Can faster access to real-world data actually change patient outcomes, or are we still too reliant on controlled clinical trials to see the full picture?
In this episode, I sit down with Dr. Alex Asiimwe, Executive Director of Epidemiology at Gilead Sciences, to explore a topic that doesn't get enough attention in the AI conversation, real-world evidence.
While much of the industry focuses on AI in drug discovery or diagnostics, Alex brings a different perspective, one rooted in what happens after treatments reach real patients in the real world. As he explains, clinical trials may be the gold standard, but they are still controlled environments. Real-world evidence is where we begin to understand how treatments perform across diverse populations, healthcare systems, and everyday conditions.
What stood out in our conversation is just how messy and fragmented that real-world data can be. Much of it is not collected for research purposes, which means it takes months, sometimes up to a year, to clean, structure, and analyze before it can inform decisions. Alex shares how AI is beginning to change that, not by replacing human expertise, but by automating the most time-consuming parts of the process. If that timeline can be cut in half, the impact is immediate.
Faster evidence means faster decisions, and in healthcare, delays in evidence can directly affect patient outcomes.
We also explore what Alex describes as the "analytics gap," the disconnect between where data exists and where insights are actually generated. Today, much of the evidence used in drug development still comes from limited datasets, often from a single country or region. Yet the treatments themselves are global. That mismatch creates blind spots, particularly in low and middle-income countries where data is often unstructured, fragmented, or simply not accessible. AI has the potential to standardize and unlock that data, helping to create a more complete and representative view of patient populations worldwide.
Of course, the challenges are not just technical. Trust, governance, and politics all play a role in whether data can be shared and used effectively. Alex is clear that the biggest barrier is not the science or the analytics, it is building trust between organizations, governments, and communities. Without that, even the most advanced AI models cannot deliver meaningful outcomes.
This conversation also touches on the importance of collaboration, not just between healthcare organizations and technology providers like SAS, but across the global ecosystem. Alex highlights how partnerships, open standards, and shared frameworks can help close the analytics gap and accelerate progress in areas like HIV prevention, where understanding real-world patient behavior is critical.
As we wrap up, one message comes through clearly. AI is not a miracle solution, and it will not transform healthcare overnight. But when applied to the right parts of the workflow, especially around data preparation and evidence generation, it can create measurable, meaningful change.
So, as healthcare leaders look to move beyond pilots and into real impact, the question becomes, are we focusing on the right problems, and are we ready to open up the data needed to solve them?
Useful Links
OHDSI – Observational Health Data Sciences and Informatics
Please check our partners of Tech Tech Talks Network
Learn more about the NordLayer Browser
[00:00:00] This month I'm partnering with NordLayer and it's support like this that allows me to keep bringing you conversations from across the global tech community. And if you are running a business right now, you may have noticed there's a quiet shift happening. One that most people are still underestimating and that is your company doesn't live inside your network anymore. It lives inside the browser. That's where your SaaS apps sit.
[00:00:28] That's where your data moves and increasingly that's where attackers are focusing their attention. So NordLayer has just launched its new business browser and it's designed specifically for small and medium sized companies that need visibility and control without the overhead of enterprise security tools. What I like here is the balance. You get advanced protection, better compliance and full visibility into how your team is working online.
[00:00:57] But without slowing anyone down or forcing them to learn anything new. It feels like a practical step forward rather than another security layer that adds friction. So if you want to see more about how it works, please head over to NordLayer.com slash browser and check it out and let me know your thoughts. But now on with today's show.
[00:01:18] What does it really take to turn healthcare data into better patient outcomes and on a global scale? This is exactly what we're going to be unpacking today live from the show floor of the SaaS Innovate event here in Texas. And my guest is someone who works right in the center of this very challenge.
[00:01:44] His name's Alex. He brings a fascinating perspective because where many people hear AI in healthcare and immediately think about diagnostics or drug discovery, his focus is on something just as important. Real world evidence generation. And he is a professor at the University of Cambridge and also works with Gilead Sciences where he is responsible for medical evidence generation that supports the entire drug development life cycle.
[00:02:13] So his work is all about making sense of the messy fragmented healthcare data from all around the world and then turning it into insights that can genuinely improve patient care. So what I'm talking about here is everything from HIV prevention in low and middle income countries to the challenge of trust, governance and standardization across the global healthcare systems. Because Alex is helping close what he calls the analytics gap.
[00:02:42] So today we're going to talk about where AI is making a measurable difference, why data cleaning might be one of the biggest untold stories in healthcare innovation, and why trust, not technology might still be the biggest barrier to progress. But enough scene setting for me. Let me officially introduce you to my guest live here on the show floor at SAS Innovate. So thank you for joining me here.
[00:03:10] Can you tell everyone listening a little about who you are and what you do? Alex Asim is my name. I'm an object professor at Cambridge University and I'm working for a company called Gilead Sciences. And in Gilead Sciences, I'm responsible for all the medical evidence generation that supports drug development life cycle. And there's so much I'm excited to talk with you about today. Because when people hear about AI in healthcare, they often think about diagnostics or drug discovery. But your work focuses on very real world evidence.
[00:03:40] So for listeners unfamiliar with that world, what problem is it that you're solving and why does it matter so much right now? Alex Actually, AI can be used in real world evidence generation. What I mean by real world evidence generation is we have clinical trials, which is like put people in the laboratories and experiment with clinical trials. But real world evidence is actually what the evidence we generate from data that comes from real world, which is not a laboratory. So AI can be used when it comes to cleaning the data.
[00:04:10] That's where it really plays a bigger role in cleaning the data. Because remember, this data we collect from the hospital is not collected for the purposes of research. And therefore, it's very, very messy and takes a lot of time to clean it using manual processes. So AI really can help in automating the cleaning. And that's where I really cleaning the data takes a lot of time. So with AI input, it should reduce the time it takes to produce evidence.
[00:04:38] And that evidence, the one which is used for decision making. And of course, any evidence delayed, it has an impact on the outcome of patients. And obviously, we're talking here at SAS Innovate. And you're talking about accelerating real world evidence and closing the analytics gap worldwide. So what does the analytics gap actually look like in practice for healthcare organizations today?
[00:05:01] So the gap really is about when we generate evidence within the pharmaceutical industry, we always use data we license. And that data is really a single data source. Most cases come from one country or one institution. And as a result, it takes a long time to clean the data, execute the analysis and have the results. So in one country. But when we drop the drug, the drugs are for the global world.
[00:05:30] They are for the whole world. We don't drop drugs for one country. So when we generate evidence that we want to see an impact or safety of a drug, we want to see the safety of the evidence from the whole world. Right? Not just one country. So we have to generate evidence from multiple countries. Right?
[00:05:49] So when we are talking of a big gap, mostly since we have been doing analysis in one country, now with AI, we should be in a position to actually generate evidence across the globe. Because you have standardized the output. You have standardized the processes. And that should make the analysis much, much faster. And in areas like HIV prevention, I think speed can directly impact outcomes.
[00:06:14] So how does better access to data and stronger analytics, how do these things help your teams make faster and more confident decisions? That's a good question. So HIV is predominantly in low to medium income countries. And in those countries, they speak lots of languages. So when we are analyzing data, we have to make sure that the data is standardized.
[00:06:38] So at the moment, most of the data we use for medical evidence is coming from English speaking countries and mostly in Europe and the US. So in low to medium income countries, they have a lot of data and most of the data is unstructured because of maybe not digitized.
[00:06:55] So with the use of AI, if you can go there and help standardize that data using AI, then it should be very, very easy to go and analyze the data from those countries and therefore generate evidence that's reflective of the global perspective of how HIV is. So for example, there's the Lopinavir launch, which has now started going to market.
[00:07:18] People in low to medium income countries have started to get that drug, but we don't have a capability to actually generate evidence to understand who are the people who are taking it. What's the safety impact on those patients? So, but we hope that with the introduction of AI, maybe there'll be some effort to really use AI to start getting that data ready so that we can do the analysis to understand the population.
[00:07:46] And the word game change is often overused, but that certainly feels like one right there. And clinical trials will tell one part of the story, but real world evidence often shows what happens after treatment reaches real patients. So how important is that broader view for shaping prevention and indeed future care strategies too? As I've said, yeah, clinical trials are at the sphere of evidence. They are the top sphere of evidence. But remember, it's an experiment.
[00:08:14] So doesn't why real world evidence plays a bigger role after the experiment is you start to understand really what's happening in the general population. So I hope that with access to the data sets that we have in those countries, then we are able to understand who are the patients benefiting from these drugs, who should we target to get the treatment so that they don't get HIV. For example, Cicillin Capriva is for prevention of HIV.
[00:08:40] So it's useful for targeting and understanding the populations in those countries. What would you say are the biggest barriers that is stopping healthcare organization from turning huge volumes of data into meaningful action? Is it that fragmented data that you mentioned there? Is it technology, process, trust, culture, or maybe a mixture of all those things? It's actually a mixture of all those things. I've traveled around many countries.
[00:09:05] The main issue is in Europe and the US, the data governances are kind of established. But in low-medium income countries, the data governances are not established. And there's a lot of politics. And there's lack of trust, especially if you are from the industry, pharmaceutical industry. There's lack of trust. People believe you are stealing their data. They don't understand why you need their data. Okay? The trust is really the main issue, not the analytics. The science is fine. We know we can do it.
[00:09:34] But it's really the governance, the politics, and the trust. Those three things are very important. How it has to be done, it has to be a collaborative effort across the globe so that we build the trust that the data we use is actually to help them. We use the data to help the patients themselves to improve their outcomes. But yeah, with politics, that's a big issue. And obviously, you operate on a global scale.
[00:10:00] So how do you approach evidence generation across so many different healthcare systems, regulations, patient populations, while also maintaining consistency and trust? Again, I would imagine it's quite difficult, but getting that balance right. That's why it's very important that every evidence we generate, we can demonstrate that it's reproducible and is of high quality. And transparency on how we do it is very, very, very important. So we make sure that we standardize everything that we do.
[00:10:28] We make sure that the team that reveals everything we do is in the right way. And then, of course, the data governance is always the key. And we have lots of guidance on how to do all these studies. For example, the EMA European Medicine Agency, the FDA, they have all those guidelines that we follow to make sure that the way we conduct observational studies, which is leading through wild evidence generation, is consistent across.
[00:10:54] And then, of course, you have reviewers after you have produced the results to see whether what you are saying is accurate. And although we're talking about technology here, and this is a tech podcast, as we've mentioned several times, trust is everything in healthcare. So how do you ensure AI and analytics support human decision-making without creating any risks about bias and operating compliance and transparency and all those things? Again, massive topic, isn't it? It's a very good topic.
[00:11:22] It's an interesting one, especially when it comes to bias. So AI relies on the data. If the data is biased, AI is not going to remove the bias. But we can apply epidemiological approaches that reduces, minimizes the bias. So the data is really the driver. So AI, we use AI. I believe AI is only going to be used as an aid. The person using AI is still accountable to what really is coming out of the studies themselves.
[00:11:53] That's how it works. I think, I know people are talking about a human in the loop. Yeah. But to me, I think it should be AI in the loop because the person is using AI to help him or her produce themselves. Makes perfect sense. And we hear a lot about AI transforming healthcare. But where do you see genuine measurable impact today rather than just experimentation, hype, demos? Where do you see that measurable difference happening right now?
[00:12:21] So AI is not going to do A to Z. AI, when you're looking at the workflow on how you generate evidence, for example, from A to Z, there are some bits where AI really can reduce the amount of time to generate evidence. So if I give you an example, it takes around six to one year to just conduct one study. Yeah.
[00:12:42] But 80%, 70% to 80% of the time is spent on cleaning the data and some workflows around code generation and all that kind of stuff. So if AI can help people generate the first code, the analytical code, and then the validation and the tracking, that reduces the amount of time that you would spend. So ideally, or I hope that it should, instead of six months, you should do it in three months.
[00:13:09] That's doable because most of the work is about really routine work, something that can be reused. So if you can do it first time, then the next time should be much, much, much, much easier. So that's going to be possible. I think most of the pilots that we have seen is because people just look at a bigger pilot instead of looking at sub bits, you know, the bits of the workflow and see how you can reduce it rather than the whole thing.
[00:13:35] Because when you have a pilot, you're looking from A to Z and A to Z is not going to work, but those bits should work. And if you combine those two many bits of the workflow, then you reduce the amount of time. And looking at your work in innovation and partnerships, what role does collaboration between tech providers like SAS and healthcare leaders like yourselves play in making some of these outcomes possible? Because I would imagine you can't do it all. It takes a village.
[00:14:02] Tell me more about those partnerships and what that delivers. Industries, I've worked in many pharmaceutical companies as well as Enica, Best, Gilead and all those ones. We hire the best people. Yeah. But it doesn't mean we hire everyone. And when we hire best people, they focus on certain work. But I also believe that most innovation takes place outside, especially with the academic institutions and companies like SAS and all those ones.
[00:14:29] So, if industry, we can focus on what we do best and then we collaborate with external, then we can deliver something big that normally you wouldn't deliver on your own. So, really, that's the reason why I actually introduced partnership within the company so that we can collaborate and where necessary we co-create. Because co-creation is really a big thing. We can share. We have all different skill sets. So, co-creating things after results in having something faster and something innovative.
[00:14:59] And what brings you to SAS Innovate this year? Are you doing some speaking here? What brings you here? Yeah, I had a very good talk yesterday. The talk was about use of real-world evidence and really how transitioning from the old SAS, which was like you'll have one study, you do that study and you're done, you do that study and you're done with in the old SAS, you do that study, you do that study.
[00:15:21] Kind of a good thing with it is you can just talk to different languages, no longer just talking about SAS, R, rather you can now talk to SAS, R and Python. And those are languages which are used by academic institutions here because they are cheap. If they can't afford SAS Viya, they can use R, they can use Python. Those are cheap.
[00:15:43] So if we are talking about collaboration, actually, if I do work within my company, which is a SAS Viya, which is expensive, but I can send that work to an academic institution and they just change to R or Python and then do the analysis, which really helps us because previously you couldn't do it.
[00:16:02] So you'd have work done and then you give it to an academic institution, they would have to repeat the work, which takes a lot of time. But with SAS Viya, that's very, very important. And with Gilead, we do work with a lot of people external and those people don't have the same infrastructure that we have. So, but with SAS Viya, it helps with doing the analysis when you don't have the same infrastructure.
[00:16:26] And we've talked a lot about the importance of partnerships. And I would imagine when you came off stage, there would have been a few people wanting, well, giving you feedback, sharing ideas, maybe it sparked some ideas. And then from those conversations you've been having with people here and learning from people here, anything stand out? The kind of conversations you're having with people here? Yeah, it's amazing. Everyone is excited. Everyone is looking forward to see the change. Maybe there's a lot of optimism.
[00:16:53] Yes. Um, but I think the future is bright. If we do it right, uh, with the technology that we have, I think we should be, especially from the perspective of medical evidence generation that helps, you know, patient improve patient outcomes. I think the future is bright. We can do a lot because everyone is really looking forward to do a lot. Yeah. People are talking about the data governance as a big concern.
[00:17:18] Yeah. So that's not a surprise. We're talking about how to use, how, how can you work with AI to reduce the bias because that's also a big concern, but those things are with good guardrails, they can be solvable.
[00:17:33] And if you were advising any healthcare listening today, anywhere in the world who want to close their own analytics gap, where do you think they should start first to create real impact rather than another pilot project? Any advice that you would pass down? Cause you're, you're much further along than many people listening. I would imagine. All I can advise, especially to those people who have healthcare data. Yeah.
[00:17:54] The data that they have in their system can make a big difference to improve patient outcomes around the world. All they need is just open up, join organizations that can help them clean that data and make it available for research purposes.
[00:18:11] Health data is very important and is always going to be used in a very effective and more well-governed way. So data, no one will always have access to your data. All we need is can we generate evidence that that evidence then can be used to make decisions, make physicians use it to prescribe medication and also patient understand what they are all about.
[00:18:37] Well, that's really what's important. So data, I'm just focusing on the data because that's really the bigger thing. We can do experiments very well. Companies have money to do experiments, but we don't, without data, we can't understand what's happening in the general population.
[00:18:52] Now, if we don't get the data, if people don't open up to access the data, we're always going to be using the same data we have always used before, but that data is biased to those people who are willing to share their data. And then we lose out to those people who are not sharing the data. So evidence we generate is going to be reflecting those people who have shared their data. For those who have not shared their data, we will never know what's happening. So we just want to go.
[00:19:20] So my advice would be, please come up if you have the data, work with people who have capability to clean it and make it available for research. There are guide rails on how to use it in terms of privacy.
[00:19:34] And we're recording this on the last day of SAS Innovate. So when you head to the airport to get that short flight home for you, I believe, when you take into account all of the announcements, all of the conversations that you've had around here, anything you're going to be reflecting on and thinking about as you take that journey home?
[00:19:53] Yeah, there are lots of things. I missed the tornado. That's number one. Something I've never seen. I thought I would say it, but I missed it. But in general, it's good to see the topic of AI being the main topic on how to use it. It's something that's going to change the world. I know there's a hype on it.
[00:20:14] We see those hype whenever something new that has come. But with this one, I think there are some good guide rails already in place that we shouldn't worry so much on really whether it's going to take people's jobs or things like that, because that's not going to happen.
[00:20:30] Now, if you have listened to most of the conversations or listen to some of the sessions, you can see everyone is talking about, yes, we can use AI, but there's this X, Y, and Z that may be, but those X, Y, and Z they're talking about, they are servible. They are solutions to actually make it work. So I'm just going home thinking that even though we have some problems, it's not just for one company, it's a global problem.
[00:20:57] And with a global problem, if people can come together, they will get solutions to really solve those problems. And maybe in five years, we should be able to apply AI in all our routine work. But most importantly, also, people need to be trained on using AI because we can talk about it. But if you listen to most people, they actually don't know where they can use AI and what AI can be used for.
[00:21:23] So maybe upskilling people to understand what AI is, not just giving them co-pilots and all that stuff. There has to be a mechanism of training people to understand how to use AI. Now, if people know how to use AI, then they will start thinking about, okay, where can I apply it? Maybe those pilots, I think there was someone who said 95% of the pilots have failed.
[00:21:48] And this is why they have failed is because people are just thinking, oh, AI, so let me apply it without thinking of actually where can AI help me? And then just focus on where can AI help me? Right. So maybe if people understand AI, then their pilots will focus on those little bits where AI makes a big difference rather than shoveling in everything and think AI is going to produce miracles. AI is not a miracle maker. 100% with you.
[00:22:16] And I think that is a powerful and thought-provoking thing to finish on. But for everyone listening, I will add a link to your LinkedIn on the show notes so people can carry the conversation on with you. Anywhere else you'd like me to point everyone, link to your website, et cetera? No, it's on your LinkedIn. And also, Odyssey is something that we didn't talk about. Odyssey is observational health sciences. Odyssey is an open source, which is comprised of most academic, but there are people willing people.
[00:22:45] Outside pharmaceutical companies, it's an independent body. That's a place where there has been an effort to make sure that all the data is made available. Anyone who has data, they help them convert it into a standard format so that when you want to generate evidence around the world, those people can help you. So maybe that's another place that you can, Odyssey as OHDSI.org.
[00:23:14] It's a good place, very good people, smart people that can help anyone who has data, especially in the hospitals, that they can help them clean their data and leave it with the academics or whoever owns it, but make it available for research purposes. So that we can generate all the observation studies we want. Regardless of whether you're in China or Brazil, the data is available. That community is there to help everyone to convert their data into a common data model.
[00:23:41] Well, I would urge everyone listening to check out the show notes to this podcast, wherever they're listening. There will be a link to everything you mentioned there, including your LinkedIn. And I urge people to reach out and keep this conversation going. But a big thank you to you for joining me and starting that conversation today. Thank you. Okay. Thank you so much. I think when we talk about AI in healthcare, it is very easy to get distracted by the shiny headlines around automation, futuristic possibilities.
[00:24:07] But for me, today's conversation was a powerful reminder that real progress often starts with something far less glamorous. Things like better data, stronger trust, and the willingness to collaborate. And I think my guest shared why real-world evidence matters so much, especially when treatments move beyond clinical trials and directly into the lives of real patients across many different countries, languages, and indeed healthcare systems. And I think his message was clear.
[00:24:37] AI can absolutely help accelerate outcomes, but only if we solve the human challenges around governance, trust, and access first. Because without that, even the best technology can fall short. I also love his perspective that perhaps we should stop talking about humans being in the loop, and instead think about AI being in the loop. Because accountability must always stay with people. And that feels like one of the smartest ways I've heard this framed all week here at SAS Innovate.
[00:25:06] So a big thank you to Alex for sharing his insights and helping us better understand how analytics, AI, and global collaboration can improve healthcare outcomes for everyone, not just the few. And for everyone listening, I'll leave links in the show notes so you can connect with Alex on LinkedIn, learn more about the community he mentioned there, and also a question for you. How do you see AI helping close the healthcare gap where you are? I'd love to hear your thought. You can get a hold of me at techtalksnetwork.com. There'll be links to Alex as well.
[00:25:36] Let's keep this conversation going. But that's it from the show floor here at SAS Innovate. I'll be back again tomorrow with another guest. But thanks for listening as always, and I'll speak to you again tomorrow. Bye for now.