SN 2 EP 9: Data Stewardship, AI, and the Future of Health Research in Kenya with KEMRI’s James Ngumo Kariuki

SUMMARY KEYWORDS

data, evidence-based, research, county, kenya, hackathon, technology, big data, years, health researcher,  health, challenges, output

EPISODE SUMMARY:
In this podcast episode, James Ngumo Kariuki, a senior research scientist at the Kenya Medical Research Institute (KEMRI), shares his perspectives on the critical role of data stewardship, artificial intelligence (AI), and technology in shaping the future of health research in Kenya. Initially interested in medicine, Mr. Ngumo shifted to statistics and computing, eventually becoming an assistant research officer at KEMRI. In this discussion, Mr. Ngumo highlights KEMRI’s use of big data and analytics, noting challenges like siloed work and the need for critical thinking. He reflects on the accelerated adoption of digital tools during COVID-19 and the role of evidence-based policies in achieving universal health care in Kenya.

He also discusses the challenges in accessing medical data, emphasizing the importance of appreciating data’s value and asking critical questions to drive its use in predictive healthcare and system improvement.

HERE ARE THE KEY THINGS TO LOOK OUT FOR:

• Mentorship: The crucial influence of mentorship in Mr. Ngumo’s career development, including specific mentors who guided him and the impact of their support.

• Digital Tools: The rapid adoption and integration of digital tools during the COVID-19 pandemic, and how these tools have transformed health research and data analysis.

• Policy Impact: The significance of evidence-based policies in achieving Kenya’s universal health care goals, and how his research supports these initiatives.

• Data Access: The challenges in accessing medical data, the importance of this data for predictive healthcare, and how better access can lead to significant improvements in health systems and outcomes.

EPISODE TRANSCRIPT

Intro:

Welcome to Mazungumzo, African scholarly conversations, a podcast that highlights the perspectives of various stakeholders in academia, and research fields across Africa through open dialogue or mazungumzo on scholarly communication in Africa.

Joy Owango::

Welcome to Mazungumzo – African Scholarly Conversations, where we are joined by an expansive list of African policymakers, science communication specialists, innovators, and tertiary institution leads who contribute to this realm of science communication and Scholarly Communication.  I’m your host Joy Owango:, the Executive Director of the Training Center in communication, a capacity building trust based at the University of Nairobi Chiromo campus in Kenya.

We have the pleasure of welcoming Mr. James Kariuki Ngumo to our podcast, Mr. Ngumo holds the position of senior research scientist at the Kenya Medical Research Institute. He has over 20 years of work experience in health systems research, knowledge management and translation, training and capacity building for the help for the healthcare workforce, and collaboration in inter-regional projects focusing on health and water sectors. His interests lie in data analytics, health systems and strengthening and knowledge management, driving his commitment to advance in these areas in the healthcare field. A warm welcome to the program. James.

James Ngumo:

Thank you so much joy, it’s a pleasure to be part of this conversation. I’m looking forward to a nice time of free discussion, Think Tank discussion around the issues that we are going to discuss.

Joy Owango:

Absolutely. So, to begin with, could you share with us a bit about your professional background and journey? How did you end up in the field of health systems research and knowledge management?

James Ngumo:

Thank you so much, and welcome to all the listeners. Wow, my journey, I love to tell stories. So, allow me to tell the story. And we’ll pick out the key points. So, my journey in KEMRI was an accident.It was totally an accident. Just like any other graduate who has just done the first degree, I found myself, first of all, not knowing what else to do in the university, when you just pass the exams, and you land in a university.  I missed the combination of points.so, I was just thrown to do statistics and computing. It was the last resort, the fact that you have qualified to enter a public university.So that is how I found myself being statistics and computing for four years.

Joy Owango:

what is it that you missed, because you ended up doing a program that involves just pure math?

James Ngumo:

What I missed, first and foremost, I wanted to do medicine, just like any other secondary school student. And then I told myself, okay, fine, let me go and try to come BCOM now that I didn’t make the cluster, and then I was told it is full. I told myself, okay, I will do law, I was nowhere near the witnesses. So, under the University Council, they got me to do with a Bachelor of anything, as they call it, B.A they told me just go to the statistics class, after one year, we’ll think about you can transfer. And I fell in love with math. Then during the holidays, you know, the long student holidays, I would end up seeing some calls here and there regarding KEMRI looking for people to collect data. And so, I applied and I would be called in to collect data in the field letting go of collecting data. There was this big four national surveys taking place. And the fact that I had some skills in computing. At that time, it was all about questionnaires but I had some skills on computing and those are the days when you do a computer program virtually, it is not like what we know today that time you write it from zero. And therefore, they took me to be like a data manager. And upon my graduation from Kenyatta University, they took me in to be now assistant research officer on the biostatistics and health informatics program. That was my entry. Yes, so up to that point that is how I came into KEMRI Interesting Because the advent of computing was moving from the command-driven programs, where you type the command to these beautiful icons. Also, it is passion that drove me to get into deeper programming, a bit of computer networking. So, I found myself in that Department of Health Informatics squarely in analyzing data and also installing computer systems. That is a humble journey 25 years ago today.

Joy Owango:

And then how did you find yourself in health systems research and knowledge management?

James Ngumo:

Thank God for people who are called mentors. After 11 years KEMRI requires that, you know, when you come as a research assistant, you’re required to go back and do a master’s. So, the research assistant posts, you are under training. And so, you’re not a scientist, you’re just under training and you’re working under somebody, so that you go for your masters when you exactly know the path. And that is when I met somebody who was a very senior scientists who was about to exit and said, Look, you can continue pulling computer cables for your life and you can continue analyzing statistics, it’s good, you have the statistics skills, but from the way I’ve worked with you for four years, why don’t you try the health systems, the future of the world is going to be health systems in another 10-15 years. And then combine it with your health informatics and your analytic skills, you’ll be good to go. And, therefore, I went back for my masters. And they decided to look for where I could do health systems, and they went out to the country to Israel, to do the health systems management course. it’s a very interesting life, you know, being mentored here and there.

Joy Owango:
mentorship is very important. It’s actually key to any researchers anyone who wants to succeed in life, having good mentors is so important. So now, and this leads me to the next question. So, considering the diverse range of diseases and outbreaks addressed by KEMRI research centers, how does the institute leverage big data and data analytics to enhance research and processes? Are there any specific initiatives or projects you are currently involved in that utilize AI technologies? So basically, just take us through how you’re leveraging on data and data analytics, to address some of the outbreaks of diseases, the various range of diseases that KEMRI is working on.

James Ngumo:

Let me say that KEMRI collects a lot of data of different studies. They collect a lot of primary data, a very useful source of data. Clinical Data, patient data, vector data, pathogenic data. So, the tragedy is that we work alone. in silos, we work in silos. So, it is about Kariuki, getting a grant and working in silos. That means I choose who to work with, I go to my laboratory, and shut myself in. So, there is a lot of primary data. And that is a very useful source. On the other hand, there is plenty of unutilized secondary data. It has been in the shelf for 8-15 years. Now, what KEMRI has done six years ago, they realized that we have such a big pool of primary data, can this data be repurposed? And that is when I was tasked together with two other colleagues to think about it. And for the first time, two years ago, we hosted the first data hackathon that primarily focused on repurposing biomedical data, to give it a new lease, and see whether we can use it together with, evidence, publications, and see whether we can model new things to inform UHC and other disease, pandemic preparedness. That was a unique thing because when you do a data hackathon, you’re basically asking yourself, who is clever enough to set the first set of questions? And which tools are you going to use? I’m happy to report to you that the biggest problem we have is not the absence of tools. We have enough statistical tools. And they are all open source. However, the critical questions that will require a critical mind to critically come up with an innovative statistical model is the deficiency. And that is what we saw in the data hackathon. So it  was very interesting. We had interesting results, but I was disappointed because only 5% of the 90 students in Kenya were able to give us something meaningful. So that is innovation that KEMRI did.

Joy Owango:

So now when you are taking us through this, especially the hackathon, did you leverage on imagine AI technologies? Have you started using AI technologies to address some of the outbreaks and diseases that KEMRI is focusing on?

James Ngumo:

No, we did not use the generative AI that is now currently in place, but what we did is, we use something called machine learning, where you train the machine on sample data, and then you test it. And you see what kind of output model it’s giving you. So that’s the furthest we went. We anticipate that from January next year. Again, we’re going to have the second data hackathon. And the focus will be on AI now. How we can leverage most of this open source AIs  in combinations to inform evidence, we will talk about it as we go on. And I will tell you some very interesting stories that I am personally championing.

Joy Owango:

Yes. So, in your career journey, what have been the most exciting or impactful technological advancements that you have witnessed in the healthcare and research domains? And how do you envision these technologies shaping the future of healthcare and research in Africa?

James Ngumo:

Joy? Can I say one thing? Yes. COVID was a blessing.

Joy Owango:

Oh, yes. Yes, that is true. We saw the innovation coming out of this continent. Yes, that is so true.

James Ngumo:

Who could think we could use Zoom; we could train students through zoom. The biggest part is the breakthrough where the government of the day in attempts to control the pandemic they realized frustration from the international community. And they started asking, can we have an internal think tank that will advise us on how to treat and manage COVID? The think tank that came together, started producing evidence and extracting the implication of that evidence to inform decision making, the face masking, all this was because of a group because the government had decided to demand for evidence.

So I want to say that the most interesting thing I’ve seen is the big rapid change in the use of technology. Honestly, you can’t keep up with technology. I used to program in the very old basic and the rest, the command driven, then they became object driven. Now we have apps.The apps have been a wonder, and then you can’t keep up because the young generation that were we thought are young nowadays, you give them a computer they generate for you outputs or make you an app.

We are like where were you trained? We trained ourselves. So, for me that that rapid change of technology, and especially exploring how this technology can be customized for public health intervention. My goodness, I’m so excited. Joy, I forgot to tell you that when we were doing a hackathon, it was open to anybody in the public. We will give you the data, give us the output. You sort out in between, that is where we attracted 470 participants, individual participants out there. So, imagine if that is the skills that is unutilized out there. How much more Can we use that skill to customize the open-source technology to inform public health decision? I will expand on it. When you look at UHC, I’ll tell you what we were doing at the institute. Very interesting thing in this group.

Joy Owango:

Interesting. So now that you’re talking about UHC, that is actually the next question that I have lined up for you. So, the National Health Policy strategic plan in Kenya identifies the need for universal health care. And this has been considered as a priority. But in your opinion, what are the key challenges of barriers to achieving this? And how can evidence informed policies and research contribute to addressing these challenges, including the adoption of technologies?

James Ngumo:

Thank you. I just love the UHC. I was in the initial setup of the UHC. For us It was a research component we were supposed to track how UHC is performing. There was the four pilot counties of Machakos Nyeri, Kisumu and Isiolo.They were all strategically chosen And I knew we were going to have some really wonderful data. Second part that just to set the stage was the fact that health had been the devolved So the decision making was actually at the county levels. So, you give them the big roadmap, the guys sort out the operational part of it. The tragedy is, we never learned from research. That is why the pilot failed., we went and expanded because it was a politically driven process, a national government process, we did not do a thinking session with the policy actors. So, the researchers continued generating useful research evidence that was not being consumed. In the next of August, we are going to launch UHC for the third time. Now, what is going to be different this time from my own assessment is the fact that the counties are asking for evidence, what has worked, what has not worked, And they’re going to tie that to the big data. So, if they tie with the kind of data they have, that they draw insights and perception, then they will need customized solutions  like to chat GPT you asked the question, it will give you an output. But we need to customize that output for that decision maker at the county level. For UHC to succeed, it has to have homegrown solutions. County grown solutions, we cannot assume that it’s uniform, each county is so unique. So is a strategy, but technology is uniform it can support and that is where we are now concentrating by providing for them something called The Interpretive Afya, which is a dashboard that summarizes evidence and gives you in the form of small summaries which they can act upon, recommendations that they can act upon.

Joy Owango:

So they are already leveraging on technology on how you could actually use universal health care to support the community. Right?

James Ngumo:

Yes, Through registration of people, tracking of medicines and the monitoring evaluation. They are very nice. They have very nice dashboards. Beautiful. But can I tell you something’s on silently there. The Vested interests in the procurement process. That is why we cannot see the benefits of the UHC we need to separate the vested interests from service delivery.

Joy Owango:

Just like any government or national service, it does have its challenges because there’s only the human element and the administrative dynamics which make it a bit difficult to achieve what could otherwise be a noble goal.

James Ngumo:

Very true. But it is a self-inflicted wound. You know this is good, It’s like me telling you just take a pinch of salt but because you love putting a lot salt in your food you go ahead. There is enough research recommendations on where we need to adjust. But because somebody is going to lose, especially on the pharmaceutical and non-pharmaceutical product, they will rather destroy the government’s supply system. It is so intentional. The big insurance companies, the big companies don’t want people to register an NHIF, because it’s going to shift the dynamics of the market and they will lose out. So, block. So UHC is good on paper, but poor in implementation.

Joy Owango:

Interesting. So, this actually connects to the next question, what role do knowledge brokers play in bridging the gap between research and policy and facilitating evidence, informed decision-making processes?

James Ngumo:

I will do it this way; I will give very practical things that I have been engaged in and have been championing as a knowledge broker.What we decided to do is that for each county, we decided look at what is the research evidence that has been generated from all the teams that do research regarding a specific county, so we took Makueni, eight years of Makueni Care? what does the research say? And we generated summary evidence briefs, three of them. That’s the first thing that we did. And then we had an engagement with the new county health management teams. And we told them, hey, look, this is what we have learned over the eight years. They received that information very well.The next thing is the, and that is what we call the know part. Then they ask, what can we do about this recommendation? How do we turn this recommendation into actionable? We took them to we look through literature for each recommendation, can we have a success story in Kenya within East Africa? So when they’re seated in the same forum and you have a participatory approach to creating actions based on evidence, that’s the experiential learning, based on successful case studies, and based on the evidence, and they come up with actions. And you also tell them, how can we measure the progress now that you have this action? What are the indicators we’re going to measure. That is what we were telling them Welcome to the world of evidence, informed action, where everybody is a co-creator, in this evidence ecosystem We tested it in Taveta, and they were so excited, they started generating other critical question. So one thing led to another one evidence led to another. And what we what we were so proud about, is the fact that we have taught them how to ask critical questions, not just a question, a critical question that is meaningful that we will be able to get literature out of, we have been successful in three counties, we have looked into the Institute, and through the Council of governors, can we spread it out to another county? That is how we want to minimize the know- do gap. They know from recommendations, but how do they act on that recommendation? That is the next big frontier, the know-do gap.

Joy Owango:

So now, in your experience, what are the some of the key challenges or limitations in accessing and utilizing medical data for research purposes? And how can these obstacles be overcome?

James Ngumo:

For Okay, number one, the critical parties, they don’t know the value of the data.

Joy Owango:

Is it value or do they don’t understand the data? Because data has no meaning at that point.

James Ngumo:

The data that they have, they do not appreciate the value. The value is that you’re sitting on a goldmine but you don’t know. They can actually revolutionize the whole healthcare system, if they value the data that they have. So, that is why whenever you go to any of the public hospitals, they take your history, everything and they put it aside, it will never be used again. They do it because it is a requirement of law. That is what they do. Number two, they do not have a critical question. This issue has come up over and over again. You see if you don’t have a critical question, you don’t go look for data. So, for them, it’s a treat, go home treat go home. Is the data making sense? No, we just draw a bar graph and the line chart. So that is where they are. But can we move away from curative to preventive. That is what my team is working on. Telling them look this patient data that you have can be structured in a way that it can actually be used to predict, maybe somebody’s going to be unwell. And they said please tell us a case study. Did you know that the Ford vehicles in the US in the last 15 years. And this is a case study, I always tell them in the last 15 years, they have really improved the features of their vehicles to minimize road traffic injuries. But where did that data on impact on car injuries come from and don’t get the fact they knew they cannot get hold of the Chinese or the Japanese vehicles and take them through, they needed something real. And what happens? The motor vehicle sponsored Johns Hopkins University School of Public Health to help the Kenya government study for three years, the road traffic injuries, what kinds of trauma, what is the outcome? And finally, these guys took the data and they went, and they left us with a very clean server with new equipment but the data was not there.

Joy Owango:

Hold on, we do not have that data?

James Ngumo:

it disappeared. I tell you what they did. They removed the external disk drive, and they put a new one, they increase the memory of that computer and everything.

Joy Owango:

And then they went to the most important bit the data,

James Ngumo:
that data because to us, as long as I was getting some form of top ups, as long as I was going to the field to collect data, as long as I was doing what the protocol requires. I didn’t care about the data. I didn’t even see the value, the end game, as long as we drew a little graph and publish it to publication. it did not have any meaning to the Kenyan public. Nothing with the Kenyan public,

Joy Owango:

Weren`t there any systems set in place on how that data can be released?

James Ngumo:

15 years ago? those things weren`t there, it was just giving them the data

Joy Owango:

Oh, goodness. And it was so easy to work in Kenya because you remember the rate of accidents at that point. So this was a very good place to do that research.

James Ngumo:

And the data we gave them free data, and we`d tell them to take more and  can we even expand to other districts. So they covered 65% of the country, massive data! From hospitals and police files. So, you can imagine the things that we’re working on it. So, when NTSA wanted this data much, much later, they were told hakuna.

Joy Owango:

Is it possible to go ask them for it?

James Ngumo:

No, those guys they know their rights. And by the way, who are you to go and ask for them?

Joy Owango:

We willingly gave it out, But it also comes to literacy on data protection and data security, which we did not have in the Global South. We did not have t adequate information about it in Africa. And it leads directly to the final question because you’re talking about loss of data. Data security and privacy are critical considerations when working with medical data. Just the example you’ve just shared with us how you give up the data and there’s no you can get it back. So how do you ensure the protection of sensitive information while maximizing its potential for research and innovation? bearing in mind right now, we are all talking about Open Science, open data, open infrastructure, you know, so how do we factor data security and privacy when it comes to medical data so that number one, we don’t lose our data. And number two, we protect whom we are getting data from.

James Ngumo:

First, let me say, first and foremost, the coming in of the data commissioner was a blessing. Number two, the work that they are doing to sensitize all public officers, irrespective of which ministry is very, very important. I got annoyed. I almost shed tears. We`ve just seen the KDHS report. And everyone’s clapping hands. But let me tell you the tragedy, the tragedy is this. This data is not hosted in Kenya.

Joy Owango:
Where is the data hosted?

James Ngumo:

In the US, we were actually asking for data tables. So, they have this guy from ICS. And as we were collecting and uploading, the data was going there. But one thing I thank God for, during the time, we are going to write to analyze the data, people started asking questions, just release the data to us, instead of waiting that you give the request, it goes overnight, you get it the following day, No, no, no, we just want to have the data here. And therefore, we started asking ourselves, you cannot get raw data, semi processed data about the Chinese population.

Joy Owango:
You can’t even in Tanzania, you can’t and that’s next door. You can’t.

James Ngumo:

But in Kenya, we are waking up so late. So first thing we need to do before we continue making more plans, we sensitized everybody

Joy Owango:

on the data management,

James Ngumo:

Yes,We start knowing the implication of giving personal data out there? And then what is the good? What do we get as outcome. What are the benefits, to us as a whole? So, there’s a lot of discussion and awareness creation, and that is what is actually happening right now. If we really have to make gains with the Data Protection Act, we need to punish those giving out personal data illegally.

Joy Owango:

Especially, now that there is the trend on open science and open data, there is a need to educate the community, on protecting their data, the degree to which they can share their data, understand the data laws, understand their respective national data laws, as pertains to what is being proposed within the open science community so that you don’t lose critical information and, in this case, data that would otherwise belong to you to a nation. So, because you know, once data is gone, that is it, retrieving it is a bit of a challenge.

James Ngumo:

Yes, Joy let me tell you something, for the last three years you cannot get primary and secondary school childreninKenya’s data. The Ministry of Education nor the State Department of Education. First of all, they critically look at why you require all this information about our children, and then they start determining. We were even being told that When you’re going to select these students for a study, you have to code and it’s the Ministry of Education that will code for you, because they have the registers of students. So, you want to go to x primary school, they have, you want to interview class sevens, we will have who is in that, and then they will give you based on your criteria. They will randomly pick for you, but not tell you the name and the contact and everything. And then the Ministry of Education is very, very careful. It says that for this and this parameter, you will not analyze but you can this part and this part. Now, they’re protecting the subjects.

Joy Owango:

Absolutely. And those are very vulnerable. subjects Yes, that is true.

James Ngumo:

You can’t even write the name of schools in an output you. So you write Nairobi School X Y, Zed J K?

Joy Owango:

This is such a good move, and that is where we need to head towards because, we are all pushing for open science. There is very little conversation on the degree to how we can protect our data. We also, need to look at data sovereignty, you know, like the case of what we just shared with me of the loss of data to John Hopkins University School of Public Health, School of Public Health. I mean, that loss, the one thing that was very clear is loss of data sovereignty. And today, there are some that technologies that are aiding in, in protecting our data, you know, with the rise of digital object identifiers, and different types of persistent identifiers, checked to me synthesize data, but also raw data. And that is the way we need to go, we need first of all, the awareness on data protection, and data sovereignty. Also, we need to also understand to what degree we need to share our data, we need to be very critical in the kinds of partnerships we create, especially in our contracts, which will lead to the sharing of data. And all this is true, you know, unfortunately, we burnt our fingers. But we learn from that, we learn from that and look at how best we can mitigate that in the future. And you know, from what you’ve shared with me, it’s quite an interesting journey that you’ve gone through. And I love your views on what we need to do some of the challenges we are facing and the views that you have on what we need to do to protect our data. And what we need to look forward to in the future. You know the new technologies. And I thank you so much, James, for your time.

James Ngumo:

Thank you. Thank you so much, even for having me in this discussion. This is work is progress but we are at a turning point.

Joy Owango:

We have a turning point. And that’s a nice way to wind up. Thank you so much, and goodbye for now.

Outro

Thanks for joining us on today’s episode of Mazungumzo podcast. Be sure to subscribe and follow us on all our channels for more updates and for candid stories by researchers, policymakers, higher education leaders, and innovators on your journeys. See you in our next episode.

Listen to the full episode and explore more episodes from the #Mazungumzo- African Scholarly Conversations podcast on the following platforms:

• Buzzsprout: https://www.buzzsprout.com/2140692/15249435

• Spotify (Available Globally: https://open.spotify.com/episode/6NUSbRdJKvrqRVQunTcFYq?si=580e5c07e1aa45ad

• Apple Podcasts (Globally: https://podcasts.apple.com/us/podcast/sn-2-ep-9-data-stewardship-ai-and-the-future/id1652483621?i=1000659023258

• TuneIn (Available Globally): http://tun.in/probT

• Player FM (Available Globally: https://player.fm/series/3450840/423554746

• TCC Africa-Training Centre in Communication News Room (Available Globally): https://www.tcc-africa.org/category/mazungumzo-podcasts/

Listen, Learn, and Share! Let’s amplify the voice of African scholarly contributions.

#Sciencecommunication

#researchvisibility

#Sciencecommunication #researchvisibility