Trials with Maya Z

Hello everyone. Welcome again to Trials with Maya Z. I'm your host, Maya Zlatanova. And today I have an awesome guest. His name is Stefan Schroeder, he's actually the CEO and Co-Founder of EnRoll With AI.
So, there will be a lot about patient recruitment and AI at the same time. But before we start with our conversation, Stefan, tell us a little bit more about yourself and your background.
Yeah, hello. Thank you, Maya. Thanks very much for inviting me here I'm very happy to be here and I'm in good company. I've seen several of your podcasts and I think the guests have had very interesting, provocative, and thoughtful conversations. So, hopefully, I can contribute to that as well with this kind of perspective, but yeah about me, so, I've always loved science.
I think the king himself, David Attenborough, really drew me into loving and having a deep appreciation of the natural world. And through that, I went on to study biology at university, not necessarily specializing in any particular area, but always love that kind of deep understanding of science.
I studied at the University of Manchester and they were very flexible in the modules you could look at. And I studied everything from, you know, bioethics to molecular genetics to neuroinflammation to parasitology. And I think through that, I just really loved science, but I was kind of a creative person, and I really wanted to tie in my kind of love for science and my creativity and fortuitously at my graduation, the kind of founder and CEO of Langland when it was an independent agency, Phil Chin spoke at my graduation and that was a kind of a bit of a eureka moment.
I spoke to him afterward and ended up working at Langland as a first job, and they were a very creative kind of healthcare advertising agency, creative kind of communications agency that also looked at a problem and developed a solution, be that with communications or be that with a digital transformation as it were.
Langlands also were a brilliant kind of place for me because they also focused on clinical trial recruitment, which I guess 10 years ago was very heavily kind of targeted within CROs kind of looking at clinical trial recruitment and actually having this creative agency look at problems in a unique perspective was an amazing place to be. I was lucky enough through working there to transition over and to live in New York for three years, expanding the kind of clinical trial recruitment business over there with them. I was always on the client service side. So I worked with them, various different clients across the world and in recruitment campaigns for pretty big, big status kind of compounds, you know, the likes of aducanumab, which is obviously a huge, huge potential target for Alzheimer's.
And, you know, it was really, really interesting. Then, I've bounced around and worked in various different large kind of multinational, communications agencies, always within healthcare, communications, and digital transformation. And sometimes in a niche within clinical trials.
So worked a lot with the team at Havas phase. Um, who I think, do a phenomenal job in Making sure that patient experience is kind of at the forefront of everything that is done and they have a really intelligent referral management platform, but through that, I became a bit disenchanted with the clinical trial space and realized that it's just a bit, it's a bit broken, to be honest.
And we're probably diving into trials straight away, diving in the deep end, but, you know, there's data to show that up to 85 percent of trials fail to recruit on time, there's data to show that the trials still have a vast majority of white participants.
I think 70, 77 percent of white participants in the U S. They're not representative, you know, they're not diverse. We're not recruiting on time. They're not necessarily working in the best favor for sites, for healthcare professionals, for patients. There's a lot of problems and all of that has fed into actually, you know, stagnation within drug development.
I think there was a study within McKinsey that the number of kind of US novel drug applications has remained stagnant, I think there's an average around 43 a year, you know, and that's remained stagnant for the last kind of four, six years. So the industry as a whole is kind of flawed.
Not for one of a lot of people doing great things, you know, and really wanting to work hard, in the best interests of patients, of healthcare professionals, of sites, of drugs, you know, and of the end kind of individual and getting drugs to market. But I think there's just a lot of opportunity. To do something different or to evolve the industry that we're in.
And I think through my background within working in these interesting, fast-paced environments where strategy is such at the forefront, I think I've kind of wanted to see and try and champion things that can change the industry. And this is where AI, over the last, year and a half, especially, since the launch of chat GPT, which was in November of 2022, but last year really gained momentum has really brought, the capabilities of AI to everyone.
Of course, artificial intelligence has been something we've been speaking about. We have in films, you know,we've been utilizing in drug discovery for so long, but really now chat GPT, I really think open AI opened the gates to us being like, wow, this is a transformational technology.
What can we do? And because I've seen that, and also I've seen the capabilities of various different technologies. I think that that's what I've built kind of in role with to champion those tools, to try and really change an industry that I think really needs it for everyone involved.
Stefan, before we dive into the AI world, because that's one of the things that I would definitely like to further discuss with you. I want to understand, you said something absolutely true. That we work in an industry where every single stakeholder actually cares for their job because their job has a real impact on people like healthcare, even very often when I'm speaking about clinical research, I'm not even giving an example with other people.
But like even knowing that me working clinical research impacts my children's lives. My grandchildren's future and so on and so forth because we hear we speak speak about how we define health care, but I agree with you let's say the majority of people working in clinical research Definitely care and realize that this has a tremendous impact, but then you mentioned that clinical trials are broken.
So can we start actually? You named a few numbers. We've all heard these numbers. But where does the problem actually come from? From your perspective.
Yeah, it's a very poignant topic, right? I think, as you say, clinical trials is such an interesting area, such a relevant area, such an emotional area because of the potential for it. And especially when you're working within phase three trials, which a lot of my kind of career has been focused on, like there's true potential in these compounds, in these therapies to change people's lives, not only in the future, but actually on the trial.
Yeah, yeah, exactly.
Because of the way trials are run nowadays, it's very infrequent that you get a blank placebo. So even if you're not going to get this potentially revolutionary drug, my use of potentially will disperse this conversation because I'm so used to having to have that in communications,
But, you know, even if you don't get this potentially revolutionary drug, you're going to get standard of care. You're going to get free healthcare. You're going to get kind of monitoring, your journey and your experience is going to be positive, hopefully, but I think, yeah, to go back to your question, where I think the challenge lies is, I think for me, trials is the precipice between the medical world and the human world in a way, or from between science and like humanity.
It's really the transition from when you have scientific rigor, scientific studies on cell lines, hypothetical models, and that then transitioned to reality, to humanity, to our everyday lives. So you go from this almost in vivo process, if you will, to kind of outside of that.
And I think it's that gap and that jump that causes a lot of challenges. I'll give you an example, protocols, clinical trial protocols. Probably everyone who listens to this podcast has seen one, has reviewed one, you know, and they're quite exempt of human touch.
And we talk about patient centricity and looking at and looking at protocols, but I saw Someone on LinkedIn talk recently about a visit to SCOPE. So obviously the SCOPE clinical trial conference, huge in Florida every year. And, they were posting about this revolutionary advancement within clinical trials.
And the revolutionary advancement was 'finally, we're putting a protocol and we're simulating in front of patients and getting market research from patients before we start running the trial'. And I'm sorry, but if that's kind of the revolutionary advancement that we have in 2024, that we're kind of doing market research or testing a protocol with people before we run a potentially billion dollar trial, there is something wrong there.
Like clearly that gap between people and science is not being bridged appropriately. I mean, we run market research for like Coke versus Pepsi or what flavor M& Ms are preferred. So can we not do that on a clinical trial protocol and spend a bit more time looking at that and evaluating that to see if it works for everyone involved rather than trying to truly see if that drug candidate is most efficacious.
Of course, that's important. That's the primary goal, but you know, people are in that journey and that's really, really important. I think that's quite a fundamental kind of consideration that's necessary.
And Stefan, where do you see, by the way, I love your analogy with the Coke versus Pepsi and like the market research and like hundred percent like here, though I spent quite some time in the space. I've been researching, protocol development for protocol design feasibility. These are all areas that I'm working on a daily basis.
And even though it sounds very simple, actually, it's not that simple. There are processes that are, let's say, today's status quo. And so we need to change the status quo. We need to change the mindset of people before we even go into something more sophisticated. But yes, I agree with you. Come on, guys, we are working on billion dollar projects here.
We need to do a proper customer discovery, market research, whatever you call it, before we actually spend this money. So, agree. But where do you see AI, actually, in the whole, because one thing you said is that you have this background in patient recruitment. Then you see where like one of the biggest pro, like there are multiple problems, but you see this problem with how we develop the protocols.
That was the main problem that you mentioned. But also you want to champion these new AI applications. So, where do you see AI playing a role in making clinical trials more efficient?
I think it's interesting hearing what you say there because there is a status quo in place, and I think that that also is something that we need to challenge. I think that in trials we're so frightened of technological advancement, you know, and any potential kind of ethics review or challenge that we kind of limit ourselves.
Interestingly, again, at SCOPE as well, and I'll talk about some AI advancements eventually, I promise, but at SCOPE as well, the FDA were present and the FDA said, we actually have an open door policy. If you want to talk about AI with us and utilizing AI in clinical trials, we have an open door, we have an email address, we have a phone number, you can call us and see if we think it's valuable.
At the end of the day, the FDA said we want patients, you know, patients health is our forefront, you know, that's our key consideration. So if there is an artificial intelligence tool that can help that fantastic, but let's talk about it. So I think the status quo needs to be challenged a lot.
And I think that that probably is a challenge that we have is that we're moving along, still defined by principles that have been set feels like years ago and actually look at how quickly the world is moving. I think I saw today for the first time in our history, we truly do not have a clue where we will be in 20 years time because technology is moving at such a fast rate. But diving into specific AI tools, I mean, it's hard because artificial intelligence is a bit nebulous. it's easy for me to say, we help you use AI for clinical trials, but I kind of liken that to saying we help you use the internet for clinical trials, right? It doesn't really truly mean anything because artificial intelligence is a kind of a methodology, is a technology that underpins certain tools.
So, where there is a problem, there is a solution. Probably an AI solution. However, looking at particular tools and particular companies and organizations, there's so many across the board from a protocol standpoint, for example, evaluating a protocol we actually built using chat GPT four and the GPT agents, we built a kind of a clinical trial evaluate protocol evaluation tool, where you input the NCT number of your trial, and it will evaluate it from a kind of a patient HCP site experience standpoint, and evaluate like how positive or kind of areas for improvement there are for this trial from a patient perspective.
So, something as simple as that is a really, really quick and fast tool that you can utilize to start advancing your trials and that impact of using that, you have a minor change that means that you don't need a protocol amendment later down the line could save so much money.
Also increasing the experience of people, but it's difficult to kind of isolate individual tools because I could talk forever, but again, looking at the initiation. So, let's kind of split out this conversation into looking at tools for the initiation, kind of the recruitment, enrollment period, and then the actual trial period.
I think that we can focus on that. And there's so many tools that we can talk about, but I think breaking it down in that way kind of helps and I can communicate.
I agree with you. There are many areas, but is there one or two areas where you see that AI will have the biggest impact?
I think initiation is a really interesting area looking at the protocol, looking at site feasibility, looking at site locations. I was working with kind of a partner called Pearl AI and they're an organization that they don't focus on clinical trials, but we've been talking about how we can adapt it for clinical trials.
And what they do is they basically have billions of data points that they can map, but it can map demographic data on top of behavioral data to understand the best places to open sites for a trial. So you're looking at the key demographics. What kind of key therapy areas are, you know, within those different regions, so we can focus where a site is, but also the behavioral data is really important.
So of course you might have a condition, but are you primed and wanting to go on a clinical trial? You feeling like you are a bit of an innovator, or an early adopter. And it's those people who are important to target with the condition to then go on to a clinical trial. The same if you're looking at healthcare professionals in that region, you know, those individuals who are going to promote a trial that are a bit more forward thinking in that space, and therefore using this tool, you can map all of these data points to then, focus a trial in a location.
Again, evaluating protocols, there's a British market research organization that I have been speaking to called Day One, and they do a really fantastic job of looking at synthetic data and looking at kind of using large language models and such as CHAT GPT and Claude and so on to deliver market research.
We all know within clinical trials time is of essence, budget is of essence. And also, how do you recruit patients for market research that are hard to recruit for the trial themselves? You know, you're trying to get the same cohort people to research and go through the protocol that you would on the trial.
And that's a challenge, but through using the likes of GPT and Claude, all of these different kind of large language model tools. You can actually gather data to analyze protocols and to deliver market research. There's an interesting case study where they did it in a late stage oncology patient group.
And they asked a variety of questions, which they already had asked true individuals with that sort of condition. And, they built a discussion guide and prompted the LLMs appropriately. And actually the output from the LLM and versus the output from the patients was pretty much exactly the same, exactly the same in terms of themes.
So everything that you need. To change the protocol or change the way that you run the trial actually is there if you're looking for specific quotes, or specific experiences, they weren't quite as refined, but if you were presented with the output from the LLM, that would have a very positive influence on the way that trial, the trial was planned or recruitment was planned.
And so that's another example of utilizing a tool in that initiation phase.
Let me stop you sharing and ask a few more questions around there because, I agree with you and I've seen firsthand how these LMS can support, will empower, let's say the old fashioned market research and make it more actionable. Let's put it that way. However, I've also witnessed that in many cases the teams and the people who are behind the protocol design are not, and the protocol development afterwards, are not necessarily people who are used to actually analyzing feedback coming from patients.
So, I think that here, it's not only about having this information. But it's how you translate this information into what it means for the protocol at the end of the day. Because it's one thing a patient telling you, Hey, I'm suffering from nausea every single morning. But what this means for the, for the design and the protocol, that's actually the big question.
Yeah, yeah, yeah, I completely agree. It's making sure that, you know, the insights that are gathered communicate clearly to evoke positive change within the protocol, for example, is there a site visit that requires you to be present at the site for eight hours a week? Or is it necessary to be there for one hour every four weeks, you know, and those subtle differences, although they might not seem huge, and for certain individuals might be different, but for different therapy areas, that could have a profound impact on kind of freedom, flexibility, and movement.
And those subtle changes in the protocol can be on earth from, you know, talking to these people at the end. So yeah, I agree with you. We do need to translate that into the protocol or also into the communication strategy, you know, into the recruitment strategy, into how we liaise with sites.
But, we now have an ability to gather this sort of detail and data rapidly. So, how do we now weave that into the flow? Right. And of course, kind of why I've kind of built this, this consultancy is to bridge that gap, show the opportunities so that no longer is the roadblock actually we can gather this here.
So now how can we start to evoke change there? So with your company, Stefan, Enroll With AI, you're consulting different companies how they can leverage AI for the sake of faster patient recruitment. So, can you give us any examples how AI actually supports patient recruitment outside of the protocol development? Are there any cases that you can share?
Yeah, absolutely. Protocol like valuation and that insight gathering is just a very small section of the capabilities of AI. Before we actually talk about patient recruitment, I think there's a really interesting idea and some tools to fit that also can change the shape of kind of how we recruit patients as well.
So, I was speaking just last week to the founder of Veil AI, they actually worked with Bayer and what they do is they've built an AI based kind of engine, which helps develop synthetic data, helps mass anonymize, and develop, and then generate synthetic data which is very representative of a population.
They actually worked with Bayer and analyzed vast swathes of data for a particular trial, which then meant later down the line, when it came to recruitment, actually there was 20 percent less people required to recruit because you can extrapolate from the data, anonymize that data and generate synthetic data.
So before we even talk about recruitment, there's technology available that can expedite the recruitment process by reducing the amount of people that we need, which can be fantastic. Of course, I want people to have access to these potentially breakthrough medicines. So having people on trials is positive.
I don't want that to go away, but if you can streamline the process, especially reducing the need for people on a placebo arm. You can do with this or, you know, develop a digital twin with AI, which is another capability. And that's also really, really interesting advancement that we need to consider as well, with 20 percent less people.
That is huge, if we want to talk about not only the human impact, the economic impact is significant. Yeah.
Fully agree. So-called silico trials, there are quite a lot of companies. Actually, I don't know how many, but I know about several companies that are really fighting on this battlefield for minimizing the need of patients and you're absolutely right. If we need less patients, then patient recruitment happens faster.
And then the retention is also better. So yeah, yeah. That's very interesting. And when you work with different companies to consult them on AI integration and adoption, how do you feel, Stefan, are companies open and ready to integrate AI? And are there any areas that where they're more open and less open?
So, I've experienced a whole range of different, acceptances, acknowledgements, or stages where particular organizations are on their artificial intelligence journey. Drug discovery has been an area of focus for artificial intelligence for some years.
And actually, I would be surprised if you could find one that wasn't using AI to underpin that drug discovery. So there is acceptance and uptake of artificial intelligence in different areas. However, when it comes to recruitment, I think there's challenges, you know, I think one of the problems or one of the areas that we're in right now is that actually because of this huge advancement and AI becoming this kind of buzzword is that a lot of sponsors and CROs, Bringing in large consultancies to try and advise them on where they can start utilizing AI in their practice.
Not just within clinical trials, but within pharma, within communications, within digital innovation as a whole. And, you know, those evaluations are therefore meaning people on the ground from a clinical trial perspective are waiting to hear what is going to change and how. And therefore, actually adopting some more tactical tools is less pressing as it might be. So that's something that we're experiencing. However, you're also having, the likes of you seen Pfizer have released their own kind of large language model internally, advancement there, you're seeing AstraZeneca have got Evanova, which is a unit focused on clinical trials and AI, they have different sections within their business, so we're seeing a lot of organizations pick up artificial intelligence because it can improve productivity, can improve efficiency. However, the reason that we exist is because I still believe that further down the line there is going to be this gap, as I've always experienced within trials, that tactical, innovative technology is not necessarily adopted for kind of recruitment, patient experience and further down the line we're only just really grappling the use of social media to help communicate and recruit patients.
So how are we gonna adopt artificial intelligence is a challenge. However, that's kind of our challenge to discuss with clients. And, we've built a database of hundreds of different AI tools specific for clinical trials, also specific for health. But then can be used in trials, but then also specific for just consumer that can then be utilized for clinical trials.
And I'll definitely, at some point in this conversation, list off quite a few groups of different tools that I think are useful to explore for various different people at different stages. But, yeah, I'm seeing and we're experiencing kind of adoption, but also you run a clinical trial, you're trying to engage patients.
Now AI is here. Like, how do you have time to understand what it means and what it is when you're hearing LLM and NLP and AI and, you know, artificial general intelligence, it's confusing. And that's kind of why we exist - to help support people along that journey and show them the true utility of these things and to not be afraid of it.
I think we've kind of scorned AI by calling it AI and it becomes this scary kind of nebulous thing when actually it's just technology that people have developed to help any other technology. Yeah. Well, yeah, it's powerful technology, but I understand what you're saying. Stephan, you mentioned that you mapped hundreds of different solutions and at the same time you see these companies trying to explore the world, like the AI world. So do you have any advice for these people that are exploring for new solutions, especially the ones who are using AI to achieve different solutions for different problems?
Do you have any tricks and tips? How can you assess a solution that you should spend time on versus one that's just using the AI as a marketing buzzword?
Well, get in touch with me and I'll help you out. That's a shameless plug. Yeah, it's a good question, but how do you kind of evaluate, you know, particular AI tools and understand whether they are kind of transformational or just because they have maybe like a chatbot that uses AI, you know, are they actually transformational?
I think, what I would suggest is don't use it for the sake of using it, look at where there are problems, or challenges, or areas of opportunity that you need, or require, or see some potential improvement, and then seek out organizations that you think are appropriate to fill that gap, because whether they're underpinned by AI, which a lot of technology is, is becoming or not, at least you're looking for a solution to a problem that you have.
Right. So I think I would start there. I really don't think you should sit and think, ah, everyone's using AI. How do we use AI and just jump onto some tool because it promises to be the be all and end all. I think you need to evaluate really where there are areas of opportunity. And clearly, like we discussed there are a lot of areas of opportunity, AI can help with recruitment. It can be very good at recruiting more diverse populations. It can help make sure that there is more gender equality in trials as well. Very, very important, it can support setting sites up in certain areas.
It can also help you communicate better and more clearly and more accurately from a client side, they'll be happy, but also more appropriately with certain people. I don't think using AI actually is going to remove the human touch. I actually think it will increase empathy.
Because it allows people to focus on the things that are truly human, which is human connection. And actually it's going to augment practice for us to be more present, but also more accurate. Humans aren't perfect. We're not perfect. We see that, I think there's like 800, 000 misdiagnoses in the US a year, which can lead to quite negative consequences. If we can bridge that gap, that's fantastic. Let's do that. You know, and within clinical trials, clearly, there are areas of opportunity. But it doesn't have to be AI,
but it probably will. It probably will be soon. But don't start getting scared because everyone's talking about it and you need to use it.
Actually, what are your problems, your challenges and, and start from there. That's where we start when we're speaking to people. Look at an evaluation of where there are particular challenges within the flow and then offer solutions. We offer solutions that are based in artificial intelligence, but, for example, patient communities can be a really good area to get an engaged group of individuals, who are interested in trials or interest in the therapy area and start to reach out to them, we've been liaising with a partner called Evidation.
They are underpinned by artificial intelligence to manage their community, but so we will recommend them for clients because I believe in what they're doing for one. And yes, they use AI as part of their practice, but is it like, would you traditionally call this an AI tool?
Maybe not, but actually I see the value, you know, and that's why we would recommend it or that's why I'm talking about it.
Yeah. People say fall in love with the problem, not the solution. That's actually what's important. Wow. Yeah, I'm pretty sure that many people will reach out to you and ask you for your AI guidance. But it's awesome what you're doing, Stefan. And, we need these pioneers and people that are exploring what's the situation out there and give us guidance.
Definitely. I have one last question, though. It's a question that I like to ask my guests because it provides me with this valuable perspective from each one of them of what will make clinical trials more successful, patient friendly as well.
So if there is one thing that would bring the biggest impact on making clinical trials more successful, what would that be? From your perspective.
I think it's a very good question. I mean, we're talking about artificial intelligence and one thing that I personally have kind of got really excited about with AI is that we might potentially be able to speak to animals within our lifetime using AI, so if it can do that, I think that It can probably increase the success of clinical trials in the long term, but saying artificial intelligence will help trials become more successful isn't really that kind of focused. I think, maintaining a level of humanity, a level of kind of presence, a level of understanding, of compassion, of empathy, I think those things are truly, truly important to making sure that this industry is successful. And so to drill it down to one thing, I think it would be empathy. I think that is really integral to this industry. We need to think of why this industry exists and it's to bring potential, you know, life saving, life changing therapies to people across the world.
And if we have empathy at the core of that, I think that we'll be moving in the right direction. Empathy can also ensure that when we're looking at technology like artificial intelligence that we do it with such positive intent. So I think empathy would be my response .
Thank you, Stefan. Yeah, we do need empathy. I agree with you. You also mentioned something that I want to mention once again and wrap up our conversation to bridge the gap between science and humanity. And maybe this bridge is exactly called empathy. Like you mentioned, thank you so much for bringing your perspective.
Thank you so much for sharing about the world of AI in clinical trials. Good luck with your company. And, yeah, next time, I hope soon we'll have another episode where we'll go through another round of AI applications for clinical research.
Yeah. Thank you so much. Yeah, I don't think I gave enough guidance within specific tools to use so there's always time for another episode, but thank you for the thoughtful conversation and appreciate your time. Lovely to chat to you and hope we get to speak soon.
Same here, Stefan. Thank you.