Andy York — 30 Years in Pharma Programming, AI Guardrails & the Future of Statistical Programming

**Adam Walker:** I'm Adam Walker, a biometrics consultant, and this is the Pharma Prescribed podcast, where leaders, innovators, and hidden voices in healthcare open up. No sound bites, no spin. Just raw insight, one prescription at a time. In an industry driven by data, protocols, and pressure, we rarely pause to ask the human questions.

What drives us? What breaks us? And what truths live behind the titles we wear?

**Adam Walker:** Today's guest is Andy York, one of the most respected voices in global statistical programming and clinical data science. Andy brings more than 30 years of experience shaping how the industry designs, validates, and delivers statistical outputs for regulatory submission.

His career began at Roche, and has since spanned senior leadership roles across major pharmaceutical organizations. Most recently serving as vice president of clinical data science at Novo Nordisk, where he played a pivotal role in modernizing data workflows and advancing best practices in clinical reporting.

Andy is now an independent consultant, acting as a strategic advisor at Verizion, contributing to their mission of accelerating clinical trial evaluation through AI, graph analytics, and next generation traceability frameworks. Andy brings deep historical perspective, technical authority, and a forward-looking vision.

I have had the pleasure of working with Andy twice across our careers, and I'm delighted to welcome him to Pharma Prescribed today. Andy, for those of our audience who are not familiar with you, who are you, and what is the mission you're

**Andy York:** on? Yeah.

**Andy York:** I'm a statistician by training. I actually, did my master's, whilst working, at Roche.

So I did it through, part-time. prior to that I had a degree in mathematics, stats and computing and OR. And, yeah, master of business, I think you'd probably describe it. But that's where I built my, interest programming where I learned languages like Fortran, which, nobody's probably ever heard of these days.

But, that's the way we used to do it in the old days. And, part of that degree was sixth sandwich course, and I had to spend a year in industry. And I actually spent a year working for a pharma company and, learning SAS and, got quite good at it within a year, and had a bit of an aptitude for it.

So although I'm a trained statistician, my preference is towards programming. Quite a few other people, I think, in the industry as well. But, I've got that statistical background knowledge, I kinda draw on.

**Andy York:** So why the industry? I've always had a life science interest, and it fell into place that, working on medicines for people is a really noble cause.

You think about, improving the life of humans around the planet, and giving them a better quality of life helping cure diseases or at least alleviate the symptoms of diseases. helping people to live the healthiest lives that they possibly can given the, human condition, if you like, and the things that do go wrong.

And that's always been my mantra. And throughout all the companies that I work for I really, look back and feel quite proud on the number of different medicines, for example, that I've worked on over the years and the advancements that we've have brought and are bringing, to the world.

I can't mention names drugs really for confidentiality reasons, but, I've worked on some of the biggest blockbusters in the world. And you're talking hundreds of millions of people that have taken those drugs, which is fantastic number, and hopefully improves their lives as a result.

So that's pretty much what drives me. so that's who I am. Quite an average kind of person really, to be perfectly honest. really privileged to be working with, this industry and so many great people. I'm still driven by trying to advance medicine and trying to advance the way we do medicine and how we actually, submit the data that we analyze, how we actually make the case for the, the medicines that we've developed and try to get approved with the authorities.

to me, that's the ultimate end goal is, increasing, the options for patients,

**Adam Walker:** Thank you for summarizing that, Andy. I absolutely wholeheartedly agree with, that mission statement and similar to you, it's the thing that drives me as well, making a difference for people, for patients, for family and as we get older, dare I say it, for ourselves as well.

Over, the years you've seen an awful lot of changes and, from a 30,000 foot view, you've seen so many things and experienced them firsthand in the trenches, in the doing and the advancement of thinking about technology, data collection, all the way through to, the reporting process.

What are the things that you take away from those learnings that you can pass on to perhaps the next generation of life scientists behind us? Because we have an awful- Yeah ... lot of young people listening to this who are keen to get into the industry, but what they're seeing is just the now. How did we get here?

**Andy York:** Yeah, several things there in terms of, how we get there. But, my key message to anybody looking to join the industry is don't put yourself in a box. Don't say, "I'm gonna be X, and that's all I'm ever gonna be in the industry." Make sure that you have, or take advantage of the opportunities that come along to reinvent yourself as you go along.

Learn new skills and adapt to the, the changes that the world brings. we're all seeing the kind of the AI revolution at the moment, so it makes sense if you join the industry to be ready for that, and, what that may mean for you in the future, depending on the role that you have.

**Andy York:** in terms of what we're seeing, obviously when I started, we had paper CRFs, three-part NCR where the, doctors wrote on the top page the nurses behalf of the doctors. And, there was like carbon copies made of that information, one of which came back to the pharma company or the CRO who then entered the information.

And, you spent an awful lot of time checking and cleaning and validating, with investigator visits to confirm, "Did you really mean that?" You know it's a lot of manual effort to try and, clean up the data as you, went along. And I think if you look at innovation, I think there's some significant steps in how we collect that data and how we treat that data.

And that kind of started funny enough at Covance with scanning that, they introduced within that organization, so that, the CRF pages were scanned and sent into the company for data entry. And then you start to get onto, EDC so that actually you're capturing the information electronically.

And that then gets transmitted back to the organization and, stored in, our data warehouses there.

**Andy York:** And at the same time you've got, from the programming perspective, you've got a lot of layering of standards and new and different ways of working to, support standardization in the industry and working with regulators.

when I started, we didn't have things like double programming. For example, where you had an original programmer followed by a validation programmer. And that quickly evolved into double programming because it was seen as an effective way of, producing quality control of what you did, particularly within the CRO industry.

And then you start to evolve things like standards, things like CDISC standards, SDTM, ADaM and even beyond that. Of course, all of that was done in SAS, and then now we're evolving away from SAS, , which is now, , a 40-year-old language into using languages like R, which is like a what?

20, 30-year-old language. So we're kind of slowly in terms of that, and I think it'll move on even further into, , future languages. Julia might be an example of something that we use even further in the future. So it does evolve. It does evolve slowly though, and I think, , even where AI gets involved, I think we're really only just at the beginning of that journey, and there's a lot of, , challenges around that, but I don't know, , how you see that yourself,

**Adam Walker:** yeah, I ... you make some amazing points there and I think broadly, those experiences sit alongside my own as well because we worked at Covance all those years ago when we were both young men, and I do remember the scanning technology coming in and just by very nature, as I've mentioned many times in this podcast we work in a very conservative industry where things do happen very slowly.

And I think around that point that you make with regards to, , the future and the future adoption, things are moving so quickly now, aren't they? They appear to be moving quickly on the outside of our industry and the outside of pharma. But I wonder how quickly those new technologies will really be brought into organizations, and the structure around which those are controlled and managed through the processing out to reporting.

**Andy York:** Yeah.

**Andy York:** I think that, is the key question that I see at the moment. I think, , certainly in, companies we know recently with the AI focus is that they would like to do it as quickly as they can, and it just doesn't work that way within the pharma industry because of the complexity of what we do.

, it's not like the banking industry perhaps where, you're pretty much collecting the same information over time throughout, the last 20, 30 years. So you can very readily apply AI methodology to, , the finance industry, , or banking or whatever. But in pharma we deal with some quite

diverse data, small amounts of data compared to, say banking data when you compare the two.

It might seem a lot to us, but it's not. And, it just gets really hard to set something up that's reliable. I heard the other day something like, we can achieve 80% of what we need to do using AI today. That's actually 20% failure to me. And if I look at that, from a Six Sigma perspective, if you had 20% of the work you do is wrong, that's actually a process that's out of control, and why would you ever adopt that process, so until AI reaches, competence level in the high 90s, at least 95% accuracy, I just don't think we can really honestly say that we're anywhere close. And, that's, again, part of the reason I joined a company like Verision because they're striving for that.

And you do see somewhat ridiculous things as well with the AI, such as descriptions of how to generate an output and somebody will type out an English language paragraph to say, "Please generate this table showing adverse events with, these conditions," or whatever. And by the time they've typed it out, a skilled programmer would actually probably have programmed it, so it, it's not really bringing a lot in that sense. So I think we need to really decide where we're gonna use AI as we go forward.

**Adam Walker:** I think that's a really rich perspective you give there around the confidence levels around 95% accuracy. You say that. I know, you've got a statistics head on your shoulders, and

That is such a pivotal point that you make there. 80% is not good enough. And in our industry, where patients are absolutely at the center, and patient safety is central- Yeah ... to everything that we do, you absolutely nailed on there. So how do we get from 80% to 95 and beyond, and how quickly can we get there?

Because you mentioned around, your work in advising a company like Verision. How is that gonna develop, and how will that become a reality in a shorter Yeah ... space of time?

**Andy York:** I think the trick is putting guardrails around the AI. So if you just, put nothing around it, it's gonna give you, a lot of hallucinations or garbage and, be fairly unconstrained.

And, you see that already If you go online and ask an AI agent to, give you an answer to something you can't be 100% sure it's gonna give you the right answer. And there's various stories about, inaccuracies that kind of come out of that. Uh, And of course it's improving all the time and will continue to improve, but if you put guardrails, and those guardrails might be your company's standards for output generation, your company's standards for ADAMS and SETMS you can actually start to achieve a much higher rate of success and accuracy with what the AI is trying to do.

And I think that's part of what I see and like within the Veriscian software is that they're putting those guardrails into it. So their AI they're quite open about it, their AI it's not perfect. It's not gonna do 100% validation correctly. It may get to a very high percentage, you know, of that ilk of, uh, you know, 95% or more.

But they do that because they're doing it, you know, smartly or smarter than other people are perhaps doing it at the moment,

**Adam Walker:** so following on from that, the human in the loop, as far as I understand it, checks and identifies any, errors in the AI programming around specific standardization.

There is clearly a distinct change between that double programming that- ... used to happen that we really did have enormous confidence in, to where we're heading towards. What happens in the interim?

**Andy York:** In the interim, um, and I think you're always gonna leave that human in the loop, and I think there's nothing to stop you continuing to do double programming.

And, you know, checking the quality and reiterating around the learning that you get within your large language models that you're creating to get them as good as they can be. And, you know, tho- you know, those first generation models probably aren't gonna be all that great in terms of, you know, they, they're gonna hit that 80% perhaps and, uh, that's about it.

So you, have to iterate. Uh, and until you get there, you know, you've got to rely on the human in the loop to, make sure that they're checking everything. And I know, uh, for example, the FDA requires that you say whether anything's AI generated these days when you submit. You know, um, it's kind of, how far do you want to push the boundaries there?

What do you want to risk? Because historically we've all seen the, bad days when drugs like thalidomide and so on, and the issues that we saw back then. We certainly don't want to have another one of those occurrences in the industry, so There's no room to get it wrong for our patients.

In terms of, humankind and getting things, as good as they can be. So it's gonna take time, but the, FDA are embracing AI. they're starting to use AI methods to, read submissions. We may find ourselves actually, the traditional kind of shape of the submission that we give may change in order to become easier for their AI to actually read over time, for example.

So it becomes more accurate. We're also starting to move away from huge piles of paper. I remember that, like 40 boxes of paper going into a submission, all being sent off to the FDA or whoever, and, these days, we have apps. I think why aren't we working towards having a submission app that actually packages everything up for the FDA and they can go in there and explore the data rather than look at, 500 pages of lab listings or something like that, so but it's small steps, and I think it takes, people to innovate around that and actually show that they can do that and take some chances around that. But until we get there, we're still stuck with the, manual ways of checking everything and making sure that it's right, I'm afraid,

**Adam Walker:** yeah, I have a couple of thoughts on that as well. As you were describing that I was thinking about having parallel streams. So parallel streams around the standard way, and then running something concurrently where you're doing a compare and contrast and- ... validating new practice against old practice.

Now, that's obviously gonna be a fairly heavy lift. So in the short term, to prove that AI and large language models are working, there's gonna be more effort and more requirement for involvement of highly qualified people who understand this data- Yeah ... and what you're trying to produce. Is that fair?

Is that accurate?

**Andy York:** I think that's very fair. I think what we have to do initially is create space for ourselves. So at the moment if you look at your typical stat programming team, they're so overwhelmed with the amount of work that they have to do and the timelines that they have to achieve it in, that actually they can't really stop and think about, "Can we do this better?

How do we improve the process? How do we utilize AI?" And I think by utilizing tools like, the Verizen tool, and if you can eliminate things like the need for double programming through that tool, you think about the amount of resource that suddenly frees up overnight. You go from two FDEs to one to, program and one to validate, to having maybe, 1.2 FDEs to do the same work.

And at that point, I think an FTE can be used so much better to help you think about actually now we have all that free time across our teams, how do we move things forward? The other point about that, and is people, perhaps miss this if they're outside drug development, is that the opportunities for pharma companies are often restricted by the amount of resources that we have.

In one of the companies that, I work for, we just had a drug approved, and we were looking at the opportunity space around other indications for that drug, and there was something like 30 plus different areas that they could actually, run clinical trials on using that medicine.

But, the company, it reduced to four or five because the company said we've only really got resources enough to do four or five." And if you think, you go back to what I said at the start about helping humanity with medicines if you could look at 10 instead of five, that's a massive increase in opportunity for patients and for the pharma companies to, drive revenue.

And, you can do that in a much more efficient way obviously, and actually achieve that with the same amount of resources that you have today just by being smarter with, AI. and that I think is a huge point to, make when looking at the rationale for this. So yeah, so there's a lot of work to be done now, and I think it's gonna take, years knowing our industry.

you're probably looking at 10 years, I think.

**Adam Walker:** Thank you for elaborating on that particular point. Again, I was thinking around the role of the regulator alongside transparency with pharma companies and contract research organizations and the supported technology services. Clearly there are a lot of tech companies that wanna get into life sciences, drug development, and- Yeah

and see pharma as an opportunity to really get the churn going into their models and into their technologies.

**Adam Walker:** When you've been at conferences over the last year or two are there some patterns that you're seeing? Are there conversations that are happening around that? Because I've certainly heard and seen various things.

I'd love to hear that from your particular point of view as well.

**Andy York:** Yeah. It's interesting.

**Andy York:** I think most of the conferences I've been to have been focused on moving away from SAS to, to R or even Python as a programming language, and there's been an awful lot of focus on that in statistical programming

There is a

almost a little bit of a cult-like feeling around the use of R and some of the things that people are doing with things like PharmaVerse and so on.

And it's great, Joining Novo when they did their R submission was absolutely fantastic to see, them pushing the boundaries and seeing a company where they allowed them to push the boundaries into things like that. And I think, it's good also to, challenge the establishment, so it hopefully will make SAS innovate more themselves because they're now having to compete in an environment that actually is becoming multilingual rather than just largely based on a single language.

And of course, when I started, SAS killed off a number of programming languages that, we use, or statistical languages if you like, that we use in the industry, like BMDP or GLIM and stuff like that, which people haven't even heard of these days. but they were tools that we were using when I started.

and it's almost like full circle, there's now something that's pushing SAS a little bit, but they're big enough to innovate, and I think they'll move on. For tech companies, trying to break into the industry, I think it's really important to talk to the regulators.

And I know that's something that Verition have been doing. And they have some open invites to go back to the FDA and present on some of their tools. I know that other companies who are successful have done that as well. So I think there's a piece around selling , your tool into the industry, but actually, , there's a piece around, , getting the regulators to, understand what you're trying to achieve as well.

I think one of the keys to success, as you go forward, but yeah, there's just so many different dimensions at the moment. It's gone from a very easily defined kind of pathway for what we're trying to do, , moving to standards and then go on from there to drive automation and things like that.

But, , it now seems to have gone off in, , a much wider direction than it, was previously. It's really exciting. , hopefully you see that,

**Adam Walker:** you

**Andy York:** know.

**Adam Walker:** I definitely do, and I wanted to just dig into that point around programming languages because I think your perspective is rich on this in the fact that certainly when I joined the industry it was always SAS, and I'd installed and validated several SAS systems into companies.

They had, and probably still do have, fairly strong handle around the industry. But with respect to many of the young graduates that are coming out of universities now, they are able to program in R and Python, and many of these young people have a very active interest in future-facing programming languages.

What are some of the key differences, from what you learned and- are able to program in SAS compared to R and Python because if I'm understanding correctly, R, Python, Shiny, many of these are open source models. And that I remember being particularly uncomfortable for people to accept, that we were- using open source programming tools that were not necessarily fully system held- ... and system validated and were openly accessible to others.

**Andy York:** Yeah. It's, , a very different way of actually working. , from the point of view that SAS was always trusted, , by the regulators in terms of what it did.

So if you used SAS, you were in a very safe place, uh, in terms of your statistical analyses and results that you drive. Whereas with R, because it is community-driven anybody can... write packages of their own essentially within the system. You have to have a great deal of trust or certainly you, would need to validate the packages that get generated.

And that's where things like the, you know, PharmaVerse is, is so critical or R Consortium even, to driving confidence around some of the packages that get produced within those places. Uh, you know, I'd encourage people to, you know go and check out, you know, the work of, uh, PharmaVerse, for example, just to see what exactly exists there.

But it's this whole, um, notion of validation by community means that the more a package gets used, the more you can rely upon the results that it's generating. So it's like crowdfunding for packages really. And the less a package is used, the more you have to validate it yourselves to make sure that you're comfortable with what it produces.

Um, and I think there's a group in, Fuse maybe in PISA, I can't remember that kind of validates between the different analyses even that you get between SAS and R. And a lot of where they see differences can all be explained by different options within the statistical analysis that, uh, the parameters that get specified to do the stats analysis.

And you can actually reproduce, uh, all of the SAS analyses using R at least that's my understanding to date, uh, just by applying the right options within R. So everything gets kind of like transferable and reproducible. So there, there's no real concerns there I think regulators are now more, uh, accepting of, uh, you know, these open source languages that, uh, they are actually okay, uh, to use.

And what they don't want to see is custom packages that you've built in-house. They'd much rather see you using the open source versions that are available to everybody because of that reliability, which is quite an interesting take on it from my perspective.

**Adam Walker:** That's a very different approach to what, happened in the past, isn't it?

Because that's effectively, as you're describing Uh, validation through, pure volume, through- Yeah ... volume of use and volume of users. And that's a very different mindset, particularly from a regulator's point of view,

isn't it? That's something- ... that I would say is significantly different over the course of, let's say, 20-plus years.

That, if you put yourself into 20 years ago, could you have imagined that happening any more than today and what that next era looks like?

**Andy York:** Yeah, and 20 years ago nobody would've taken the chance. And I think it's only through a few companies pushing the boundaries and just trying it to see what the outcome would be that we've reached the place where we are now, where actually we're starting to get a rich diversification of languages.

My view, you know, in terms of where we'll end up is I think we'll still be in a multi-language environment, but it'll be the right language for the right job the right task, if you like. SAS has some great tools and abilities within it, particularly around data manipulation. R is great for analytics and graphical output.

Python I suspect can do it all, and that may be the one that eventually takes over. But it's, you know, again, I don't think we've seen any Python-based submissions, although, you know, uh, Novo was dabbling with, uh, you know, some regulatory work using Python around the time I left but in a very small way.

**Andy York:** So in, future it may well crystallize and harmonize back onto a single language, but I think what we are more likely to see is that the AI actually starts to do the majority of the programming for us in the future, and the humans actually move from code developers to validators to make sure that the AI is generating the output correctly, um, and that you know, they're using their skillsets there to understand the data far better than perhaps they do at the moment.

Because of the pressures that we have today, I think a lot of programmers just follow the specifications without looking at the results. Uh, and I think actually, you know, it's pretty key that programmers understand the results of what they're producing, because so many errors can be caught that way, um, rather than take an approach of, you know, the standard says, you know, the specifications say something, I've done it to that so it must be right.

You know, I don't know how you see it, but, uh, yeah.

**Adam Walker:** Yeah, it definitely requires an elevation of the role of understanding- Yeah ... around this. And one thing that I always admired, particularly in statisticians and statistical programmers when I worked alongside them was just the eye for detail and the way that you have the ability to tweak a program that just makes the outcomes and the outputs a lot clearer and a lot more understandable.

Yeah. But you and colleagues like you have a different perspective, mindset, and capability- ... to program those outcomes. Now, what I'm wondering is whether or not those skills are being passed down through generations to the next- ... generations behind you and I, because that was the thing that gave me confidence, that the people that were sat alongside me- Yeah

understood what they were doing and why they were doing it. And like you just said there, I would be really uncomfortable with the fact that those skills and that eye for detail was not being brought forward in the work that we're doing in the future.

**Andy York:** Yeah. I think there's a real risk, and I think you're absolutely right.

I know some time ago they, did a poll on the average age of the statistical programmer and this is perhaps 10 years ago or so, and even then it was in the 30s late 30s. So it's a demographic that was getting older and older. And, it's not the easiest industry for people to join, if you compare earning money working for Google to earning money working for the farmer, you can earn a lot more at Google

it's, attracting people in is, has always been a an issue. I think the onus is on us to make sure that we're training people the right way for the future. And it may be that the statistician role kind of comes back in with the programmer role 'cause the statistician's always been about detail and what the data is showing.

Uh, the programmer has been about making sure that we present that information correctly. And so it, it's kind of almost there's this more hybrid, I think as you, go forward, where even the stats analyses may in the future be done by AI anyway. So it's, you know, entirely possible that, uh, you know, you kind of combine these two roles into, you know, what we at Novo call a data scientist as opposed to a stat programmer,

**Adam Walker:** following on from that point, I'm just wondering, how you see that next generation and the skills that they're gonna have because, I'm gonna make a very broad statement here, but people of my kids' generation in their early 20s- ... not my kids particularly, they're not gamers, but gamers, people who are spending a lot of time online and looking at strategy are the kind of people that I think today will make for very good programmers, will they not?

**Andy York:** Oh, absolutely. I have to confess, I play computer games. I've been playing them- ... since I was 18. and I was playing computer games since before there was an internet. and I was playing games online before there was an internet, so you know. So I go back a long way and, there's a kind of strategic mindset that comes about when you're playing computer games to, that you would apply to, programming, to get to the end result.

I think they will, and I think they're gonna be more comfortable with the technology as well. 'Cause they're already working with AI today. as long as they're not cheating on their exams by getting the AI to write the answers for them and stuff like that.

But, you still need that original, thought to go into what they're doing, of course.

**Adam Walker:** Effectively within the next, 10 years, we're looking at kids that will be AI native. Yeah. Not only have they grown up with a, phone and tech in their hand but AI will be the thing that they go to before they probably speak to their parents, their brothers, their sisters, their peers.

**Andy York:** Yeah. It's gonna be a very different world, I think, in, respect. And, I do read a lot of science fiction and stuff like that, and you can, see the older science fiction starting to come into reality now. when you look at, robots and stuff like that.

and there's a lot of AI in, science fiction that's coming through into what we do. it's quite fascinating, really. But I think

there's also some challenges there, I think schools are already seeing some challenges of say, kids coming in who don't know how to interact with humans properly when they come into nursery can't read or write or stuff like that.

So I think we, as, guardians of that generation, I think we need to be really thinking about those kind of things and making sure that we, equip them with life skills, not just, being good at using the technology that's out there.

**Adam Walker:** Yeah, I think that's a wonderful perspective you share there.

And just taking one step back, I do remember from a strategy point of view, you were always a brilliant snooker player, right? I've remembered that correctly. You were brilliant then. I bet you still play really well now. And I thought I was pretty good, but you were really good on a full-sized table.

But also darts. Do I remember you played darts really well as well?

**Andy York:** Yeah. All the vices, snooker, darts, and pool. I played- You know I played for a pub pool team at the age of 14. So yeah, very much misspent childhood in that respect,

**Adam Walker:** yeah, so that strategy has always been there with you and that was what I experienced particularly, in, those nights out and team building days-

and afternoons and evenings. But , it just came to me while we were talking about that, I just had to share that point. you make some wonderful points around science fiction being science fact these days.

**Adam Walker:** And as I say, I think from the perspective of what you've seen and what in the future, can we just bring that centrally back to, some of the tools that you're developing now with Verision?

Are you at liberty to share any of those kind of future-facing things, what we can expect coming down the line?

**Andy York:** I think the one thing... And I've been working for them for five weeks now. And I think the amount of energy that I see in what they're doing is huge.

And I think they are looking at where they can go all the time. So at the moment they started off with something that did straightforward traceability. You read in your SAS code, it created this big map or graph of all the connections with your code and the data. And they build a lot of interfaces around that to make sure that, understanding that traceability and working with it is fully understandable and usable by skilled programmers.

They're actually building, or they're working in partnership with CDISC at the moment to build out the core functionality within, their tools. So they're doing that on an open source basis. it's again, a tool that they're gonna be providing. They'll be probably talking about that later this year once it's, it's finished.

It's still a work in progress. And I was talking to them today about some extensions of things, that I feel are possible within the software. things like, doing patient profiles, patient narratives at the push of a button, because they have all the information in there.

They can generate things like defined like XMLs, reviews guides at the push of a button from their traceability that they have at the moment. but there's so many extensions to what they're doing probably more than they can actually do with the resources they have at the moment.

they're on the cusp of something really, quite significant. if they invest, hire more people, which they're doing, that they could then add on a lot of things to their environment. and it's brilliant to see because the aim is to disrupt, the industry in terms of what we do, and they are doing it in a way that nobody else is doing, from what I can see.

So very exciting times, And, look forward to where they're gonna end up as a company,

**Adam Walker:** I

**Andy York:** mean, but

**Adam Walker:** in identifying you as an advisor, they've obviously done their homework. Uh, they don't need to listen to too many podcasts, but anyone who's involved in this industry and has had the experience that you've had comes with enormous capabilities and the breadth of understanding of how that translates into the future-facing point, and I think you've really elaborated and spoken very honestly and openly around that.

And that future piece sounds like it's going open source and continuing on a similar sort of vein into the future. So thank you for elaborating on that and, and really thanks, thanks so much for sharing your perspective on that, Andy. Um, are there any particular points that we haven't touched on today that I should have asked you?

**Andy York:** Oh, wow. Um,, I think we covered a lot of the things I can think of. My advice to people who are, you know, currently programmers in the industry, you know, perhaps if you're in your 30s, don't sit on your laurels there thinking that you'll be doing the same thing in 10 years' time.

You know, really start thinking now about how you can diversify some of your skillsets to make use of different languages perhaps, uh, be more engaged with understanding data and, you know, upping your kind of involvement across the clinical development cycle. If you're new in the industry, uh, or want to break into the industry, yeah, think about the directions that we're going in and be, again, as diverse as you can in terms of your skills, not just your technical programming skills, but, you know, how you present yourself and how you, uh, you know, would sell yourself to a p- prospective, uh, you know, employer as you go forward.

'Cause I think it's gonna be harder for a while and then I think it, it'll probably settle down in, you know, that 5, 10-year span. But I think the the signs are that, you know, there may be some job cuts and things like that in our area. It's unavoidable, I think.

Um, but the ones who succeed are the ones that are able to diversify and think outside the box,

**Adam Walker:** I think that's a wonderful point that you've made there. And just to follow on from that, the other thing I will also echo is the skills of communication. That's where this podcast came from, it was about connecting with people in a new world setting, but also putting information out there for anyone and everyone to learn, to acquire knowledge, and to really understand what happens underneath the hood. Because it isn't a closed industry. It doesn't

need to be that way. The goal that I've always had with this particular podcast has been around opening up those conversations, making these things accessible to everyone, not just the people that are in our industry, but those that want to come into it, and there are many.

And I would just echo those same points to you, which is, diversity of capabilities, broad understanding. It's a bit like, learning sport. Do as many different sports as you can. Have multiple capabilities- Yeah ... like you, snooker, darts, pool. Darts and pool, yeah. Yeah. Yeah. Oh, that's brilliant.

**Adam Walker:** At this point in the conversation, Andy, I always like to conclude with a quick fire round. Sure. So if you would be so kind, what is the one piece of advice you would give to your younger self?

**Andy York:** My younger self. Continue to take risks. I've always taken risks in my, jobs at the companies I work for.

and it's a lot of fun doing it. what's the worst that can happen? So yeah, take a risk.

**Adam Walker:** I absolutely wholeheartedly echo that point as well. And as I've got older, I've got better at doing that, by the way, as I'm sure you have. What are the top three qualities you value most when building a team?

**Andy York:** I think ability to listen, ability to present and ability to learn.

**Adam Walker:** Nice.

**Andy York:** So it's the communication.

**Adam Walker:** Yeah. And we've touched on a few things that you enjoy outside of work, but what is the favorite thing you enjoy outside of work?

**Andy York:** these days yeah, I'm not playing too much, pool or snooker at the moment, but, we go out, looking at old houses and stuff like that, and seeing how things were done, back in the, early years, like Victorian or even Edwardian kind of eras.

Uh, so we do a fair bit of going around and looking at different houses, stately homes and stuff like that. My wife's into gardening, and she's just brilliant at it. She's kind of RHS qualified, so we're now going to a lot of gardens and stuff like that. And yeah, I still play computer games, as I say,

**Adam Walker:** what's your favorite game at the moment?

**Andy York:** Uh, June Awakening, I'm playing at the moment,

**Adam Walker:** and finally, what is your number one golden rule in life and in business, Andy?

**Andy York:** I think don't fret too much. I think everything sorts itself out. Uh,

Everything is a lesson,

**Adam Walker:** absolutely agree with that.

**Adam Walker:** We've talked so broadly around your experiences over the last 30 years, and I think you've really given a very clear perspective on what the future might well look like for people coming into our industry, and I cannot thank you enough for being so open and honest around that, Andy.

If anyone wants to make contact with you, what is the best way of contacting you?

**Andy York:** Uh, probably through LinkedIn initially. I do... On my LinkedIn I publish my Variscite email, so if you want to contact me in relation to Variscite you can contact me through that. But, uh, yeah, LinkedIn is the first point of call, I think.

**Adam Walker:** Thank you ever so much for taking the time today to be on Pharma Prescribed. It's been a pleasure to reconnect with you, Andy. And, um, let's hope we can work together again in the future.

**Andy York:** Yeah, always. Yeah, really enjoy working with you and, uh, yeah, really enjoyed, uh, the podcast today. So thank you.

Thank you, Andy