A Practical Guide to Incorporating AI in Pharma with Anne Heatherington, R&D Chief Data and Technology Officer at Takeda Pharmaceuticals

[00:00:00] Hey, everybody, Saul Marquez with the Health Matters podcast.

[00:00:12] I want to thank you for tuning in again to the podcast from Health 2023 here in Las Vegas.

[00:00:18] Today I have the privilege of hosting Dr. Anne Hetherington, R&D Chief Data and Technology

[00:00:24] Officer and Head Data Sciences Institute at Takeda Pharmaceuticals.

[00:00:30] She's an extraordinary leader in the space and leads the overall strategy and execution

[00:00:35] of all quantitative sciences post-close, including biostatistics, global outcomes research, epidemiology

[00:00:42] and digital sciences.

[00:00:44] And thank you so much for being with us.

[00:00:45] My pleasure.

[00:00:46] Delighted to be here.

[00:00:47] And so, Anne, before we get started on some of the work that Takeda is up to and that

[00:00:52] you and your team are up to, can you tell us a little bit about what inspires your work

[00:00:56] in health care?

[00:00:57] Well, actually, I started life as a pharmacist.

[00:00:59] So I wanted to do mathematics as a degree, and I was convinced by my pharmacist father

[00:01:05] to do pharmacy like my sister before me.

[00:01:09] And I loved pharmacy, actually.

[00:01:12] But then I discovered there's an intersection of maths and pharmacy.

[00:01:15] And I did a PhD then in pharmacokinetics, which is that intersection.

[00:01:20] And it's actually that intersection and that use of data that has been my entire career.

[00:01:24] So when I first started in industry, I worked in a very select part of that.

[00:01:29] But it was always about how do you design the experiments to generate the most important

[00:01:33] data that you need?

[00:01:34] And then how do you use those data to make decisions and then design the next experiment?

[00:01:39] And so what I do now is basically just a bigger version of that.

[00:01:44] But I will say my role at Takeda now is vastly bigger and vastly more involved in innovation

[00:01:49] than anything I've done before.

[00:01:51] And so that is really exciting, where I get to bring all things data, digital and technology

[00:01:57] into R&D.

[00:01:58] So it's similar, but very, very different.

[00:02:00] Sure, sure.

[00:02:01] Yeah, I appreciate that.

[00:02:03] And it does seem though you enjoy it.

[00:02:05] I could see that you get excited when you talk about it.

[00:02:08] And what would you say is the most unique way your organization adds value to the health

[00:02:13] system and patients?

[00:02:14] Yeah, so Takeda is actually a really unique company.

[00:02:18] It started over 240 years ago as a Japanese company.

[00:02:23] It moved to a much more global enterprise less than a decade ago.

[00:02:29] And the current CEO has a real vision around Takeda becoming the most trusted digital biopharmaceutical

[00:02:34] company.

[00:02:35] And we lead with our values at Takeda.

[00:02:38] We lead in decision making, we lead with what we call PTRB, patient trust reputation business.

[00:02:45] So every decision we make is made through that lens, starting with patient, what's the

[00:02:50] impact on patient?

[00:02:52] How does it help with our trust in society?

[00:02:55] What does it do for our reputation?

[00:02:56] And lastly, what does it do for our business?

[00:02:59] And so all of those things combined make Takeda a pretty unique company to work for.

[00:03:04] And so our CEO is leading this evolution into digital.

[00:03:08] And part of my job is to help realize that vision.

[00:03:11] And so the uniqueness for Takeda is really driving with patient first, driving from a

[00:03:16] very, very kind, inclusive culture, and then recognizing that this evolution is what we

[00:03:22] need to do, which is why we stick around for 240 years.

[00:03:25] And so all of those things combined, I think, make Takeda very unique.

[00:03:29] That's fantastic.

[00:03:30] Yeah, I love the value based approach and it provides a compass, right?

[00:03:33] Of what direction you should go, what you should research, what you shouldn't and what

[00:03:37] areas to enter.

[00:03:38] How are you incorporating innovative digital technologies such as AI, data lakes, wearables

[00:03:44] into your day to day operations?

[00:03:46] And what benefits have you seen from doing so?

[00:03:48] Well, I could talk all day about what we're doing here.

[00:03:52] Just to take a step back, two big things happened recently within our world.

[00:03:56] One is we had a transformation in the way the organization is designed, particularly

[00:04:01] with regard to what is traditional IT, such that it was decentralized and within R&D now

[00:04:07] we're now responsible for everything to do with, as I said, data digital and technology

[00:04:13] within R&D.

[00:04:14] And that decentralized, the decision was made to bring that into my organization so that

[00:04:18] we had a single vertical.

[00:04:19] So those that were working on the project teams, thinking about how we use our data,

[00:04:23] like our statistician say, are vertically connected to those that think about data

[00:04:27] architecture, data governance and AI, as well as those that build up our technology infrastructure.

[00:04:33] So that whole vertical connection is very unusual within pharma and creates a really

[00:04:38] unique opportunity for us.

[00:04:40] And then the other thing we did was we undertook a whole overhaul of the way that we run our

[00:04:45] clinical trials.

[00:04:47] And we've had the opportunity to totally rebuild systems and processes.

[00:04:51] And we have led with our kind of mantra around leading was quality by design, automation

[00:04:57] first, real time access.

[00:05:00] And so a number of those things combined have meant that we actually have a plethora of

[00:05:04] use cases around how we're using data, how are we using AI, but everybody wants to go

[00:05:09] to AI, but it's all about the data.

[00:05:12] And so what we're doing right now is really focusing on having a really secure data infrastructure,

[00:05:17] making sure that that then becomes accessible for AI.

[00:05:20] But I can tell you a few things that we're doing regardless.

[00:05:23] We started actually five years ago, a collaboration with MIT.

[00:05:27] And we have kind of leapt into AI more as a business case, as opposed to going in thinking,

[00:05:34] what can AI do for us?

[00:05:35] We led with what are the biggest problems that we need to crack.

[00:05:39] And so we identified those business problems, and they could be problems in our manufacturing.

[00:05:44] They could be problems in diagnosis of patients or any number.

[00:05:47] And we have collaborated with MIT over the last nearly five years now to help figure

[00:05:52] out what some of those business problems are by combining our data sources with some of

[00:05:57] our brains, but a lot of their brains to really start to crack some of these problems.

[00:06:02] And so some of the areas we're looking at, for instance, are some of the terrible neurological

[00:06:07] diseases like frontotemporal dementia.

[00:06:10] We're looking at developing new biomarkers, including speech.

[00:06:13] So like we're doing here, can we use speech as a biomarker for some of these terrible

[00:06:17] diseases so that we can run more efficient clinical trials?

[00:06:22] Yeah, a great application.

[00:06:24] And I think fundamentally what we're hearing here from Dr. Hetherington is it's about solving

[00:06:30] problems.

[00:06:31] So don't just throw AI around like a tool that you need to use.

[00:06:34] Start with your problems.

[00:06:35] I have a mantra in my organization that if what you're telling me helps us get good drugs

[00:06:41] to patients faster, the answer is probably yes.

[00:06:44] And so we go at it that way.

[00:06:46] So we're led by that as opposed to led by the technology.

[00:06:49] That's fantastic.

[00:06:50] Thank you for that.

[00:06:51] And where is the hype?

[00:06:53] Simply just hype.

[00:06:55] And what can be improved to make these tools and technologies more suitable for the pharmaceutical

[00:07:00] industry?

[00:07:01] Yeah, well, as I said, we don't really do the hype.

[00:07:02] Yeah, which is great.

[00:07:06] And I think it's because I have an organization who are really skilled data scientists, be

[00:07:12] they statisticians, programmers, epidemiologists, they all kind of fall under the umbrella of

[00:07:18] data science.

[00:07:19] And so developing algorithms is something we've all been doing for many, many years.

[00:07:23] And machine learning and AI is an extension of that.

[00:07:26] Granted, recently it's really exploded, but we're very comfortable with that space.

[00:07:31] But I think in the world of pharma, some of the areas we have to be really, really careful

[00:07:36] about are the considerations around data privacy, around consent, around representation, appropriate

[00:07:43] representation and around tech ethics.

[00:07:47] And so as we really think about the application of AI, we are bringing those experts in our

[00:07:54] organization, be they legal for some of the privacy aspects, be it our ethics experts,

[00:07:59] we're bringing all of them to the table with us so that we can lead with appropriate design

[00:08:04] elements so that we're not retrofitting.

[00:08:07] So as an example, we created a framework, we call it our ethical AI framework.

[00:08:11] We've created a framework internally that we adhere to whenever we're developing any

[00:08:16] AI algorithm.

[00:08:17] And with that, we assess the potential bias in our algorithms.

[00:08:21] We assess the potential to do harm or not.

[00:08:24] We assess the risk.

[00:08:25] We assess the bias in the data sets.

[00:08:28] And so everything we do is run through that lens so that at the end of it, we can stand

[00:08:33] behind what we're doing and realize that everything we do directly or indirectly impacts people.

[00:08:39] And so we want to make sure we can stand behind that.

[00:08:42] Thank you for that.

[00:08:43] It sounds like a well thought out structure and governance for the use of AI.

[00:08:47] And so what skills and competencies do you think are necessary for the drug development

[00:08:52] workforce to succeed in the sector of digital transformation?

[00:08:57] So I would parse the skills into three buckets.

[00:09:01] I would say the first bucket is recognizing that we do live in a regulated environment

[00:09:06] to all the things I just spoke about.

[00:09:08] And you have to want to live in that environment.

[00:09:10] You have to be willing to live there and recognize that to a greater or lesser extent, everything

[00:09:14] you do impacts patients.

[00:09:16] And therefore, you have to be appropriately trained and live up to the privilege that we

[00:09:21] have of testing new molecules, new chemical entities in patients.

[00:09:26] So you have to want to be in that world.

[00:09:27] I'd say that's the first bucket.

[00:09:28] The second bucket is we will always need foundational scientific and clinical capabilities.

[00:09:35] We will always need chemists, biologists, pharmacologists, physicians, statisticians.

[00:09:41] We will always need those amazing brains and capabilities.

[00:09:45] And then the third bucket is how you bring in the innovation on top of that.

[00:09:49] How do you bring in the data science or the algorithm building or the thinking around digital

[00:09:54] tools for use in our clinical trials?

[00:09:56] And then the magic is how do you bring all three of those elements together?

[00:10:01] Because in many organizations, those disciplines all sit in different parts of the organization.

[00:10:08] And I do think that we have done a better job at ICATER at integrating those elements,

[00:10:13] either many of them within my own organization or certainly very closely within R&D.

[00:10:18] And having, I should say that I think an overriding requirement within R&D is both

[00:10:25] two requirements, I would say, is both curiosity and resilience.

[00:10:28] And if you bring both of those, you can be successful.

[00:10:31] That's fantastic.

[00:10:32] No, I love that.

[00:10:33] And so is that part of the reason for the restructure of the IT that you guys did to

[00:10:38] be able to hit the sweet spot with those capabilities?

[00:10:41] Yes, that's a good way to put it.

[00:10:43] It was actually driven by our CEO about 18 months ago.

[00:10:47] And what he wanted was really exactly that.

[00:10:50] He didn't want it to be some distant force in the center of the organization.

[00:10:54] He wanted to bring the expertise much closer to where it was needed, be it in my case

[00:10:58] R&D or manufacturing or commercial.

[00:11:01] And so we're seeing the benefits of that now within the organization.

[00:11:05] That's great.

[00:11:06] What an innovative approach, utilizing organizational structure, governance and the importance of

[00:11:11] staying within the regulatory framework to deliver on some of the promises that ultimately,

[00:11:16] to your point, getting drugs to patients faster.

[00:11:20] And this has been an incredible interview.

[00:11:22] I can't thank you enough for being with us today on the podcast.

[00:11:25] If you had to leave the listeners with one closing thought, what would you leave them

[00:11:29] with?

[00:11:30] I would say drug development is really, really hard, hence the resilience.

[00:11:34] However, it's a really incredible space to be in.

[00:11:37] And right now we are at a cusp point where if we don't learn to marry the traditional

[00:11:43] with the new, that we will become obsolete.

[00:11:46] And so I do think being able to do that and leading with that and really figuring out

[00:11:50] how we do that is critical.

[00:11:52] Thank you, Anne.

[00:11:53] And folks, appreciate you guys tuning in for this podcast with Dr. Anne Hetherington of

[00:11:58] Takeda.

[00:11:59] And for anybody that wants more information, Anne, where can they visit?

[00:12:03] They could visit Takeda.com is our website, and there's a lot of information there.

[00:12:08] Outstanding.

[00:12:09] Folks, check out the show notes and you'll be able to get more information on the work

[00:12:12] that is being done at Takeda there.

[00:12:14] And thank you very much for your time today.

[00:12:16] Okay, thank you.