JUNAID KALIA, MD: Today I'm very excited. Uh, my friend Sarah Haque is here. She works for Pfizer, leads health informatics and is very interested in policy and governance. And we're going to talk to her about AI and how larger companies, especially pharma companies, are utilizing and integrating it in their workflow. Again, thank you so much, Sarah, for joining. Uh, if you can introduce yourself to the audience so that I'm not mucking up some things.

SAIRA HAQUE, PhD: Well, thank you for having me. It's a pleasure to be here. Um, as you said, my name is Syra Haque and I am a health informaticist by training. So, um, I'll start with my journey a little bit. After undergrad, I went to Michigan. I did my master's in, um, health services administration at the school of public health in the department of health management and policy. And so really from the beginning of my career I've had a focus on policy and its application in the real world. After that I worked in management consulting for a few years when I went around the country implementing different types of implementations, uh, for information systems. And that was before we had a whole host of big EHRs. So we did things like design and develop custom billing systems, which doesn't really happen now, or you know, implementing Meditech widescale or things like that.

JUNAID KALIA, MD: So you are the reason that I hate the medical environment right now.

SAIRA HAQUE, PhD: That's true. You know what's funny is, um, I feel like at the end, a lot — there's so many compromises that are made that everyone is mildly dissatisfied. So, um, that probably is true for a lot of things in life, uh, not just health information systems. But, um, you know, there's a whole host of considerations, and I think you don't truly appreciate it until you're leading something like that because there's considerations across all the stakeholders. Um, there's certainly regulatory, payment, all the kinds of things that all have to be considered. And, um, I really enjoyed building the compliance office. It was really fun to start something new and develop the systems and processes, but um, that's not where my heart is. I actually didn't enjoy being the compliance officer as much as building the office.

SAIRA HAQUE, PhD I went to RTI, which is a research institute, after that and I started as a health IT research scientist. And when I left 11 years later, I led the data interoperability and clinical informatics program. I then moved to Pfizer and I've been there almost five years as a clinical informaticist, and now I lead the advanced medical solutions team, which involves both clinical informatics and practical research. And so throughout all of these experiences, they are different in a lot of ways — working in a hospital as an administrator, management consulting, different types of research, and now. However, at the thread and at the core is using technology to improve outcomes. So regardless of the setting, regardless of the different way, that's been the common thread throughout my career. And so I wanted to talk through my journey because a lot of people listening might be coming from different types of backgrounds and not really see how their own educational background or employment experiences can lead to different types of careers. So, um, I thought it was important to share that for people who might be listening.

JUNAID KALIA, MD: No, love it. I mean, it's interesting. So, uh, what is your current role at Pfizer and how are you sort of, uh, looking at healthcare AI in general?

SAIRA HAQUE, PhD: Yeah, that's a good question. So I work in, uh, US medical affairs. So that means, uh, that's the department that I work in in general, and I lead a group called advanced medical solutions, which is comprised of clinical informatics and applied research. In terms of the applied research, we're really looking at solving real world problems in, um, identification of patients, management according to guidelines, and quality improvement initiatives. I also have a group of clinical informaticists who are trained and we work across, um, the spectrum really — um, looking at things like, um, EHRs to help identify patients for clinical trials, EHR innovation once, uh, people are in the system, real world data and evidence and different analytical techniques, and then evaluation and continuous improvement. So there's really a whole host of things that the team does.

SAIRA HAQUE, PhD: Now, one thing that I would like to kind of point out is that AI is, um, I guess the latest in a series of technical innovations. Throughout my career, I have observed, um, I guess like a little hype cycle of different aspects. And so with all of these things, the technologies are different but the underpinnings are the same — the underpinnings of, uh, what needs to happen to make something successful beyond the pilot is pretty much the same. And that's actually something that I really enjoy — is taking something that we're studying as perhaps, um, a pilot study or an initial feasibility study research, that kind of thing, and then how do we take it broader? How do we scale? And I like the innovation, obviously, as you can tell from my career. And I also really enjoy thinking about what do we do to take those things and bring innovation to more people. And so it might not be that everyone gets bleeding edge right away, but by taking things, uh, broader, even one step at a time, we can move towards better patient outcomes, which is really at the end of the day why we're all doing this.

SAIRA HAQUE, PhD: Love it. Love it. Go ahead, Harvey. Uh, floor is yours, or if you have any questions.

HARVEY CASTRO, MD: No, no. So many things I want to say. For me, I've taken the torch of saying, you know what? Uh, I'm going to do some TED talks on the same thing because if we don't know what we don't know, how do we know what to fix, right? And if we don't know these tools exist, then how do we implement them? And so my goal has always been sharing.

EDWARD MARX: Yeah, this is a super fascinating discussion. I've written down several, uh, questions for you that I think might resonate with the audience. And you, you sort of touched on the edges of some of these, but, uh, how — you know, this is your profession and so you have a little bit of extra pressure — how do you keep up? So I'm just curious if you have any personal like hacks or things that you do, or does it just come natural because you're in it?

SAIRA HAQUE, PhD: Oh, I don't know if anything comes naturally to anyone in terms of all this kind of thing. So, even when people have natural ability or talent, you need to have the work to back it up.

JUNAID KALIA, MD: Yeah.

SAIRA HAQUE, PhD: Um, but in terms of how I keep up — I agree, I mean, the pace has just been, um, astronomical. I think in tech the pace is always fast, but over the last five years I have observed that, uh, the pace of innovation, the types of innovation, the stakeholders involved — that is all really, um, accelerated. And I will say that, um, I use them a lot to keep up because, uh, there's a whole host of things like conferences. I'm on the industry, um, council. We get newsletters every day. So there's things like that to help keep up with the forefront of what's happening. I've also been, um, pretty involved in policy, and so I subscribe to a number of governmental type newsletters to keep track of what's happening on the regulatory side, which is a little bit different because the pace of technical innovation outpaces the pace of regulations, policies, um, organizational governance, etc., which is to be expected. One reason that AI is even possible is because of technological advancements. So, it's possible to analyze more data. It's possible to integrate across more data sources. And so, it's important to keep track on all different types of fronts. And yes, it's a lot, but um, if we don't keep up, then you'll get left behind. And so, I really encourage people to keep up with what's happening, but also, um, in different arenas.

EDWARD MARX Talk about a little bit about career advice. So people look up to you. You know, you're a well-accomplished person, professional, and they want to be like you. They want to be in your role. What is maybe one or two words of advice that you would give someone who aspires to be, you know, in a role such as yours?

SAIRA HAQUE, PhD: Yeah, it's a good question because I think that career development and even the way the career landscape now is, is very different than it was when I first started my career. And I think that's important to note. The environment is different, the context is different. Um, one thing I'll say that I'll start with is to be open to all different types of opportunities. When I first did my master's, I wasn't really, um, thinking about technology that much. I liked technology. Um, you know, I thought it was interesting. I did some things with it, but it really wasn't on my radar. But, um, you know, as I came in and did my coursework and then as I, uh, did management consulting where I started the strategy practice, actually, I, uh, really enjoyed it because I came to see that technology, systems, processes, and strategy are really all linked together, and, um, I actually came to enjoy aspects that I didn't even know existed. So, one thing I'll say is to really be open to different experiences.

SAIRA HAQUE, PhD: Another thing I would say is I've observed, um, a lot of people who ask me questions will often cut themselves off before they apply for something or before they put themselves into consideration. And I always think, you know, you don't have to cut yourself out. Let them do it if they're gonna cut you out, you know. Or, um, you know, put your head in the ring and think, because everyone has a whole host of different experiences. And I say this all the time — if you do what you always did, then you're going to get what you always got. And the world isn't the same. So, we need to move.

SAIRA HAQUE, PhD: And then the last thing I'll say is if you're going to do something, commit. I say this all the time, too. If you're going to ring the bell, ring it. Don't just, like, half-heartedly ring it. Don't, you know, kind of go half in. If you're gonna go in, go in and go in full force. Put your best effort in. Don't go in half-heartedly because that will influence what comes out. It might not be that everything is successful. The very tenets of innovation mean that they won't all be successful.

HARVEY CASTRO, MD: Yeah.

SAIRA HAQUE, PhD: That — otherwise you're not being that innovative. If it's 100% successful, if everything goes 100% according to plan, you haven't pushed the envelope enough.

JUNAID KALIA, MD: Yeah.

SAIRA HAQUE, PhD: So, we need to be comfortable with that. And then also to think about what can we learn from things that didn't go as planned.

JUNAID KALIA, MD: Yeah. Thank you so much. It was like a commencement speech. So good.

SAIRA HAQUE, PhD: It was great. I mean, that is a great compliment.

HARVEY CASTRO, MD: Yeah. I got to jump in. In the age of AI — and I'm gonna bring this to AI — we're all scared that we're losing our jobs, and oh my gosh, AI is gonna replace me. But listen carefully to what she said. She put herself out there and she hired people and she wanted people that were different. Because if you think about it — to, again, to her point — if we're trying to do the same thing and we hire the same people that think like us, we're going to get the same answers that we were going to give out. But if we get someone different — i.e., your personality, your clinical gestalt, what makes you you, that secret sauce of you — plus now we throw in the modern AI and you put it together, you're going to get some crazy good output.

SAIRA HAQUE, PhD: But one piece of that is also being comfortable with having discussions where we don't all agree on what the core problem is. But that's because of, um, our experiences and background. And it's important to shape that and really embrace it so that you have a better solution and that you have better answers to the things that you're trying to address.

HARVEY CASTRO, MD: From one to ten — ten being definitely in the next three years and one being total [bleep] — how would you rate that this is actually going to happen? Because I'm just going to be honest. Is there really a drug out there developed by AI? I mean, I know of one, but what I'm saying is — what do you feel in the industry, from one to ten? Do you think this is [bleep] or...

SAIRA HAQUE, PhD: You know, um, I don't think that things are so, uh, black and white in that sense because there's aspects that are happening right now. And, you know, will there be a 100% AI-developed drug without people involved? That doesn't seem likely. Will AI be used to influence various aspects? Yeah, that's happening as we speak, um, somewhere probably right now. Um, and so I think what we could really think about — and one thing that I really am excited about — is how can we use AI to, um, really make things more relevant for particular populations and subpopulations? And now, um, from the health system level, more people are capturing data around different types of subpopulations. So when you marry that together we lead to, um, what Harvey was talking about earlier, which is moving us more towards precision medicine.

JUNAID KALIA, MD: So in your opinion we are getting closer — from what we used to say like, okay, statin for everyone — to a specific statin for this particular population, and then eventually you're going to say that Jane needs this particular statin? And we are probably going to customize it for her. But, uh, what do you think — from a global regulatory alignment on AI — where are we at? And what do you think, how would you govern, let's say, a biosimilar or a statin for me? And how do you actually like deploy it?

SAIRA HAQUE, PhD: That's a great question. And so a couple of things. One thing I was thinking about as you, um, asked your question is — so, um, yes, there could be a specific statin design that is for you, for instance, but perhaps you have a different LDL target than Harvey or Ed or me. Perhaps the target that we're even aiming for is different based on our particular health and our particular background, our particular demographics. So, even before we get to the medication, we have to think about — well, what are the goals for us and why? And so right now we're relying on providers to keep up with their knowledge of what's happening in the literature to understand, um, perhaps to look at — well, you know, what is this patient cohort for this particular person? So, like everything, there's positives and negatives. I'm not casting aspersions. The intent was great. You know, the intent was good, but then, you know, there's things that happen.

SAIRA HAQUE, PhD: Um, and so, going back to that funded study — um, a lot of people don't know about it for a variety of reasons because it's more in the research realm. And that's one reason I'm so passionate about taking things from pilot to practice, because what's the point of having that cohort? That's just one example, but in general, what's the point of doing all that if we're not going to scale it? The other one I added was courageous leadership, because I know so many times people are just afraid to make a decision. And I had to be — you know, I've got some scars. You get scars from it, too. But that's what leadership is. You know, if you don't have scars, you probably weren't a very good leader. Um, we've got to be the ones to go out there and make things happen within your organization.

SAIRA HAQUE, PhD: Um, I did want to highlight one thing before though — about the bias and representation of data. That's something that, uh, we really need to spend a little bit more time on as an industry and as thought leaders, researchers, as people in this space. So many models — even models that are implemented and widely used in clinical practice — are developed on local data sets. And what I mean by that is they're developed on a data set at one institution in one region. And, um, the amount of times that those models are validated on a national data set, on a more representative data set, and then adjusted before going into practice — it's not as high as you think. Um, then we have to think about the data and we have to think about the provenance of the data. We have to think about who's in and who's out of data sets and why. And then going back to the policy that I cared about in the beginning — well, why aren't people in data sets? And there's a whole host of reasons that it's good to use them. But every single data set — there's no data set in the world that's a perfect data set. There's no such thing as a fully representative data set. It doesn't exist.

JUNAID KALIA, MD: Harvey, what are your thoughts on, um, synthetic data? That's the right term which people use. I don't like that word. I use the word augmented data. And the reason I use the word augmented data is because it comes from actually the basis. Um, Hari, first you go ahead with your digital twin and everything, and then I'm gonna ask Syra — uh, do you think it's ever going to be implemented in industry?

HARVEY CASTRO, MD: Well, this is a big controversial area. I, I'm biased when I say it. I, I think synthetic data — doing it correctly and being a data scientist from that perspective — I, I see a lot of weight to that, and I think it's a good thing, because again my focus is the patient. How can I help this patient? And if I can get over the culture and create a new paradigm way of doing things and helping my patient, then I personally feel ethically it's a crime that I'm not using this technology if I see someone dying because I'm not using it, per se.

SAIRA HAQUE, PhD: Yeah, I think that there's a lot of promise, and I don't think it's really a matter of if it's going to be used, but the degree to which these data sources will be used and the degree to which people will use the digital twins. They're there now, right? Like they're there. There's, um — I do like your idea of augmented data versus synthetic data. I like that term.

JUNAID KALIA, MD: If you had a magic wand and a wish list to actually use personalized medicine in one year, what would that actionable item be?