From Gut Feeling to Granular Insight: PLD in Action

Hello, and thank you for attending today's webinar. From gut feeling to granular insight, physician level data in action, presented by OptimizeRx, CMI Media Group, and Fierce Pharma. I'm John, and I'll be your host for today. Now before we get started, just a few housekeeping notes for everyone. To learn more about our speakers, just visit the speaker bio window on your screen. Additional resources can be found in the handouts window on your screen. If you accidentally close one of these windows, you can reopen it by clicking on its name in the top navigation bar. This webinar is also being recorded and will be available on demand within twenty four hours. And there will be a q and a session at the end of today's webinar, so feel free to submit your questions at any time using the submit questions window. And now I'd like to introduce Seth Chaney, senior director commercial enablement at OptimizeRx, who will be moderating this discussion today. Seth, over to you. Thanks so much, John. Seth Chaney, as John mentioned, newly, new member to OptimizerX, coming from a decade of, working inside pharma. And, my background is biomedical informatics and digital health cloud infrastructure, and most recently working with, launching cardiometabolic brands and overseeing their kind of brand management market. I'm joined today with Asya Abada, and she is a two time DataIQ most influential top one hundred, persons winner. And she's the EVP of business insights at CMI. And she recently led a team of innovators, including data scientists, engineers, and developers to, execute on decision support solutions. As well, we have Karen Hayes, who is the EVP of insights and analytics at OptimizerX. She is a recent recipient of the twenty twenty four PM three sixty elite one hundred award, And she is my co collaborator on, the PLD white paper that we'll mention. So the basic structure, the way that we've kind of formulated this presentation is to build on the, to build essentially on the structure that we had of really using and activating your PLD, getting the applications of that PLD, like frameworks, competitive benchmarking, predictive analytics on top of that once you've set that foundation, and integrating into an om omnichannel where you get those data and insights about what's driving your response. We'll go ahead and and transition into, essentially, our first question here that's gonna drive. I wanna pass it to you, Karen. What makes physician level data different from other HCT? And why does it matter more now than ever? Thanks for the question, Seth. For those of you who may not have read our white paper yet, I would suggest that you do. But when we talk about PLD, what we mean is that it is campaign impressions aligned to the physician, specifically linked to the NPI, the physician, and it will include data on the channel and also tactic for that particular promotion. So very useful physician level promotional data. A couple things to note. First, since the data is tagged and resolved to the MPI, it can be combined with other insights to really understand prescriber characteristics, as well as diagnostic treatment patterns, most importantly used to measure campaign performance and ROI, as well as to optimize campaigns. One last consideration or uniqueness of PLD is that it is often aggregated or de identified and not available at the physician level from every, provider of or vendor for, campaign activation. And this is really, really important and a tremendous advantage when you can get the campaign data back at this physician level, especially since in our case, we are able to activate and link to nearly eight hundred thousand different physicians in EHR as well as even more in in digital promotion. And why is it more important than ever to have this information or be growing in importance? Particularly because pharmaceutical companies are being asked to determine return on investment and really, money for promotion is, further scrutinized. And I think all of us want to ensure that programs are not only effective, but that we understand why that they're effective. That's fantastic, Karen. I'll go to you for this next one as well. What do you see the biggest disconnect between PLDs potential and how it's being used today inside pharma companies? I think, one of the biggest connect disconnects is that many of our clients use PLD, of course, to look at impressions, which doctors actually saw the campaign or were exposed to the campaign, and then they look at clicks. And one observation is that clicks aren't always relevant to promotions. Sometimes they may, but a lot of times they don't really reflect the actual health outcomes. In other words, what does that physician behavior or how does that physician behavior change as a result of that promotion? So one of the other big disconnects is that if the campaign did not have good return on investment, sometimes the client will say, well, I don't wanna continue it. Whereas, there are really opportunities to further analyze that data. A couple situations that we've had or scenarios where, client had about a one to one return on investment. And what we observed was that geographically, the physicians in rural and Midwest sales in areas were much less responsible, had a less than one to one ROI versus, physicians in urban and coastal cities, for example, Seattle, Los Angeles, New York, Philadelphia had much better response to the campaign. And that insight really gave the client, a good idea then of further analysis they needed to do to adjust the campaign so that we become more effective in the Midwest, as well as in the, along the coast. Thanks. Asya, I went to you, but the the end question there as well. Where have you seen the biggest disconnect between PLDs with your customers from an afternoon. I think one key disconnect is the underutilization of PLD as a strategic signal generator. Too often it is used retrospectively to validate past performance rather than forward looking to pursue engagement in real time. And there is also a mismatch of granularity depending on the channels where we can have some, data coming back from performance, at the NPI level versus the aggregated. I also have seen operational silos that are major blockers where PLD is often buried inside the analytics team and isolated from the creative field. So, so that, you know, often, what we see is that brand missed the opportunity to translate individual ACPA behaviors into content triggers. And, and there is also a a widespread assumption that PLD is only, is often used for postscript analysis, but there is a a possibility to have a continuous optimization tool across all the in, entire campaign life cycle. Fantastic. For this next one, we're gonna talk a little bit about the intersection of PLD and and breaking into broader data intelligence. Karen, I'm gonna go to you, first here. What other types of data can you integrate with PLD to help unlock the incremental value that customers wanna see from it? So that is a really big question, Seth. And that should really be driven by what the strategies and the business questions are for them. So considering that, I think we typically think about PLD and linking it to prescription data or medical claims data to see, you know, how that campaign performed, comparing, you know, a test versus a control group, and then seeing if there were lift there was lift in the positions that were supposed to be the campaign. So that's kind of the first step to looking at the types of data that's available. Secondarily, like Asya just mentioned, it's really important to understand some of the drivers of that campaign, and there's a lot of other types of data that can help you do that. So when considering, for example, the patient journey, some of the other types of data that we may wanna consider are, number one, lab data. Many of the newer oncology medications often require genetic tests in order to start therapy or in some cases or genetic tests, that help to find the condition in in better detail. So more of our clients are often looking at lab data as an outcome metric to see if, you know, ultimately, that data led to treatment with the medication or better diagnostic flow. Some of the other newer data are recorded and transcribed dictation of electronic health record. These datasets tend to be smaller and unstructured, but they really give some great insight into intent. So why was a patient started on? Why was a certain, test given to a patient? All of these insights are very helpful in the way you're saying what worked and what Next data data type is pair formulate formulary data. While our clients often talk about using this and use this in pure pair analysis, They don't always use it in these types of campaign analysis. And, really, we see that the campaign may have been very effective and that the doctor may have actually prescribed the drug, but the payer ended up having the product not on formulary or it's very difficult for that physician to overcome prior authorization. So some additional payer analytics are often required to ensure that payers are not a barrier to treatment. The other great set of data, is source of determinants of health and cancer data and Sorry, Karen. Just a second. I think I think it might be your microphone. They're getting a lot of noise feedback. I don't know if there's anything you can do about it, but just wanna pause here. We'll get a lot of Can you hear me? Can you hear me now? Yeah. Much better. Yeah. Thanks so much. So the final the final data is, as we're talking about, social determinants of health. And this type of data is really useful in understanding some of the health care behaviors and limitations of the patient. And I think we'll talk a little bit more about the patient as we move further into the discussion. But if the patient maybe has difficulty getting to that physician's office or they have difficulty understanding some of the information on the disease state or treatment, that can really hinder that doctor's opportunity to treat that patient, number one, but also for that patient to remain compliant on therapy. So that summarizes some of the key types of data that are available, for these types of analysis. Love that, Karen. Thanks for that. And, Asi, I'm gonna touch briefly on the on this as far as connecting to broader intelligence. Can you talk a little bit about the DTC side of that? Karen heavily weighed in on the HCP side. Hoping you can shed some light there on the other side. Yes. To Karen's point as well is that we are now able to really create what I call a multilayered health graph by integrating PLD with patient level claims, digital engagement, behavioral intent signals, EMR data. This will allow the brand to move beyond the exposure matrix, right, to understand HCP in a context, what the patient they are seeing, what are the treatment journey they are inter interviewing, and what contest direct are also engaged across channels. So we do have this full holistic, what I call the health graph. This integrated view is really something that is, going to give a three sixty view and layer the foundational work in term of, providing the foundational data structure in order to create, a little bit more predictive analytics, that would could be based also in, patterns and behavioral patterns that have linked all these data together. We do see, that, the convergence. Right? There is, the convergence with DTC becomes more powerful because now we have really a, what I call, a golden matrix that puts all these, data signals together. So you can think also of this as not only we can add on multiple data assets or multiple data signals, the online symptom sources, the education site visits. We can add demographic overlay. But for most importantly, this health graph would allow to put HCC, DTC together to really understand the journey, in a more precise way, because then we can identify with precision the moment when both HCP and patients are aligned or misaligned for that matter. I love that. Let's we're gonna drive some action here. Asi, I'm gonna stay with you. What are some first steps after you you're able to kinda marry those two things up? What are some of the first steps that some of the people listening here can take today, to basically get more, to to basically just start using that PLD more effectively. Yes. I think, one thing that, we would recommend is start with orchestration, not just reporting. Build the PLD in your planning from the outset, not just from a KPI dashboard perspective and targeting and creative and, and, of course, the representative engagement that will be starting with the orchestration in mind between HCP, DTC, the orchestration of the campaign. How are you aligning these campaign rather than having campaigns? HCP and DTC, having different, campaign objectives, different, the the the creative need to to be orchestrated in some fashion. This requires a lot of coordination operationally, but, that orchestration is key because then we have the possibility to really hone in into the ecosystem as seen by the HCP and from a point of view of the health care provider as well as the, the patient. We also, I think, use signals such as digital activities, early adoption, prescribing philosophy to really define HCP cohort by behavior, right, not just by specialties are important. But I would say within this orchestration where we have seen, what we could, would, define this ecosystem that I'm talking about, the idea would be to develop a framework, right, that is able to sustain that orchestration from two point of view. The orchestration is happening at the campaign level. Right? This has to happen. Behind it, when it's supported by a data orchestration graph that is able then to support any new data that we can add on real real data, health determinant, what have you. And then you have then a structure that allows you to automate insight and to develop predictive engines on top of this. Fantastic. I'd like to oh, sorry. I'd like to just add on to what Asya said about that. And we'd like to think about five key types of elements for your health graph because that's really fundamental to building any type of orchestration engine. The first of which is traditional data like Asseo said. So your specialty endocrinology versus PCP versus, you know, maybe it's a multiple sclerosis subspecialty. Prescribing behavior, are they early adopters versus conservative prescribers versus loyalists versus splitters? That's always critical, but basic information, you know, of course, volume and new patient starts for prescriptions. What tends to be the trends, affiliations, whether it be academic medical centers, IDNs, private practice, and location. This traditional information really forms the kind of, like, base foundation for some of these predictive models to be able to understand the context of that purse that prescriber's behavior. The second, number two is engagement propensity. This is really, really important, and I I don't see a lot of companies storing this in a way where they can access this data in in any type of not real time, but quickly. We've done projects with customers where it may take weeks or months to put all this information together, but it's really valuable to understand engagement propensity. So what that means as you are, running campaigns to be able to standardize channels and tactics, EHR alerts, digital banners, you know, rep calls and samples, emails, all of the this data into a single repository as Asia has tested. The next is payer influence, which we've already discussed. And keeping that real time is important because payer, formularies can change over time. Presiders can shift their prescribing extremely quickly as a result of the introduction of the new biosimilar, for example. So be able to being able to track that and know what's happening at any point in time is gonna be critical to properly predicting, a doctor or patient behavior. So that's number three. Number four is peer influence and networks. And the reason why I'm saying that is because we're seeing that our clients are doing triggers and next best action at different points along the journey. So it's not only getting the doctor who is going to be the prescriber to write that brand, but it's also helping the doctor who may first see that patient, whether it be a PCP or maybe a specialist that's not a subspecialist, to be able to identify that that patient actually has the condition of interest and get them to a training specialty, and then triggering along the patient journey. So the diagnosis is not this treatment. And then the last, number five, are physician therapeutic interests. So that can be a doctor who perhaps is treating obesity and type two diabetes versus just treating type two diabetes alone. It's one example. The patient population that doctor is seeing, are they pediatric, geriatric, Medicaid heavy? Can the patients afford the medication, or are they gonna need support from, pharma company assistance programs? And finally, the preferred, format and channel which goes back to engagement propensity, but this is more just knowing, you know, where that doctor wants to receive messages and also the sequence by which, that doctor tends to be most responsive. So I would suggest that those are kind of the base foundational elements for the doctor. And then on the patient side, you want to be able to know when that patient has started medication and if that treatment is related to that physician visit, or if that doctor is even visiting that doctor or remaining compliant with the medication. So those are kind of the key metrics, on both the physician and the patient side to start your journey of, being able to do next best action and going from not just a reactive to more of a prescriptive approach. Love it. We're gonna get into synchronizing. Thanks for that lead in there, Karen. Stay with you here. What role do you see bill PLD playing in building a more connected marketing ecosystem in the next two to three years? So I think the role of PLD is it's gonna move away from being just position level data. I think for many years, we've seen clients with marketing teams that have focused on physician level and then a separate team focusing on patient level, which Asya will address in a little bit. But the challenge here is really being able to put all of that together and synchronize that. And if you go to the next slide, the next slide is just a little example. Yeah. And what this shows is that there is a journey that takes place from and across the physician as well as the patient. So while the physician is treating that patient, the patient has to decide or make that decision to visit the doctor initially with their complaints. And then from there, that journey begins whether it be with a PCP or a specialist all the way through treatment. And so the idea here is to really coordinate and collaborate not only omnichannel to the physician, but also to go across and coordinate and activate patients in a way that's going to facilitate that journey. In other words, the patient is gonna be prepared to visit that doctor. They're gonna understand their condition. They're gonna understand the potential treatment and what questions to ask that doctor and have a very productive, not only visit to the doctor, but really a productive interaction or engagement with that doctor. So both the doctor and the patient feel comfortable with treatment. Love that. Hasey, I'm gonna kick it to you here. Why is it so hard to synchronize these campaigns today? What are some institutional or structural barriers that you've observed, and, what what can be done about them? To what's the role for the PLD to Karen's point also. I think PLD is fundamental to within the two, three years to two aspect, real time orchestration. These two keywords, right, are not happening today. And that with identity resolved, infrastructures, we can really connect the between exposure engagement outcomes. Right? But I would say, we have today within CMI solutions to to look into these aspects already with next best action across HCP and DTC's touchpoint. But what we wanna convey for two, three two, three years coming, and focus on or recommendation is to be ahead of the pack is really to understand that I think everyone who has been working with data understand that, it's not just more about data. It's really about intelligent action that the data can provide. And, therefore, to your point, synchronizing what we see between synchronizing HCP and DTC campaign is an operational, aspect that needs to be taken out. And the challenge, that we've seen is different team, different budget, different agencies sometimes, running in parallel and not in sync. And then there are some planning that can be in isolation. And having a a a coordination at that level first is is, paramount to the success of the synchronization in understanding also the patient journey versus the HCP. Yeah? Because often, we can see, okay, this is the journey this is the HCP journey through digital channels. This is the journey of the, of the of the patient, but in integrated way is important. So a unified map, to fix this need of the treatment journey. I think sharing the signals across HCP and DTC audiences as as ex as also touched by, by Karen. And, finally, a data driven, operation operating model, that allow this omnichannel sentence orchestration is key. So this means what? It means, if I have to summarize it, it means that rethinking the workflows, rethinking the tech stack when needed, and rethinking the KPIs as well. Because now if you are talking about the orchestration, then those KPIs that were in silos need to be remapped to really handle to address, the health graph. Fantastic. K. Push it here. K. Let's let's look ahead a little bit. Karen, I'll start with you, and then we'll kick it over to Asia for the same question. How can our industry shift from a reactive to a proactive approach and then from a proactive approach to predictive? So I think we are at the phase now as Asya suggests that, many pharmaceutical companies have put together this data asset or their analysis of putting together this data asset. And they're beginning to use more sophisticated AI to build out these frameworks or, graphs as Asya suggested to be able to really activate along the patient's journey. But the one gap, or could be a gap is that, really all of the pieces need to be put together all the way from the planning to the audience to activation through measurement. The I think the one consideration is, is there a productionized approach? I see a lot of pharmaceutical companies now that are piecing things together. You know, they may have one company do a predictive model and then feed the data all over the place to try to get it to activate against that, and then it's difficult to measure. So is there and can there be a a productionized platform that makes this journey easier. Right? Can you go from your audience to activate, as Asya said, with both physicians as well as consumers and coordinating that activity? That's something we here at OptimizeRx do with our next models. We have a platform to link both the physicians as well as the consumers, to activate both simultaneously, but also in a way that's gonna, I'll call it, nudge them toward, you know, the the right treatment package. So that's really, I think, the really one of the most critical things is productionizing that activation and engagement process. Love that. Asya, same question. Yes. I I think within the if we go in term of predictive, I think we really need to the future will push us to go into the predictive. Today, we do see regression models, machine learning. I think the agentic AI will be a key, a key component on how, how this will come to life because we are talking about a predictive. So the predictive can be in two phases. The first one would be to really develop a true predictive engines, and then, second, leverage agentic AI to really power this to the next level. And this will depend into key three key components. Clean real time data is actually the key driver for such a for such a, an end over. Identity resolution across all systems is very important. And then, of course, there are a plethora of today as everyone is talking about AI. There are more plethora of engines and methods to go after the agentic AI. And I think the success will be really, anchored into the execution of these agentic AI specific to the company business call, strategy, data maturity, and the operational execution. So in my opinion, if I have to recommend go first into predictive without the real time, then you go into the real time, phase two. Fantastic. What are some other you talked about agentic AI, their ASIA. What are some other modern tools and applications that that, to be aware of in this space that could help build this foundation? Yeah. So within today, the AI, as everybody knows, is the buzzword. Ai has been existing for a long time in term of machine learning. Right? So machine learning was part of the AI. When we talk about agentic, we are talking also about the generative AI where while you are automating your workflow so there is a lot of open source out there and then the big, you know, you have NVIDIA, as a as a provider who is providing hardware where you can really, take a large language model and, get your data to learn so that you can own your own large language model or two and fine tune the large language model based on the enterprise data so that you have your proprietary information in within your ecosystem, you can develop that. From in from the agentic, workflows, you have a plethora of online, of open source workflows that you can create is and and there are many, many ones of of thumb. But the key is really, for me, the tools are important, but also in order to, be able to develop agents, I cannot stress this enough as clean real time data and integrating the data together in a proper way that is not a technological exercise or a tech exercise, but the data is coordinated from a biz from the end, with the business in mind, with the end business use in mind so that the data is not, is not connected just with with from a tech perspective. Yeah? So, that's one. I think there is an there is a a the scale by which you can adopt these, agentic, AI is also important. How are you starting to, to get them to get this different agents aligned, in order to, be able to claim, a real time orchestration, will require human understanding, how to stitch it together, and business involved, and then the tools to do so. But I think the tools are so abundant into the the marketplace. I don't think the tools is an issue. Right? It's you you we hear all the time about AI agents coming up. I think the the the key is how to link your own data to your business imperative and to your operating model. Thank you. Awesome. Asi, you said something there, and I wanna kick it to Karen here. You talked about the human element. Karen, how does the human element come in with all this, like, pro like, this proliferation of, all these, like, agentic AI and these, generative AI models and all this stuff as it's growing. How does the human element tie in? So the human element ties in in three ways. The first is we need the human element because agentic AI is great, but sometimes and, LLMs are great, but sometimes there are hallucinations. Sometimes AI is a little too, dogmatic about what it says, you know, something is or is not. And so you've gotta correct either add or subtract what it's saying. So you need a human eye to make sure that we're refining the prompts and also refining the information that gets put into the LM LLM or into the agentic AI, in order for it to give better answers. There is, this concept of what's called kintsugi, which is this Japanese idea that you have a data foundation here or in this case, pottery foundation. And, what Japanese, Kintsugi is being able to fill in the cracks of the data with either AI or human intelligence, I e, fill in the cracks with Kintsugi pottery. You see these lovely Japanese pottery, and they have painted, gold leaf to put together some of the pieces of cracked pottery. And the cracked pottery, once reassembled, looks even better than the original. It's very, beautiful in its unique way. And so the idea here is that the data is imperfect. While we want real time data and we're getting closer to that as Asya suggests, when, we resolve impressions and then link to the physician or link to the patient, sometimes we can lose data. On the patient side, you can lose anywhere between thirty and seventy percent of that data, so you have to fill it in in some way. Right? So, number one, we can use AgenTic AI to fill in data, and need that human element to make sure we prompt and edit that data properly. The second human intelligence piece of this is really being able to establish, strong guidelines for the use of AI both internally and with for external tools that you may bring into your organization, and just giving making sure that there's training and guidance for the folks on your teams that are being able to that are, actually using this data. So that's the second critical element. And then the third is really around, developing an infrastructure, a production infrastructure for being able to use this. When you are setting up your data asset, there are certain ways to set it up so that, that data can connect via API to either LLM or agent tech processes. So practically speaking, those technology considerations need to be need to take place. So the, more structured your data is, sometimes it's harder for AI to be able to interpret it. So you've gotta kind of create that balance between structured and unstructured data. Love that. Go ahead, Asya. Yeah. Yes. I'd like to to to to add on what, Karen was rightly so highlighting. I think when it comes to the AI, I think the first step is prompt engineering, but the future will require prompt engineering, yes. But then, actually, you can actually leverage your own LLM. You can you can take your LLM and term of machine versus human, I think as our CEO says also, AI is here to scale you, not to scare you. So the scale of scaling I think there's a mic. Scaling scaling AI AI is is here to scale you, and it's a tool. Understanding it's fundamental is is super important. And I would say that, in order for the human feedback to come on to the AI engine, this is call this is done by reinforcement learning that can be developed specific to pharma. And because today, you have tools today. If you look at any LMM that is available across the spectrum, whether it's Gemini, GPT four, on you know, or or cloud or anyone, they have already integrated some reinforcement learning, meaning integrating the human feedback within the, AI engine. That was done already, but it can be done for pharma for specifically into a specific area where we have PLD, we have, DTC, we have that health graph, and then we can have a a reinforcement learning, that is incorporating the feedback from the the from the expert within the field. Fantastic. Let me go. We've got one question from the audience, but before that, I'm gonna I'm gonna, I've got one follow-up question here. If you're if you're watching this, feel free to drop a question in as as we're answering this next one. We'd love to just see what you guys are thinking about. Karen, if you could redesign how pharma brands activate PLD or even any kind of broader intelligence from it, what would you have them do differently? Good question. I think the one of the biggest gaps, just a very, very basic gap, even before we get to predictive and, prescriptive, you know, LLMs and be able to have agentic AI is very simply going beyond return on investment and creating optimization models, creating that dataset, and using optimization models. One example is we recently worked with a client on a, campaign, and we made some recommendations to them on how they could be increasing and decreasing impressions for certain channels and certain audiences. Well, it turned out that the agency actually was, doing their own optimization. We're not sure exactly how, but when we ran our optimization models for the client, we saw that they needed to, adjust their promotion from one form of television advertising to another. So the client was doing a lot of CTV, but their customers or their sorry. Their consumers were really heavily into cable and and satellite, and therefore, addressable TV would have been a better option. And so we didn't see the client actually change until the very end of the campaign, which was toward the latter part of the year where they saw a tremendous increase in, prescriptions as a result of that. So what I would suggest is just doing some of the basic blocking and tackling, being able to run optimization models earlier on so you can adjust the campaign to yield better results. That's really the low hanging fruit. Love that. I'll see if you were gonna have them do anything differently. Same question. What would you have them do? I see. I'm having some mic trouble on this side. Sorry. I just wonder. So yes. Can you hear me? Much better. Yeah. Much much better. Okay. Thank you. Thank you. So I think, if I could design how we form a brand reactivate PLD from scratch, what we'd be doing differently is maybe, again, thinking about this orchestration ahead of time. And for me, it's really, work backwards, by defining the patient journey then the PLD on top of it. That what I would do. I think the, also, I would think that, today, we have seen with our clients there are optimization in place. And, I think the feedback loop from the optimization and what we learned from measurement from the PLD coming back into the planning and coming back into the segmentation of the audience from a PLD would be something that need to be systemic, not one off, not ad hoc, analysis. Love that. Okay. We're gonna kick it to a question. We've had it for some time, but we were holding them for the end. So thanks for your patience here. What ROI lift have you seen when shifting from traditional to PLD based segmentation? Is there any benchmarks or examples you can share? Karen, I'll start with you, and I'll kick it to Asya as well. Yes. So Asya and I have been talking about predictive modeling and being able to apply that, along with PLD data to do, for example, alerts or triggers based on when patients are ready for treatment. So it's really an alert based on patient readiness, for therapy, but it's physician level in that it's the activation is with the physician, and it could be, you know, through a variety of channels, primarily in this case, EHR. And, keep in mind that ROI is gonna vary a lot because it's not dependent only on the prediction, but it's also dependent upon, a number of other factors like the creative that's that's built and, the amount of impressions that the the customer may decide to purchase or apply to that one particular channel and tactic. But we tend to see anywhere, you know, from three to four times ROI to an average of ten to twenty times ROI. And sometimes if it tends to be a more, expensive specialty drug, oncology product, rare disease product, we may see ROIs as high as, you know, fifty fifty times and above. So, ROI, like I said, can vary a bit, but, you know, if you're doing as Asya has suggested, you know, the right data integration, planning again for predictive modeling and predictive use of AI in your advertising promotion, then, you know, oftentimes, if you've done all that right, you can get really solid returns on investment. Awesome. Asia, same question there. I said I, I agree with with with Karen fully. Rois are really coming from media mix modeling and how, it's not just media. It's the creative aspect of things. So PLD segmentation helps because you you are after more precise target. But when it comes to ROI, we, the engagement that is going through the creative is critical. So it's a it's a it's really a a mesh of media impressions and spend and engagement and the creative engagement, as well as how PLD's, segment have been created and targeted, and how the personalization has occurred. So we see, we see a good, ROIs across the board, so far across our brands. Love that. We have a question actually coming in from a startup, which is pretty cool. Is there a way to harness some of the AI models to find, and this is emphasized reliable PLD for us to use? So maybe someone looking to, drive some business cases, and they're hinting on the reliability of PLD. Karen, any thoughts on that? Interesting question. From a start up. Yeah. Yeah. So I think it really has to do with speaking with your vendor and making sure that if that data's like, where the data comes from, what the providence of that data is, and how that data is collected. Sometimes we have seen vendors, other vendors who model that data, Whereas in our case, we receive that data. We tag every impression. It gets attached to an NPI, and then because that impression is tagged, we get that tagged impression back, which is attached to the same NPI. So in our case, it's like a one to one of, what the physician is actually exposed to from the advertising campaign and what we get back. That may not be the case in every situation. Sometimes you won't get the full PLD back. So it's really a conversation with your vendor and making sure you understand the provenance and reliability of that data and how that data is collected. Fantastic. Asya, any additional thoughts there? Not really because they are saying to I I think I I don't have anything to add. Yeah. My last clearance, I think she answered it very well. I mean, completely aligned with it. I don't see what I can add. Awesome. No. Hopefully, that was helpful. I would encourage you, to to reach out. We'll have a slide, actually coming up next. Please feel free to reach out, directly to Karen or I to talk about, any questions you have on activating your POD. And that goes for we have a a good mix here. Yep. And, maybe the last thing. What do you, you know, Asiya and Karen, what are you guys most excited about, you know, kinda on the horizon? What are some things that you're working towards? I'll say you talked about a move that you've really spent, CMI spent the time with their customers building a foundation and PLD and getting into predictive. You just talk a little bit about that, and then Karen, I'll kick it to you as well to just chat, a little bit about, some of the things that are on the horizon here. So, we have, within first of all, we created a personalization engine that is called AMO, today that is really a fully automated and powered by AI and machine learning that allows to serve, the leverage PLD and leverage, the HCP segments as well in order to serve, the HCP, the right message at the right time. And we are also developing one for the HCP, for, patient as well. And then as we are moving into the AR realm, we do have, actually, this week, an AI tech tour going done by our CTO and our tech team, where we are looking at integrating AI across the different aspect of our business. Very excited about what we are doing there. Within the within the predictive, we're starting, slowly into a a pilot. So we are at CMI integrating AI in every piece of it, so from an audience perspective to the planning, as well as, within the activation as well as the measurement. So, more to come on that as, our products going to the market. Fantastic, Asya. Karen? So I come from Medix, which is the, consumer audience and activation company, which was acquired by OptimizeRx. And I am very, very excited about the combination of both of our assets, which has come together over the last year and a half, because we have a fully integrated platform where we're able to activate physicians and be able to identify when a patient is brand ready. We say brand ready, and that can mean a variety of things, but brand ready, ready for treatment, or ready to be diagnosed in many cases. It and those models are custom created by our AI team and then put into production in a way where we can activate both the physician as well as that patient in order to ensure that the patient and the physician come together, for the best opportunity to increase the likelihood that that patient ultimately will get the brand of interest. So that is super exciting. Our patient activation, which is called micro neighborhood targeting, which is integrated with our DAP platform, allows, in a privacy safe way, patients to be activated not individually, which is a no no, but within micro neighborhoods or, ZIP nine areas. So if we activate everyone in that ZIP nine area, the household plus that, small geography, then, you know, the individual person's not getting the ad, but we're making sure that most of the patients that are actually eligible are linked to that, doctor or nearly all the patients that are eligible plus a privacy safe, scaling are going to be activated in addition to that position. So that's super exciting. The other thing that we're doing, which is very cool, which I'm part of is, as Asi was talking about, I think both of our organizations are heading toward greater use of AI. We're beginning to use AI in an agentic way to help profile and, segment patients as well as physicians and be able to use that for more informed media buys as well as, more informed development of creative and messaging. So really aligning the planning with, the actual activation and collection of data, all to AI, and, predictive modeling modeling. Fantastic. We actually had a question come in, Karen, as you were, responding to that. I'll I'll read it off to you. Do you explore physician behavior and response? I kinda know the answer to this one because of our collaboration on the white paper. Do you explore physician behavior and response as part of the patient journey and or doctor referral networks? Karen, I'll stay with you for that. Yes. We have done both. In fact, on the referral network piece, we have worked, more recently, I would say in the last six months or so, with medical departments or medical organizations within the pharmaceutical company. As you're aware, they can't focus on conversion, but they're focused on informing doctors, you know, on some of the benefits of the product and the science around that product or the condition of interest. And so we're using these kinds of tools and platforms and proceed along that journey by, triggering or a learning doctor that are not not just the treaters, but, again, are the refers or maybe those initial doctors are beginning to see that patient with those symptoms during. Gotcha. I'll put a a quick plug on that as well. If you look at, in the structure of the white paper, if you check out measurement frameworks, we start to get into analyzing response across, channels, and then you can see, the behavior and what's really driving that response. That really helps to start, diving in, as well as, feel free to reach out, if you wanna have an expanded conversation on what, we do there as far as the behavior and the response. That is all the questions that we have for now, unless anyone wants to drop in any one last second. I'll give kinda ten, fifteen seconds for someone to drop a final question in. Okay. Doesn't look like anything's coming in. I would just like to say thanks to everybody who attended today, probably a lot during their lunchtime. Please check out resources for the white paper, or you can also download it directly from the OptimizerX site. I wanna thank specifically Asya for joining Karen and I to co present this this session and as well as the Fierce Pharma team. Reach out with any questions. Here's our emails here. We'll keep that on. And thank you, guys. Appreciate the time. Okay. Thank you, everybody. Thank you for attending, today's webinar. That is all the time that we do have for today. If we were not able to answer your question, we will do our best to follow-up following today's webinar. As a reminder, a recording of this session will be available on demand within twenty four hours. Thank you for joining us today, and we look forward to seeing you next time.