can really serve us. 

Chaitali Sinha [00:12:31] Great, and you know what, this really is a wonderful segue into what I wanted to ask you about next. You mentioned already about the solution that you're working on, which listens to mosquito wingbeats and helps with collecting that data and doing AI-enabled analysis to support preparedness and response against vector borne infections. You talked about inclusivity and making sure that communities feel engaged and active members in shaping what AI is being used for and how it is being felt in their own communities. Can you give me a little bit more insight into how the use of the AI solution you have developed and have deployed across Ghana is influenced by and perhaps influences gender relations and social dynamics, and also things like access to digital communications. You've mentioned inclusion. What is it that you would say, from your perspective and experience to date, are the real benefits of these AI-driven tools? And at the same time, what should we continue to keep in mind to make sure that we're not creating new inequities or reinforcing existing one when it comes to things like social and gender inclusion? 

Rose-Mary Owusuaa Mensah Gyening [00:13:53] So the AI solution that we have is a mosquito tracker, which first of all is able to work in a trap. So as the mosquitoes are trying to enter the trap, the device will pick up the wingbeats or the sounds that the mosquitoes are making, and it's able to detect the specific genus of mosquito that is going into the trap. Why do we care about this? We do care because we know different genus of mosquitoes carry specific diseases. So the mosquito that causes malaria, for instance, is different from the one that causes dengue and even zika. So in Ghana, we deal with malaria more often. If we know that a lot of... Mosquitoes are being recorded in a certain region at a certain time period, then the government can come up with targeted interventions that can be used in those specific areas. Now we can have this wonderful idea, but still the issue of gender and digital access can limit the solution. So, what are we doing or what have we done in the past? We have gone to the communities several times. We have been to several communities, and then we have spoken to men and women and the vulnerable groups. We have made them understand the solutions that we are coming up with. And they have also given us our concerns with respect to internet issues, even mobile phone access. And even the time it would take for them to contribute to the solution that we are coming up with. For instance, somebody told me that if it's going to take a very long time to mount the trap, and get the mosquitoes to come in, then that person may not be willing to be a part of it. So if the person will have to be taken to the classroom to be taught how to set up the trap and then how to do the recording and all of that, then it might be difficult. So what did we do? We have automated most of the things. The most important will be for the people to set up the trap which will not take a lot of time and then AI does a lot. So you just set up a trap and then the machine learning model running on the trap will do the recording of the sound and then also do the interpretation and send the data to an external dashboard which is connected to the solution. So the people in the community do not have to carry any information on a pen drive or record any from the trap to send to whoever. We have automated most of the things. For me, that solved the problem of time. In Ghana, our women are very busy. They manage the home and all of that. So if you are going to take an intervention to them and it's going to take a chunk of their time, they are likely to resist. So we have automated majority of the things that we are doing. And the only thing that we'll teach them to do will be perhaps to set up the trap and then keep monitoring it as and when needed. So, and with respect to the internet access, it's a general problem in Ghana. So we have made sure that the models that we have deployed on the trap are lightweight and they don't even have to connect their phones to the trap for it to work. So we would provide the internet. In a certain room within the community or at a certain location and then the internet source will be connected to the trap then the trap would upload the information to the dashboard. When it becomes very necessary for them to send us any information we have a very simple interface where they can record whatever they want to send to us in their local language and that takes very minimal data. 

Chaitali Sinha [00:17:53] It's just so important to really make it meaningful for the people who are using it. Thank you for sharing that. And for walking us through what it means to really use the solution that you developed on the ground so that the data is collected in a respectful and meaningful way and then used by those who need to use it. Now, this solution sounds so thoughtful, it also sounds very robust. And relevant to local communities and also national health priorities. Does it work in isolation from other solutions? Can you tell us a little bit about how you and your team share experiences and also learn from each other and from others outside of your network? 

Rose-Mary Owusuaa Mensah Gyening [00:18:41] So our project is under the Ghana hub of the AI4PEP project, which is the Global South AI Network for Pandemic and Epidemic Preparedness. The beauty about what we are doing is that mosquitoes can be found in so many places. However, the diseases that these mosquitoes carry may be different. So in Ghana, we have malaria being prevalent. And then if we go to, let's say, a different country, such as the Philippines, we might have Dengue being prevalent over there. So the beauty of what we are doing is that with the support of AI4PEP, which is being funded by Canada's IDRC and UK's FCDO, we are able to marshal the resources that we need. To first of all gather local data in Ghana and local knowledge and train the models to adapt to our environment and then we would go to Philippines to also first of all check how well our model is performing over there and definitely we expect to make some modifications so we are also going to collect local data on mosquito wingbeats and also local knowledge which we are also going to feed into the model. So with the support of this global network that we have and the wonderful sponsorship packages that we have from IDRC and FCDO, we look forward to a time where the intervention that we have can be deployed in many environments and with just few adaptations. 

Chaitali Sinha [00:20:18] That's great, and I think that speaks to, as you were saying earlier, about how communities need to understand how AI works. But given the novelty and the fast pace of evolution and change within AI models, I think literacy and capacity strengthening goes across the board, whether it's people developing models, decision makers, and also community members. So I think the network model that you discussed of sharing is quite interesting. Let me now go to a final question for you. Some people might argue that the growing focus on artificial intelligence risks distracting attention from the deeper structural challenges facing clinical and public health systems. What would you say to someone who sees AI as part of the problem rather than part of this solution? 

Rose-Mary Owusuaa Mensah Gyening [00:21:15] So what I will say is that we have to understand that artificial intelligence cannot replace hospitals, artificial intelligence can not replace nurses, it cannot replace health workers, it cannot even replace the health infrastructure that we need to manage these facilities that already there. But we should also not dismiss the impact of AI. Let me put in this analogy. See AI as a flashlight which is going to help you to see the potholes on a road which you might not have seen in the night. So AI is a solution that is going to augment what is already there. It's not going to say that okay we have AI so we don't need the infrastructure or we have AI we don't need the nurses or we don't need a community. AI is a tool that is going to support them, going to help them to see the things that they are not seeing within a very short period of time. So don't see AI as competing with health systems, but rather strengthening the health systems that we have. 

Chaitali Sinha [00:22:23] I really like that analogy of the flashlight, the headlights and the cars shining light on the potholes. We hear a lot of analogies when it comes to AI and health around AI as a fancy car on roads and infrastructure that simply don't exist or are far too dilapidated to be used by any car, albeit even a fancy. And your analogy takes that even further. It talks about the headlights on cars. Which allows any car, whatever maker model you have, to be able to illuminate the potholes on the road that lies ahead so that you can drive safely and securely. That's great. Now, before we finish, I'm gonna ask you for your reflexion on one final thing in terms of the future, what lies ahead. You mentioned that AI cannot and should not replace health system actors. It shouldn't replace health systems infrastructure or information flows, but rather enhance them. For the people who are sceptical about AI, and there are many, and also for those who are quite concerned, what would you say from your experience would be the best way to ensure that there are sufficient protections and safeguards? Built into health systems using AI to ensure that AI does what it says it does. It supports those it's meant to support and it strengthens health systems rather than fragments them, especially in low resource settings. 

Rose-Mary Owusuaa Mensah Gyening [00:23:57] If we are okay to accept that AI is here to support, I think we can really go far. We can solve a lot of issues within a very short period of time, instead of waiting for years to deal with issues that AI could have helped us to eradicate in few minutes or few months or in few years. So we cannot blame the people for being sceptical. I mean, it's normal with everyone. If somebody is introducing a new technology to you and says use it I think is going to help you, you have some questions that will have to be answered. So our duty as researchers is to help the people in the community, the local people, especially in the remote areas, understand why we need these AI interventions and how we are going to protect their data. At the beginning, I mentioned that some people feel if they give information, it might be used for other things. We need to help them to understand how we are going to protect their privacy, how we are going maintain their trust in whatever we are doing. We need to be open. About how the data is going to be used and as I mentioned earlier feedback is really important. They need to be assured that whatever data we picked from them has been used for something and whatever solution that we have come up with eventually it's going to benefit them. If we keep going to them every time and we keep asking for data and they don't see solutions their trust will not be there. If you don't tell them how we are going to store and use data, we would not have met the privacy requirements. So safeguarding is very important in the research that we are doing and we absolutely do not take it for granted. 

Chaitali Sinha [00:25:45] That is wonderful, Dr. Gyening. Thank you so much for such an insightful discussion. You have reminded us so eloquently that there is a lot to be excited about when it comes to using responsible AI. So that's AI-enabled solutions that are rights-respecting, sustainable, inclusive, and ethical. You also have reminded not to forget about the basic public health health research ethics and standards. That we must uphold when it comes to using any sort of AI-enabled system in health systems so that it strengthens inclusion and it helps make sure that the health systems are working for those that it's meant to serve. 

Rose-Mary Owusuaa Mensah Gyening [00:26:33] It's been an absolute pleasure to be here to talk about AI and I know that many people will be listening and life will be transformed by the solutions that we are coming up with. 

Chaitali Sinha [00:26:44] As we bring this conversation to a close, I am struck by a theme that ran through everything Dr. Rose-Mary Gyening shared today. Meaningful innovation begins with people. We heard how AI can help connect complex pieces of information, from mosquito wingbeats, climate data, and community observations to strengthen disease surveillance and preparedness. We also heard a powerful reminder that the most important data often comes from the people who live with these challenges every day. Communities know when illness is spreading, they know where risks emerge, and they hold knowledge that no sensor or algorithm can capture on its own. Rose-Mary referred to a jigsaw puzzle that needs to fit together to show a picture that resonates. With people in communities and with policymakers alike. These examples demonstrate that participation is not simply a box to be checked. It is a fundamental to building AI systems that are trusted, relevant, and effective. Whether it is engaging women and caregivers in the design of interventions, adapting technologies to local realities and constraints, or creating feedback loops so communities can see how their contributions lead to action. Inclusion must be built into every stage of the process. This approach of leveraging responsible AI for resilient health systems and equitable health outcomes is at the core of the AI for Global Health and AI4D initiative. As AI continues to evolve, our challenge is to ensure these technologies strengthen participation rather than replace it, support health workers rather than sideline them, and help build health systems that are more responsive, equitable, and resilient. As we were reminded by our guest, the promise of AI lies not only in helping us see more, but in ensuring that everyone is seen. Thank you to Dr. Rose-Mary Gyening for sharing your insights and experiences. And thank you to our listeners for joining us. 

Garry Aslanyan [00:29:07] To learn more about the topics discussed in this episode, visit the episode's web page, where you will find additional readings, show notes, and translation. Don't forget to get in touch with us via social media, email, or by sharing a voice message. And be sure to subscribe or follow us wherever you get your podcasts. Global Health Matters is produced by TDR, a United Nations co-sponsored research programme. Based at the World Health Organisation. Thank you for listening.