Building a Data-Driven Pharma Organization Through Analytics & AI with Shionogi's Anindita “Ani” Sinha

Welcome to Tech & Drugs. Our guest today is Anindita Ani Sinha, a leader whose career journey spans from the microbiology lab to the helm of commercial strategy in Big Pharma. As Vice President of Commercial Operation at Shinogi, any of us is critical functions, from analytics and marketing side to field operation, driving innovation that advanced Shinogi's mission. And it brings nearly 15 years of experience in the pharmaceutical and biotech industry. She has health-treatically leadership roles at industry giant like Pfizer, Bioselgin, where she led teams in market access and business operation, even spearheading the successful global launch of a breakthrough immunology drug. Any holds a BA in biochemistry from Columbia University, and a master in microbiology from Yale University, giving her a unique perspective at bridge trading edge science and business. Known for her photodorship, Any often shares insight on the future of pharma and how technology can accelerate drug development. She is passionate about leveraging data driven strategies to improve patient access to new therapies and has spoken about what inspires and excites her in the world of life sciences. With this blend of scientific, ecumen, strategic expertise and passion for innovation, Any is an exciting and relevant voice in pharma. And we are thrilled to welcome her to the Tech & Drugs podcast. Any, welcome to the show. Thank you so much for having me. Let's get started. The way I like to start this discussion is go back to why are you doing what you're doing today? What's been the journey like? Tell us a bit about yourself, where you're always an A student, what drove you to a world of life science and pharma? Thank you, dear Balc. So yes, I will say I was always an A student. I was very much a bookworm, very much a nerd. I always had my nose and books and was a very diligent student. But that's not why I ended up in the world of pharma. So I was starting my career in academia, if you will. As you mentioned, I had done my graduate studies at Yale. And so I was fully expecting I was going to end up in a laboratory and be a professor and kind of continue to do research. But I'll say about halfway through my graduate work, I realized that I really craved human interaction. And I know that sounds strange because you're in a graduate school. You think you're surrounded by people. But the work itself could be very isolating and very individual. And so for me, what I really enjoyed was working on a team, working towards a common goal. And so that's where I made the transition from academia to consulting. So once I was in consulting, to me, that was a wonderful way to enter the life sciences arena. Because really, it's a crash course in everything you need to know in terms of not just content wise but also work ethic wise. It got some of the best training I ever had in the four and a half years I was in management consulting. And so from there, I made the transition into industry. So I was recruited into Soljin to build a small, but mighty team of market access analytics, which back in that time, that was a pretty new concept to have an analytics team focused on market access. And really from there, I just jumped from place to place because I really was very lucky where I had people either coming and knocking on my door saying, "Annie, can you replicate what you've done here at a different company with more breath?" Or can you actually take on a completely different role that, again, you have some expertise for it, but we're willing to take a chance and have you brought them. So I really jumped from Soljin Bayer, Pfizer, Baring Grangleheim, and then more recently, I was at some smaller pharmaceutical companies that again allowed me to really truly roll up my sleeves in a very different way. But then also be really at the cutting edge of kind of when decisions were being made. So to really have far more visibility in terms of how choices and decisions and analyses that I did, in fact, impacted the rest of the company and really impacted the future of the drug or future of the brand. So all of those different roles have brought me to where I am today, which is culmination of a lot of it. I do have a very strong background in analytics, which I think made me a nice option for commercial operations, which really is, as I've described at the engine of the commercial organization. Everything that's not market access, marketing or sales ends up in my shop. And so we have a highly specialized group of individuals, highly skilled that really do everything in their power to make sure that the commercial engine continues to run. And yeah, it's been quite a journey, but it's certainly been a lot of fun. - You mentioned that you quickly went into analytics and your background, so your scientific background, you could have done things in more in the discovery, but it seems that very early on you're specialized in analytics. So what is it that drew you to this path of the value chain? - I think what drew me to academia initially is probably what ended up drawing me to analytics and more strategic commercial avenues, if you will. One thing I realized in academics is I loved basically answering questions. I loved getting a problem and then breaking it down into little pieces and answering those pieces and then putting it all back together and creating that story. A lot of what you do as a scientist is you do want to create a story. I maybe didn't love the story I was creating as a scientist, but I did love creating the story, the actual kind of process of that. If you lift and shift that into pharma or into commercial avenues, it's a very similar concept. So for me, analytics was a very reasonable and almost organic transition, where it's okay. So now instead of me being in a laboratory and looking at Southern Blots and trying to understand what that means for the greater question I have, I'm actually analyzing data and I'm trying to understand patterns and I'm trying to draw conclusions of those patterns to map them to customer behaviors. It's similar concepts, similar process, just you have different building blocks. >> So it's a great to say with the first of people that I want to discuss with you, because now when we talk about analytics and recognizing patterns and answering questions, so of course the first thing that people think about, it's not a dashboard anymore, but it's AI. Everyone is talking about AI. We talk a lot about AI in this show. So AI in pharma, let's try to separate the hype from the reality and of course it's described as transformative technology for pharma. But from your perspective, what's the reality today? Versus maybe the promise of the hype or things that are actually not working today. >> I feel like I get that question a lot and it's a good question because you're right. Everyone wants to know about AI. Everyone uses AI, every resume that I've been reviewing, these days all have, I do AI. It's okay. Well let me first like level set viewers, you don't do AI. AI is a tool, it's something that you can use and leverage to get to where you want to get to. So in terms of the hype versus the reality, and this I have a fairly strong robust opinion about, the reality is that in pharma, we've had very few sustainable, repeatable successful use cases where we can point to that and say you know what, that's a method that we can definitely use artificial intelligence to benefit us. I can give you a couple examples where we're starting to develop those use cases and kind of those repeatable examples. But it's taken us a good, I would say, six to 10 years to really get to that point, which to me speaks to the delay and kind of understanding and also the replicability of utilization of this tool. The hype around it is all around, oh it's this new innovative thing and it's this thing that's going to make everything better. Yes, it can make everything better. A great example is, you know, something as simple as I was writing a survey the other day and I was using Chachi PT to help me actually write some of the survey questions. It sounds like the most rudimentary example, but it ended up saving me a good two to four hours of time and it didn't mean that I was doing the work less. I think that's another misconception. People assume if you leverage tools like that, oh you're basically dumbing yourself down, you're letting somebody else do the work. Not necessarily because I'm still having to provide the prompts, I'm still having to provide the filter, I'm still having to edit and make sure what I'm getting back makes sense. So it's interesting because people keep talking about artificial intelligence and I think what I always encourage my teens to think about is, what is the actual question you're trying to solve? What's the goal you're trying to achieve? And then can you use a tool like artificial intelligence or machine learning or a natural language processing to help you get there faster? You're making a great point because there is also one of my hard-eastern English pet peeve. When people say, yeah, we need a strategy. We need an AI strategy. I'm like, no, you don't need an AI strategy. You have a business or a scientific strategy and then you have a number of tools that can help you to accelerate the strategy. And AI, in some cases, one of us, tool. And in many other cases, it's not. And I think you gave a good example of there for some things, some use cases for which AI, it's there already. It works really well. It can accelerate things by several hours or more. I use AI also just like you for writing things. As you just said, it doesn't do it for you. I think it helps you maybe do it better, structure it better, but you still have to write a lot of stuff. If you don't, if you just ask a simple question in general, you don't get something very usable. 100% agree. And I think that's where I know a lot of people ask, what's the best way to then use AI? And I'm not by any means the expert here. I'm simply a user and a learner. To some extent, a dilatant, I'm still very much dabbling in everything. But I think one of the most critical pieces that people forget, where people maybe just don't focus on. Maybe it's not they don't forget. They don't emphasize is you still need to have a really strong grounding of your own baseline knowledge in order to know when you're leveraging AI. if it's giving you something that's really truly useful versus it's giving you something that sounds good. And those two are very different things and it requires still that human brain to be able to validate the difference. Something that I find incredible concept of hallucination in AI, which is I think for us, one of the biggest problems. So very often I'm writing things and I want to have scientific references and it keeps inventing scientific references that sounds perfect. When you look at them, you're like, oh, this is a great article. And then when you look for the article, it says, the article doesn't exist. I just made it up. Why are you doing that? Yeah, that's that honestly, that's probably one of the scarier examples. This scary, and as you said, if you don't check, if you don't have a critical look at the output, then it can be dangerous, especially for industry like ours, where the steak are quite big. It's you don't want to get an entire team working on something based on on on fake reference. Agreed, agreed. You don't want to make an entire business decision based on that trust, but verify trust, but validate. Definitely, those are good tenants to have. Which brings me to another question, the importance of foundational data work, having clean, organized structure data. So I guess in your space, this is something very important. And nowadays, it's really the fuel. So it's been the fuel for analytics forever, but it's getting even more important in the context of AI. Can you tell us a bit more, maybe about the work that is going on in your space to have access to the right data and the effort and potential difficulty also having access to those data. So I think one thing I kind of want to make a little bit of a point there is I think one thing that's really amazed me, but is also part of what interests me in the space is just the sheer change in our expectations that have really come along with the sheer change in the types and the volume of data that we now have access to. And back to my time at Selgin, we were excited by getting site alerts, you know, so and that was like so great if we were excited about that. Oh my gosh, we have a new way of understanding the potential intent of a position and the concept of being able to capture the intent of a position identification of a patient before that patient is truly been identified. That was like cutting edge back then and we were so excited by it now fast forward to 10 12 years and we have an exponentially greater volume of data, but then we also have increased our expectations to the point where now we want to predict what a physician does before they even do it. And to anticipate a patient showing up somewhere before they actually are diagnosed or even show up to that side of care. So in terms of going back to your question around foundational data and how do we access it process it think about it and really what's maybe our philosophy towards it, I would say in order to really begin to think about having a really successful model powered by AI. So you absolutely need to have a very solid foundation of data and it's really comprised around two things. I think the first is you need to have a really fundamental understanding of every piece of data that you have in your box, if you will, in your foundation. And what I mean by that is it's not just understanding what the data is, but it's understanding the limitations of the data because I think people forget is you think, oh, if I give AI a bunch of data, it'll just look at it and it'll tell me great things. But then to your point, it could hallucinate it could do things that you don't want it to because it doesn't understand the limitations of the data it will only understand it if you tell it what those limitations are. And you as a prescriber of all this, you need to also have a better understanding of self, what are those limitations, which data sources should be used at a higher weight in terms of a decision making model, which data sources, I would consider to be have higher veracity than others just in terms of their coverage or their linkage. Whatever it may be. So that's the first part. And then the second part is ideally, if you can connect the data sources, that's where the real power gets unlocked. Everyone talks about, oh, let's have this massive daily and just put everything in there and all the data will talk to each other and then we'll just have this massive database. I've heard that concept thrown around for the last 10 years, let's have a massive database. And I'm like, okay, but what are you trying to achieve with that? There's value in connecting certain data sources for sure, one very simple example is if you have a pay or backbone and you can connect it to your prescription data that has managed care data fields associated with it. Yes, then you can actually start to track this is a very basic example. You can start to track formulae changes and start to draw correlations to performance as a result of that. That's a very basic example that's very US specific, but that being said, that's one connection that takes a lot of time and trouble to do that. What's the problem? Like what's the value in connecting 20 different data sets if you don't understand how they're actually going to potentially talk to each other and what value and what information you would get from making those connection. If you yourself don't fully understand that at least at the baseline level, how do you expect AI to be able to put itself on top of all that and then extract patterns from it. It'll extract patterns that potentially could be completely erroneous because again, you yourself don't fully understand the underpinnings of it. So I think those are probably the two most important categories that I always want to put front and forward when we talk about, okay, what do you need to really bring your AI kind of. If you want to get a AI strategy to live, you need to have that solid foundation first. Now we are starting to consider data as an asset. So not just something that that you use one time, but something that you will be that you will reuse. But to your point, throwing everything in one big leg doesn't really work. So you want to start with a question. And then you go and look for the data and then you do the work on the data you connect the data and then you reuse the data and use. So the way you prepare the data, you need to think, okay, those these data that I need for a certain question will certainly be used later on for question that I haven't thought about, but I need to make sure that the data labeled, etc. So how do you approach that at the Shinogi, so you were talking about we want to bring the right data connect them to answer specific question. There are literally thousands of questions you can ask in the form. So how do you prioritize and say, this is the most important thing that I want to go after. And this is these are the data that I need. This is the connection I need to make. So what's your whole process you approach and are there like couple of example that you could give in your line of work. Yeah, I think you're asking a very reasonable question that unfortunately probably doesn't have a great answer. I like questions evolve and that's part of why like my job. I do the fact that I can continue to ask questions. We constantly can we're being paid to be curious. That's how I look at it. That's a lot of what we do all day long. But that can also be counterproductive if you don't have a goal. So let's go back to your original question. It's okay. How do you prioritize. So I think for the immediate needs of the moment, you usually there's about just roughly right five to 10 critical business questions that you need to answer that will get you the answers you need for let's say the first three to six months of any kind of process that you're looking at. So let's take something for example for a launch. You're definitely going to need to develop a forecast. Okay. There's some questions you're going to need to answer that will immediately help you inform the forecast and then from the forecast you then have been sequentially can ask other questions to help you inform your brand strategy or targeting strategy where the opportunity is and so on and so forth. So it's in terms of how we set up the data and this is not unique to my company. This is I would say something I've seen across multiple farm accompanies that I've had the privilege of working at. I want to think about how you want to set the data up. I think you try to the best of your ability and it's not perfect. You often have to have several iterations of this, but like when you think about setting up your data warehouse when you think about setting up those data tables. You try to find a way to make each data set as universally accessible as possible. And really a skill. This is where a skillful data management professional or a skillful data strategist can really understand look these are the needs of my analytics teams. These are probably the fields that are going to be the most accessed and the most needed. So how can I make those fields most accessible. And then I mean them the best make sure that their quality is the highest integrity, but then ensure that they're connected to different data sets or have the ability to connect maybe not connect today. But there's an ability to connect to another data set in the near future. Build those tables and build that flexibility into your underlying data structure so that you have that. And then you have that flexibility to ask additional questions in the near future. So again, it's not a perfect process. It's iterative, but I think really putting in a little bit of that for thought in the beginning when you start to build those structures helps you in the long run. So yeah, I think I wanted to to discuss with you is when we were preparing the call we were talking about data democratization. And also for concept I've seen in our industry where we talk about, okay, now everyone blanking on the term it's decent. Okay, this intermediation. So I don't know if I've ever heard of this and pronon and prononc and prononcable term, decent, the mediation. So essentially. And over the main what I've seen is that, for instance, you want to get something done, you can do it yourself. So you need to do it via someone points as I work for many years at health severe and at health severe we worked a lot with librarians. And if you go back 20 years ago when you wanted something you wanted a piece of scientific information you will go to your librarian and then the librarian will find this thing for you. And now the role of librarian evolved to information manager and they give you the tools so that you can answer the question yourself. And I worked in the past also with business analysts where if I had a question, I would ask them and they will fetch they will fetch the information for me. So how do you see that your role evolving in this context? Where I think one of the one of the promise of AI is that there is a lot of things that people were not necessarily able to do their own. But now we can do on their own because there is this dissentimentation and we don't necessarily need to ask business to do certain type of thing. However, it doesn't mean that the work doesn't need to happen. It's just that the work is happening somewhere else. You've probably heard this term in the pharma self-service analytics. Yes. Yeah. So I think that's a great example of where my team and other teams across the country and across the world really are probably finding ways to, again, I'll use that term you use, democratize the data. So when you think about, let's say, 2025 years ago, you would have to go to a very specialized data analyst to answer certain questions because certain data sets were so complicated they were completely inaccessible unless you knew how to basically dip into them and answer the questions you do. And to some extent, we still are in that state. When I think about some of the more extensive claims analysis that I often lead, if I don't have a successful and skillful data scientist to help me with that, I mean, just programming the data, writing the code, making sure you have the right business rules in order to write that code, extracting the data, analyzing it, and then providing it, even if it's just providing it to me in tables and exalb-based spreadsheets, that's still a significant skill set that not everyone has. But I think where the democratization of data and the kind of shifting in terms of how teams like mine play a role, instead of us taking on all of that, what we're seeing value in is partnering with the I platforms like Tableau and Power BI and saying, you know what? We know that there's probably 10 to 15 questions that you as a leadership team are going to want to have answers for every month, because the business evolves. There's going to be different questions and different analyses that you're going to want to do constantly. But generally, there's a good bifurcation or demarcation of there's some more ad-hoc things that you're going to know, but then there's generally a set kind of systematic piece, a set of analytics that you're going to want to have constantly. Sets of questions that you're going to want to have answers to on a daily or weekly or monthly basis. Work with those platforms, basically program everything you need, and then make it user-friendly so that it's pushing a button, clicking on a button, clicking on a spreadsheet, and then everything is right there. And that allows someone who is not a coder, who's not a true analyst to still ingest the information in a user-friendly way. The user experience becomes very different, but then they are also able to understand this is what's happening at in the market level. This is what's happening at the product level without needing to necessarily wait days for an analyst to do all the work or worse yet to have the analyst translate all of that work again to that. So that to me is a great example of where even though it's just some extent I can see analysts getting nervous, it's when we're taking away some of my ability or taking away some of my facetime. It's been I've been hard-pressed to find an analyst who really feels that way just because I think they don't want to do the systematic kind of boring stuff every day. If they develop the right tools and the right code and the right questions to be asked, they would much prefer to have a program automatically spit out into a visual kind of self-service platform if you will, and then answer questions rather based off of that. So it's like instead of answering the first set of questions, have that automated use AI machine learning, whatever coding the simplest form, just get that all automated, and then you have a higher level of questions you can ask, and then you can actually have a much more cool some conversation. So yeah, that to me is definitely what I'm seeing, how I'm seeing us evolve, and then we can talk more about how there are certain platforms that are just taking that to an even higher level in terms of instead of just stopping at the point where it's, oh okay, here are six views of the world, take a look at that. Now the platforms are actually prompting you with questions around those six views. Actually this looks different from a did last month. What do you think about that? These five doctors have suddenly jumped up on the list in terms of higher prescribers and they weren't there for the last four months. The possibilities are endless. Now that your end users have been using Chargity for the past three years, do they want to interact with data like they do at GPT? And I'm asking this question because that's something that I got from a senior leader in one of the company I worked with and it was like I'm doing that when I'm at home with with Chargity, why can't I do it with the data in the company? And it was a genuine question. It was like why? And there was like a median reason why, but still I think it was a good point. It's like there is this technology which is becoming a bit more mature now, a bit more reliable, okay, modulo of hallucination. But I think this is a direction where a lot of users say, I don't know what we have to learn to interact with a platform, the way the platform want to be interacted with. I want to interact on my own term. So that's something that I'm seeing more and more. That's actually a really, that's an interesting point because I think you're hitting on something really critical there. People want to interact with these tools or these platforms in the way that they want to interact with them, but that's not necessarily how they're being built. They're not being built necessarily to flex to the user. They're built in the way they're built and the user unfortunately, at least this has been my experience thus far in the few, the several platforms that I've worked in, where the user really needs to learn how to query the platform appropriately. So I'll give you an example. We are partnering with a German company to basically do what we're calling unprompted market research. So they have this wonderful tool. They're leveraging artificial intelligence and machine learning to basically go out and scrape the internet and gather patient conversations. And as a result, statistically by using statistical methods, they're able to really build up these patient conversations into patient personas and help us understand like, look, what are some of the key sentiments that are out there in the universe in terms of what are people saying? What are people in an unprompted way making observations about? It's become a really valuable source of information for us in terms of understanding our consumer base. But that being said, so they've developed a wonderful tool. They call it patient GPT, fantastic tool. But I tried using it a few times and I realized I was goofing up because the queries I was putting into it, I was getting answers that didn't feel quite right. I wouldn't say there were wrong answers, but they just weren't particularly helpful or they felt kind of vanilla or they felt really linear. So what am I doing wrong? I'm asking a question that I think is very thoughtful and insightful, but yet I'm getting this very basic answer back. And it's really a lot of it is trial and error. And we really had to work with our data scientists to get to the point where it's, look, let us give you the questions we're trying to ask, but let us give you like maybe six to seven different permutations of that question. And then help us understand if you can engineer it so that you can help us understand how the database can answer our questions. It's been, and that process has taken some time. So I think what it's really taught me and it's really been a humbling experience is that this is a really powerful tool and really powerful database or in the case of kind of your examples, like it could be a very powerful platform, but it's powerful because of the way it is built, but unfortunately that building didn't necessarily take into account the user experience in that way, meaning the user can't necessarily query it the way they want to. So I don't know how to solve for that, but that to me is a great example of where if we're going to make those types of platforms more readily available, there's going to have to be a little bit of education and kind of expectation setting around it where, all right guys, we can give you a very shiny BI platform, but you can talk to just recognize like the language you use may not be totally compatible with the platform. I was talking to a company that was building this kind of platform and tools. And I was talking to one of the tech person and he was like, all the problem with users is that sometimes they ask for a bit in question. And I'm like, what do you mean by for a bit in question? And he was like, let's question that don't work. And I was like, well, that's a problem then, because just like you said, you ask a question the way you feel the other right question. And if you make a platform or tool that is very flexible in the way you can interact with it, of course, if you have a thousand users, you will have a thousand different ways of interacting with it. I think that where we are at the moment, we are coming from a pretty rigid tool like Tableau or Excel, where you have formulas, you have things that you can click, everyone needs to follow a similar workflow, the tools that are that enable total freedom, but this total freedom doesn't always work. As you said, I think we'll need to find a good middle ground if we want to deploy those tools more widely. I think you're right. There's a happy medium in that structure versus flexibility. And I think where, for example, we finally were able to get something useful out of that particular patient GPT platform is we basically had developed a questionnaire and we gave that questionnaire to the engineers and we said, and to the data scientist, we said, here's 200 questions that we're trying to ask. And I don't know how they did it. I do know how they did it. They have the whole I should make it seem like it's a black box. That's the other thing I will say. Any of these platforms, any of these tools that we use, the moment you have someone who's like, "Oh, it's too complicated to explain," or, "Oh, you don't need to know how it's done." I immediately discount half of what they're telling me because I'm like, "Okay, none of these tools should be so complicated or so occluded that you can't help me understand what the query was or what the premise was of your question or the premise of your analysis was." But yeah, we ended up providing those 200 or so questions and then we ended up getting some really fascinating insights. They basically showed us that they had taken maybe two or three different permutations at each question and then they were able to finally hit on the right syntax and the right question in order to be able to get us, again, the level of detail we were looking for it. The answers didn't change necessarily based on the syntax because I think people will hear what I'm saying and get worried. "Oh, maybe the reason you're hearing what you want to hear is because you're asking a question in the way it wants to be asked." And there could be some truths to that, but at the same time, I think the value or at least the veracity that I find in this data is that the general themes of the data were still coming through. I was still getting the same themes of the answers. It was just the level of detail around those answers that was different. Yeah, I think your putting a finger on a very important thing is that you will have 10 or 20 ways of understanding 10 or 20 ways of asking the question, but you want to get the same answer. And I think that's one of the things that people are concerned about at the moment is that, yeah, if I ask in a certain way, do I get different answer? And they are things where you don't want flexibility at the answer stage. You want flexibility at the input stage. But you mentioned something that was very interesting. You talked a lot about engineers and data scientists. Can you talk a bit about the role of talent? I think the different types of talent that we are bringing in farmer. I worked also for years now with tech people. And for many years, farmer was not necessarily the place that will get tech talent excited. They will go to to to meta to Google, etc. I think things are changing. That is my perspective that we start to see some of the top talents, even within the large tech companies focusing on life science. And that excites me a lot because you have those super smart people who are focusing on things that I think are a bit more interesting than just developing the next social network. But that's my own bias. I don't disagree with you. I am excited to see that there is a lot of tech talent that's coming our way. I think, so let me take your question into parts. I think in terms of talent, I'm very lucky that I'm actually recruiting right now for those types of positions on my team. I'm very hot on the market for analytics professionals to join my team. And what I'm finding is I'm seeing a really, first of all, a plethora of talent out there that has a very strong technical background. And by technical, I'm talking about their coders, their engineers, their data scientists. So we have that technicality. But then they also have really strong pharmaceutical, I'll say business experience analytically. So there are certain analyses. There are certain types of projects. I'll say or studies that are pretty much the bread and butter of a farmer, like field fore sizing or a promotional model, promotional mix modeling. Some of those techniques or those models are pretty cookie cutter in terms of just, those are accepted models that one needs to know if you're going to be a successful analyst in the farm industry. And you have this mix of people that have, again, strong technical expertise. And then they have that experience. But I think what has been harder to find candidly, and this is where I think the full package comes into play, is having an appreciation of how to convey that information. So the truly talented analysts that I've met are the ones that can translate that back office work into front office presentations. And I think it's that ability. That's really, it's that quality or that skill set that really allows that back off. And that is that back office analyst to come into the front office and actually give those presentations, but also really translates the quality of their analysis. If you have the most brilliant thesis, but you can't ever communicate it to anyone. How will anyone ever know you have the most brilliant thesis. So I think to me, that's one of the areas that I find just. I'll say partially a challenge, but also partially something that when I do find those individuals, I hang on to that because they're true, they're the true unicorns. I just mentioned in terms of tech talent coming over to pharma. Again, I always encourage it. I think the sooner you can move into pharma, like in terms of let's say you spend a couple of years at Amazon or meta or something, and then you transition into pharma, the better it is because there is such a steep learning curve in pharma. There's still pharma is a unique environment and that we're so highly regulated. And we're so constrained by so many different avenues, whether it's compliance avenues, legal avenues, data privacy avenues. There's so many things that we have to constantly be concerned about that if you've been born and brought up in an environment where there may be fewer restrictions or regulations around how you can data mine, data access data, strategize. And then you suddenly come into pharma, those restrictions may seem like they're actually cramping innovation. And I think what we struggle to do, but we still have found ways to do it is we work through those compliance. We work through those restraints and those restrictions to still innovate, to still make sure we're learning about our consumers, we're learning about patients, we're learning about physicians. And again, our goal is ultimately to meet the right patient with the right drug at the right time. That's really our goal. And so that will probably be one of the most, it's not a very profound saying, but it's something that I've observed with folks who have come up and can try to make that transition into pharma. How do you attract this kind of this kind of talent because I think as you said, there is a lot of competition for this type of people who are with multiple skill sets and they are they can go and work for so many different companies. So how do you attract them and flip side is what do you think make them excited to work for a pharmaceutical company? Yeah, I think in terms of how we attract them or what my philosophy is around it is so anywhere I've ever gone, I've really been very lucky that I've always had an element of building something. I've been in a big pharma company and I'm starting a small team and building that function from the ground up or whether it's my role now where I'm really building that team and continuing to add to it and grow to it as the company grows. The way I always present any opportunities I have is look, if you want to be part of something from the ground up, if you want to be building something, if you want to have ownership and kind of feel like you can put your stamp on that. This is the right place to be, this will be the right opportunity and the right team to be a part of if you don't have that mentality and that's not what you're looking to do, then can't believe this probably isn't the right place for you and there are lots of people who are totally fine not being in that group. They're totally fine. Look, I have a job. I need to have some projects. I know ABC. Okay, great. And they thrive in those atmosphere and that's totally fine. In terms of where we're also trying to attract talent from and how we're trying to retain that talent, a lot of it is why do you want to go into pharmaceuticals? You want to be part of something bigger? You want to be formed something better and again, going back to what I said about meeting the right patient with the right drug at the right time, we always need to one thing that I hope doesn't get lost as often as I sometimes worry it does. When you're in the pharmaceutical industry, the patient is at the center of everything you do. And I think that's a very unique place for us to be in from an industry standpoint. And so we should cherish that and remind people of that. That should be the reason why people want to come and work for us and work with us. It's like help, help all those patients who don't have anyone who can give them a voice, let us help them gain their voice, let us help facilitate them having a voice. That's really it should be the love of that and the drive of that that compel people to come into the industry. And then of course, I would say it's the exciting fast-paced nature of the work we do that should keep you in the industry. I think the keeping the mission at the center is something that can get lost on the day-to-day job, but I think reminding people regularly why we do that is essential. Moving on like the last part of this discussion, I'd like to look at the evolution of the role of analytics in Pharma. So what your view on how what you're doing has evolved. You worked at the number of companies. What is the evolution you've seen in the past 10 years? I think our expectations about what we can do and what we should do analytically. There's just completely changed. As I said 10, 15 years ago, me doing a small analysis on 40 clinics talking about site alerts or what have you. That was considered very cutting edge and exciting and we were thrilled that we had that level of detail. And now it's no, we don't want that. We want to know what's going to happen. So I had given a talk about a year ago at the actual aerial society. It was very interesting because I was there is like the Pharma representative. So it was a little bit of a different bunch of actuaries in a room listening to the Pharma rep. But I remember it sharing with them that how analytics have evolved. And before we were very happy with really strong historical or retrospective analysis. But now the way the world is really evolved is we need to have more prospective more productive analysis. And I think that's really probably in a nutshell, the best way I can describe how the industry and our expectations within the industry have really evolved over the past 10 years. We want to know. the likelihood of something happening before it happened. - Which brings me to the next question. What does the ideal data driven organization look like in your view? - I don't know if I have a view of what the perfect organization is, but what I can say is I think what we're certainly trying to build and what I certainly try to share as my own philosophy is, I always want to make sure that everything we do should have a purpose. Everything we do should have a goal in mind. And obviously as we get more organized, it's looked to be a short-term medium and long-term goals. But to me, a true data driven organization is one where it's really taken in three steps. The first is there is a true appreciation of the information that we have on hand. Everyone to the best of their ability should have an appreciation of what do we know versus what don't we know. Then the second part is there should be some level of systematic review of what that information is. So whether it's in the form of systematic business reviews, whether it's in the form of reporting, but there should basically be a constant study pulse on the business in a way that is sustainable and also replicable. Because I think what I've seen often happen in different brand teams or different companies is people get very quickly diverted into different directions. They lose focus very quickly. So there should be really a calm or sustainable, fixed way of looking at the world for a period of time so that you can really start to understand trends and also kind of get a sense of stability, if you will. And then the third is a truly data driven organization is one where it comes from the leadership. But the leadership really has to embrace evidence. And it's not just data, right? It's evidence. It's what is the data telling you? What are the facts that you hold to be true? And be able to use that in a way that really drives their business decisions so that there's rationale for the business decisions. I think it's sad when, again, I've worked at companies where you'll present all of the information to the senior leadership. And then sadly, they'll say, you know what? But we still don't agree because my gut tells me something else. And that happens. And that happens. And it's unfortunate when it does. And sometimes their gut is correct. And sometimes it's not. And I think as an analyst myself, I've always found those situations to be very complicated. Because on one hand, I'm very impressed with someone who has enough confidence in themselves to say, look, I'm going to go with my business acumen on my own. At the same time, it's frustrating when you put a lot of work into something and you truly believe, look, I've really thought about this. And there's a lot of evidence behind what I'm saying. But to be honest, that's a very rare circumstance. Thankfully, hopefully, it doesn't happen that often, but it does happen. But yeah, I would say those are probably the three kind of stepping stones or building blocks of truly building a data driven organization. All of them are equally important. I think leadership is incredibly important. But having a level of focus and really study, a study view of the business, and having a level of democratization and visibility into what information we have, those three all go hand in hand. Looking into the future, you mentioned in the past, getting a view on the past was already good. Now you want to be able to predict things that will overmost likely to happen. If I ask you to look into your crystal bowl, what trends do you foresee in the next five years for AI and data driven strategy and analytics in Pharma? Oh, I'm going to make two predictions. So I think the first is there is going to be-- we as an industry are probably going to start coalescing around a couple different use cases where most companies are going to start leveraging the value of-- they're going to start leveraging the use of AI and machine learning in those. So one of them is around next best actions. So I'm sure many of my colleagues will say, "Ani, that's old news. Nobody cares about that." But I would disagree because I feel like many of the partners that I've now talked to, it seems like the ability to develop an algorithm with the data that you have to predict the nature or the behavior of a physician or another key stakeholder based on historical data, that concept, even though it may have evolved from ex-best action into something else, it's still that predictive quality we're trying to capture. We're still trying to understand what is that physician going to do based on X, Y, and Z? How can we influence that behavior or that decision if we upgrade this lever versus downgrade that lever? So I think we're going to start to see the Pharma industry really coalesce around a couple of these use cases and that's going to become more of a standard offering as opposed to something that's truly kind of new and innovative. But the second prediction is we're going to start to want to leverage talent that is comfortable with artificial intelligence more so than talent that is not. And what I mean by that is I hear all kinds of stories. I personally don't feel this way, but I hear people who are concerned, oh my god, AI is going to take away my job. And then I hear people who are like AI, let's say AI. So for me, it's, look, it's probably a little bit of all of it. There are certain aspects of people's jobs that probably should be eliminated if we're leveraging AI properly. But part of me is thinking those are the parts that hopefully make sense to eliminate because they're the inefficient or the kind of grunt work sides of your job that it's great if you can leverage a tool that cuts that time in half or by a quarter or what have you. But I think bottom line is there is a new generation of talent that's coming into the market that are far more agile, far more excited, far more comfortable, like leveraging these tools and leveraging the art of the unknown. And I think that's going to be where we start to see the future of pharma going. Where the more you infuse pharma with this younger fresher, Candelene just more open to the art of the unknown talent, it's going to bring a different mindset. And ultimately, they have a different outcome than what happens in pharma. Last question. So what advice will you give to a young professional who wants to follow in your footsteps and come into the world of analytics in pharma? I think the first thing is I would say be open to all opportunities in pharma because I happen to end up in commercial. But I could have very easily ended up in R&D or in clinical development or some other area. And I think it's the more you learn about pharma, the more you probably can start to understand where you gravitate towards. But if you're brand new and looking to get into pharma, I would say be open to all possibilities. And I think the second thing is really just be curious. Continue to ask questions. Don't ever feel like a question you're asking is a dumb question eight out of 10 times. I will find that it's a question that everybody else wanted to ask, but didn't have the guts to ask. That's a great advice. I'm actually telling my daughter, this thing all the time, that I ask a question because most probably everyone has it, but we are afraid to ask. Exactly. All right, we're aligned. 100%. Any thanks that was great? Learned a lot. You're great at explaining a lot of complex concepts in a very easy to understand ways. People will love listening to this episode. And yeah, again, I hope we continue with Stay In Touch and maybe we'll have some other discussion in the future. I would love that. Thank you so much for the invitation. I had fun.