How Peer-Reviewed Evidence Drives AI Discoverability

Speaker A: So one is, if you put in a query, what is the best online AI symptom checker, it tends to say Ada Health. Now, there are bigger ones like WebMD, there are other ones that have been going for a longer period of time. There's ones from all sorts of research labs. But what all of the chatbots say is that ada's got the most scientific evidence backing it up.

Speaker B: Hello and welcome to the award winning Health Tech Marketing Show. I'm Adam Cherinis, CEO of Health Launchpad, one of the presenters sponsors. I'm really glad you're here. If you're a regular listener, thank you for coming back. And if you're new, we have a library of over 130 episodes packed with insights from B2B health tech marketers like you. And so do please follow or subscribe using whatever your favorite podcast application is, so you never miss one. And you can also access the library. And a big thank you to our syndication partners, uh, our great friends at Healthcare now radio, the 247 healthcare only radio network and podcast syndication platform. Always a special shout out to Roberta and Carol. We're so grateful for your support. Now here's a question I want you to think about for a second. If a potential buyer asked an AI chatbot who the leading companies are in your space, would your company come up? And if it did, would the answer, uh, be positive? Today we're getting something into an underappreciated area in B2B health tech. The power of hard evidence. Not just case studies, not testimonials, but published peer reviewed proof. And it turns out that in the age of AI driven search, the, uh, evidence isn't just good science, it's becoming one of the most powerful marketing assets. My guest is Paul Wicks, founder and CEO of Proofstack Health. They're a company that helps digital health companies and med tech firms build and deploy an evidence strategy that drives both credibility and also can really help with discoverability. Paul's really interesting. He's a neuropsychologist by training and a former researcher, uh, for in academia, but also working at firms like patients like me. And Proofstack has helped over 50 companies publish more than 200 peer reviewed papers. And joining us on this week's episode is my great friend and colleague, Health Launchpad's chief strategy officer, the amazing Mark Erwick. So with that, let's get into today's discussion. Paul, before we get into Prustak, your company, tell us your backstory. It's a little unusual for somebody in

Speaker A: this space yeah, and thanks for having me here, Adam. It's great to be speaking to you and, ah, Mark today. So I am a neuropsychologist by training, so it's a form of psychologist. I'm not a doctor, I'm not a medical doctor, but I work very closely with neurologists. And, um, my research training was around a disease called motor neuron disease in the UK and Commonwealth countries, or ALS in the US and much of the world after the ALS Ice bucket challenge of a few years ago that went viral. So you may know people such as Stephen Hawking who've had the condition, or more recently, Dr. Uh, McDreamy. M very sadly passed away from it from Grey's Anatomy and there would be various sports people. So it's, uh, a very rapidly fatal condition, very tragically. So most people only survive for about 18 months. M. And so, as you can imagine, the intensity of research in that field, the passion that scientists have to prove things really quickly is important. But at the same time, because there's currently no highly effective treatment, there's a lot of misplaced hope. Right? So you could imagine if you or someone you love was diagnosed with this condition, you'd be searching for information online. You might be getting inundated with people trying to be helpful. Have you tried this? Have you tried that? And it just being really difficult. So I did a lot of work on Internet tools back before it was cool. So in the early 2000s, before social media, before Facebook, I was helping to run an online community of people living with, with als. And so we'd have both people sharing hints and tips of how to live with a condition, which was super helpful because even the medically trained people, you know, they don't know what it's like to try and turn in bed when you're weak. And so sharing tips like wearing silk pajamas were absolute gold dust, but also looking at evidence. So these were the people who would be the research participants, the subjects, if you like, in clinical trials. And they had hard questions to ask of the researchers. You know, why are we using placebo in a disease where everybody dies so quickly? Why are we studying this research question? It doesn't seem very important to me. And then if someone came along with a, uh, speculative treatment like stem cell transplants, they'd all want to interrogate the evidence. So that's kind of what's brought me to where I am today. So ALS has sort of been the vehicle, but it's spread out. So as you can imagine, many, many areas of medicine.

Speaker B: Did you have a personal connection with als. Was there a reason why you focused on that condition?

Speaker A: So actually it's the other way around. I actually was really interested in autism at undergraduate, so I was fascinated by children with autism, some of whom have these savant skills where they could have eidectic memories or draw perfectly from memory, have perfect pitch. So I actually tried to get a job in autism, um, so ended up in ALS a bit by accident. In an incredible turn of fate, a close family member was actually diagnosed with a relatively rare condition of ALS about 13 years ago. So it was actually the other way around. I was in the condition and then became a secondary caregiver, applying everything that I'd learned, using all my network, to try and find a solution for a family member. So. So the answer is yes, just not in the order that people normally think that's unusual.

Speaker B: And talk about proofpoint and, you know, how the journey to get to proofpoint, it's very curious about that.

Speaker A: Yeah. So I guess all the way through, you know, science, we have tried to unpit how the world works. Where I think we've landed as scientists is this humility to say that we all fall in love with our own theories. And if you set up a good experiment, what you'll probably do is prove yourself wrong, which doesn't feel great at the time. But if you set up the proper experiment and you don't manage to prove yourself wrong, then you just might be onto something. And if you share your methods, if you share your results, if you share the limitations, and I think quite importantly, if you write down in advance of the study what you were going to do and when you publish it, you know, uh, are honest about whether or not you change the plan or you change the plan midway through, then you contribute towards a body of knowledge. And for me, this really hit home. Um, when I was doing my PhD and I was trawling through the library stacks at King's College London, going through old manuscripts from the 1800s written by French neurologists, sharing their experiences of what it was like to see, you know, a French milkmaid with this neurodegenerative disease are the symptoms that she experienced, what happened to her, uh, the analysis they did, the pathology that they had at the time, that type of thing, and thinking, gosh, there was someone just like me 130, 140 years ago sharing their knowledge. And that made me feel like now I need to share what I learn with the field, because maybe one day there'll be some scientist on Mars saying, what was this guy on about. And so that made me sort of feel, you know, it sort of shifted it from. It's my job to find the answers to. We're each just contributing a little grain of sand here and let's make sure the contribution we make is as useful, it's as fruitful as possible because we might not be the people that discover the big thing, but hopefully we're contributing towards the discoveries in aggregate.

Speaker B: And Paul, tell us about your company, proofstack.

Speaker A: So our company, Proofstack Health exists to help digital health companies, medtech, biotech, to develop an evidence strategy to take the data they may have on the shelf and actually bring it to life, particularly in the peer reviewed scientific communication world. So that might be peer reviewed papers, it might be editorials, it might be conference posters and submissions to make that go faster and smoother and into the highest impact journal that we can without taking too much time with the staff. And then the second part is once we get that publication to translate it into B2B marketing content that matters. So there's no point funding a company to just do research. That's what a university is for. The whole point is to, you know, make a claim, to make a proof point for your company and then to communicate it to different stakeholders. So if you work in say digital mental health or decentralized clinical trials, those might be stakeholders like a hospital, an insurer, uh, a clinician, a patient, a charity, and investors potentially who are trying to figure out if you are going to 100x and so trying to make outputs that speak to each of their needs is our whole service offering.

Speaker B: How did you get make the jump from essentially academia and research to say, hey, we can help digital health companies here with marketing?

Speaker A: Yeah. So back in 2000 uh, eight, I was an academic, I was, you know, slogging my way through a postdoc and I didn't find any significant findings. I spent three years doing research into a uh, condition where I didn't find anything novel. But in parallel my work in the digital world was uncovering hundreds of insights all the time. And so I decided to quit academia, move to America and start working at a company called Patients Like Me. That was taking a lot of what I had done in the online community space, but combining it with scientific rigor and on the way out my professor said, well, I guess you'll never publish again. Now I could imagine what you would do in that situation. You go, I'll show you. And so as you can imagine, patients like me published 110 peer reviewed publications in neurology and Nature and JAMA and Brain and all the places that those neurologists would read every week a little bit to say, told you so, but also because we had to prove something that was overturning a previous misconception. So, as you can imagine, doctors believe other doctors, scientists believe other scientists. We were trying to make the case that patients living with a condition like ALS, MS, Parkinson's, fibromyalgia had valuable data to contribute, that they themselves were brilliant fonts of information, of research ideas, and that they could actually do their own research. And so we felt like we had to submit it to the highest bar of evidence, to journals that we could, because then we sort of have that overwhelming proof to make our case. And so it's through those learnings, through those lessons, turning those scars and war stories and headaches and lost hair into service offerings, templates, repeatable processes that we can now offer up to many companies.

Speaker B: So just double clicking on that. So most of the listeners to this podcast are health tech marketers or B2B marketers. So what are the specific problems that you're solving for people like that?

Speaker A: Yeah, so the specific problems we hear most frequently are, uh, number one, we're very confident that our product works because, you know, we designed it very carefully, clinicians love it, we get great user reports, got great trust pilot, you know, app stores. But skeptical people demand better evidence than that. The problem is I either don't have the skills in house or I don't have the time. You know, a lot of these digital health tech companies might have a chief medical officer, a chief clinical officer, they've got an illustrious board of advisors. But writing good evidence takes time done slowly. Writing a manuscript can take a year or two, and then it sits in peer review for about a year. So you could imagine by the time it comes out, particularly in this age of AI, it is well out of date.

Speaker C: Right.

Speaker A: So what we try and do is we try and say, okay, based on what you have, who you have, who you're trying to convince, what's the quick wins here, what's your roadmap? And importantly, what are your competitors doing? So it often starts with going. If you're on a phone call with a, uh, prospect and they start typing your name into PubMed or Google Scholar, are they going to find any hits for you? They might even find some papers criticizing you. That does actually happen more often than people think. And so that can be a real starting point to go, okay, on that basis. What can we do? What can we Achieve what are milestones, like funding rounds or maybe a big conference where we want to make a splash, maybe a big partnership we want to announce. And so we structure a whole roadmap and a content calendar that you guys be familiar with, but based around evidence and where we can put those claims out.

Speaker B: Can you name any specific or give examples maybe without naming some specific companies that you've helping?

Speaker A: Sure. Well, to be specific about one, we've been supporting a company called ADA Health, based out of Berlin, London and the US over the past five or six years. So when we started working with them, uh, they had 50 doctors and scientists on staff, but only one peer reviewed publication. And one of the challenges they were facing is one of their competitors, who people might be familiar with, called Babylon, was claiming that they had this AI symptom tracker that was nearly perfect. It was 99% accurate. And I mean, nothing's 99% accurate. That's the first red flag. But the challenge is if Your competitor's saying 99% and your data looks any worse than that, then, you know, it puts a bit of a, you know, a finger on the scale. So we actually produced a study that we published in BMJ Open, really a top medical journal, comparing human doctors against ADA's AI based symptom checker and six of their competitors. And we did it with a totally blinded, double blinded, panels of physicians that didn't know each other and didn't work for the company rating these anonymous vignettes scored by another group. And importantly, they're all vignettes drawn from NHS 111 and real clinical examples where we mapped out that, you know, human doctors did best. I know that might not be a very sexy thing to say in this AI world. Doctors are really clever. I like doctors. You know, when my kid's sick, I'm afraid I call a doctor before I put it into Gemini and I think that's likely to continue for a little while. Right. But yeah, it showed, for example, that many of the competitive apps wouldn't work if you were putting in data for a child. They wouldn't work if you said you had diabetes. They wouldn't work if you said that you were a pregnant woman. And so, you know, they're massively excluding lots of people there. So on things like safety coverage, accuracy, we were able to show how ADA performed. And that study has been cited over a hundred times. Actually, Google DeepMind just published a paper and this BMJ study was one of the first ones they cited. So it's really, you know I said earlier about those grains of sand and the foundation of science. This paper developed by a digital health tech company, has informed that whole literature and everything that's come in the years, uh, since.

Speaker B: So Mark, you've been listening attentively to this from a marketing strategy perspective. As health launchpad's Chief Strategy Officer and a longtime cmo, what's your reaction to what Paul's describing?

Speaker C: So in my last role I had a number of articles that I co authored that got into our scientific papers and from my perspective we had a different challenge but somewhat similar, as you're describing, confusion versus our competitors. And I think there's two things that uh, in my reaction to Paul, one is this, like there is that concept of getting a real found solidly founded article out there that you can leverage for a variety of tools. Like you listed a number of things but like it is useful for PR purposes, is useful for sales training even to give a better understanding of, hey, this is a different view, this is a different perspective. So I totally see the value of creating that based on the experience that I have. I think the second reaction is if our listeners have never read an article like this, it's a complete different language, like 100%. Like I started writing, I started responding to similar as you, Paul. We worked with someone who was really knowledgeable, had a lot of experience in this. Um, and for me what was fascinating, like words that I use in all of our marketing collateral are banned in the scientific papers. And at the same time any fact that you have in there or any claim that you have in there need to be a hundred percent founded in order to get through your peer reviews. So like it's a different way of writing. And as a result of that, the material that you create I believe is foundational for the new way of marketing in the future. As, um, we've all been looking towards the different tools out there on the Internet, people are searching and people are finding things through Google. As that shifting towards AI search and as the AI search is going a level deeper, this is actually much more needed than ever in the past.

Speaker B: I want to shift gears here and this is uh, it's a little bit of a pivot in the conversation, but it's something that is top of mind for many of the people listening to this and that is there is a big change going on in buyer uh, behavior and that's driven by the LLMs. So what's happening is, is that particularly in complex decision making, buyers are uh, becoming increasingly reliant on chatgpt and Claude and perplexity in helping them, help them do their research. And so a lot of marketers are seeing sort of declines in web traffic declines, but also a change in sort of the nature of when a lead comes in. They tend to be a lot more better informed and a lot more qualified actually than when they were just using Google. And what we've seen, and you know, we're not alone in seeing this, is that having hard data on your website is becoming increasingly important because getting cited by the LLMs is that sort of critical choke point. I mean it's getting that citation is about as close as you can get to the being one of the top links in Google. And I'm curious whether, you know, you're seeing anybody sort of realize the importance of the type of data that you're collecting in, in that shift in behavior.

Speaker A: Yeah, so it's something I've noticed since earlier this year and it's just been ramping up and up and up. So, you know, as you'll be aware, the common LLMs of, you know, ChatGPT, Gemini, Claude, the models are originally trained on a huge corpus information that includes books, articles, magazines, Reddit, Wikipedia, et cetera in the early days of ChatGPT. So I, for example, I look after the ALS Wikipedia page. So I would ask ChatGPT three questions about ALS and it would basically parrot back the Wikipedia page. And I thought, oh, that's interesting. If someone was to hack the Wikipedia page somehow and make it saying that, you know, ALS is cured by mushrooms or something, maybe it would fool the LLMs quickly. So I was interested in the synthesis of what was going on there. So a few things have been happening. So one is if you put in a query, what is the best online AI symptom checker, which I've done on every platform in multiple countries and I ask everyone I speak to to try, it tends to say ADA Health. Now there are bigger ones like WebMD, there are other ones that have been going for a longer period of time. There's ones from all sorts of research labs. But what all of the chatbots say is that ada's got the most scientific evidence backing it up. And what's interesting is nowadays it's not just counting the numbers, it's doing critical analysis. So ADA's produced about 30 or so peer reviewed papers that they've done themselves, but also about 12 totally independent research groups have downloaded ADA 50 times and given it to, uh, patients in their waiting room without asking ADA permission, and then published those results as well, so obviously that's a bit nervous for the team, the scientific team, when someone's doing research about you without you. But it shows a bit more of a body of evidence. And just this week I've come back from M Health EU in Amsterdam and I was looking at a couple of market intelligence programs that synthesize say investor analyst reports and press releases and all the rest of it. But what I can tell is they're giving a lot of weight to top peer reviewed journals, right? And in part that's because something that's in a top journal will be cited by other reviews, it will be cited by other blogs, it will be commented on by experts, it will be commented on by analysts. So if you work in health, if you work in biotech, pharma is the big model, right? So the stock prices of a company are affected by what results they announce at asco, the counter conference or approvals get from the fda. I'm not saying it's exactly like that in every health tech session setting, but relative to blogs or M pure marketing claims that are unsourced, I believe these have more weight. Now it's hard for me to put a number to that, but just as a counter example of something that I think backs up this story, earlier this year some researchers submitted a preprint about a fake made up eye disease called Bixillomania. And Bixillomania made your eyelids turn red and it was supposed to be from holding your phone too close to your face. And it was funded by the Sideshow Bob Institute. Sideshow Bob being a, uh, character from the Simpsons. And it was littered with references to Star Trek. And at the end it said, this is a made up disease. We just made it up for a laugh. A week later, people with eye strain were being told by ChatGPT, ooh, maybe it's bixelomania. So what does that tell us? It tells us that uh, the tools are not just using year old Reddit posts, they are ingesting live scientific data. Because there was no other way that Bixelomania, which is totally made up, could have got into its answer bank. The other thing that's really important is a preprint is different from a paper. A preprint is something that hasn't yet gone through peer review. It might be going through peer review later on, but it's at a much lower level of rigour. But what that says is, gosh, if that's what the fake research is coming in, then first of all we got to be really careful that People don't flood the system with spurious findings. But also, isn't that maybe a good thing? Right? We've all lived through so much misinformation, you know. Oh, uh, my uncle on a Facebook group said, we sure were taking Ivermectin for Covid. We've all lived through that experience, I think. And so I'm actually quite positive that AI chatbots, which could have issues with being sycophantic and AI psychosis and all the rest of it, seem to be relying on science. Seems good to me.

Speaker B: Hey, Adam here. I, uh, just want to touch on something here. Everything Paul is talking about today connects directly to one of the most urgent challenges we're hearing from health tech marketers right now. How do you get found when buyers aren't googling anymore and when they're using it? The LLMs. That's what we call AI optimization. And it's a core part of the what we do at Health Launchpad. We help B2B health tech companies assess their current strategy, their current approach, the performance of their website and their content. And we help them build content strategies and optimization plans to get cited by the LLMs. And that means understanding what questions your buyers are asking, ChatGPT and perplexity, and making sure your company shows up with the right answer. Um, so if you're struggling with this issue and want to talk about it, go to healthlaunchpad.com and check out our, uh, AIO AI optimization approach. We think it can help. And then if you want to, just go ahead and book a call with us and we can have a chat.

Speaker C: May I ask a follow up question on this? Slightly on that content that's out there in the scientific papers, the articles that I supported, that I co author, they're out there, but in order to get access towards the full document, you either need to have a password or you need to pay for the articles. So that tells me it's inaccessible for the crawlers from the LLMs. In order to make that useful, have you find ways to actually recreate that content or are you republishing what, providing access?

Speaker A: Yeah. So traditionally, the way journals were funded was it was free to submit your article, and then university libraries would pay expensive subscriptions. Right. And so companies like Elsevier, Springer Nature, that's how they make their money. Over the past 20 years, that's pivoted to now authors have to pay between $1,500 or a. Really, you know, the new Nature journals could charge you $10,000 because people really want A nature publication, but then it becomes free. We're still in the midst of that sort of not being consistent everywhere. So a few things of note. I always recommend that people do pay the fee. So if you're a health tech company, you've hopefully raised a bit of VC money. When we scoop out our, uh, evidence plan that includes a budget, how much money might we set aside for this? It wouldn't be the LLMs that we need to worry about. It's journalists, it's payers, it's customers, it's potential employees who lack that, uh, journal access. And we want to make sure it's read by as many people as possible, as many patients as possible, et cetera. The large language models increasingly have licensing deals. You know, if you look at someone like Open Evidence, which is a tool for doctors, they. The reason they're so powerful is because they pull in data from the New England Journal of Medicine and jama, the full text of that. So it's not just the summaries, which are called abstracts, it's actually that whole text. But yeah, part of our, ah, sort of profiling service for customers is we try and make sure that things they do, like posters that they present at a conference, make sure you upload those in PDF form to figshare, to researchgate, that you make summaries of them on your blog, that you make them accessible, that you do webinars about them. YouTube transcripts increasingly go into these LLMs as well. So we have this saying of using every part of the buffalo. So if you spent all this time writing a peer review paper, and we will help you, but it is still going to be quite challenging to get it done. Once it's done, you don't just hit send and send it to your mum and your old professor and be very proud of yourself. If it's not a blog, a newsletter, a webinar, a podcast, uh, a handout at a conference, if it's not a poster, if it's not on your, you know, your booth, is a little citation to back up your claims. You haven't really, you know, squeezed all the value from it. And that's actually where I spend a lot of time talking to marketing and PR colleagues who have been going back to the product team or the business team, often for a long time, saying, what have we got to talk about? What's new? Best case scenario, we love doing press releases about this. And that's the algorithms love a press release. Crunchbase rates how hot you are for investors on the basis of things like press releases and they get broadcasted really easily. They show up in Google news research news results, they show up in Google search snippets and um, you know, it's a way of making the value of this go much further.

Speaker B: Talk about the process because I think for, you know, for most people listening to this sounds like a long, complex and potentially expensive process to get from hey, we need better research, we need better data to turning that into content or anything that would make an impact on the LLMs. So talk through that.

Speaker A: Sure. So because now we've done this for about 50 companies and published over 200 papers and I've sat on the boards of journals like the bmj. We have got a bunch of tools and tricks of the trade that speed this up. So first things first, understanding what data you have, you know, what stories you're trying to tell, what your timeline is like, what your team is like, that's really crucially getting to the Goldilocks zone of evidence. Right. If you're a pre seed company, you're not in the same boat as a series C company. Right. The expectations are different, your capacity is different and you may have different collaborators. A lot of people are uh, university spin outs or a lot of people have academic partners that they collaborate with and working out what they do versus what you do is really important. Where we often start with is conferences. So the reason publishing takes a long time to write is usually not the actual time to sit down and write the words. It's usually to get the eight of you around the table to all agree what it is that you want to say. I've seen that take a whole year. Right. So instead what we do is we find the key conferences where your competitors, your customers are and we find out their abstract deadlines. And guess what, if their deadline is a week on Thursday, they are not going to shift it for you. And so we get to get those eight people in a room and uh, we don't leave until we have written a tight brief. And the brief says, what is the title, what is the point? We've only got 250 words. Would better make it pretty concise. You can't throw the whole kitchen sink in there. Right. And that's actually a really useful forcing function. So we start off with that as our kind of evidence engine. We get in the habit of doing these submissions. We turn it into uh, an A0 size poster with you using your branding and graphic design. That is a whole heck of a lot easier than doing the peer reviewed publications. But it Builds the muscle, right? It builds the muscle of how we contribute, how we make this, and who makes the decision. Because oftentimes when we get in, there's not a clear decision. So within, say, your marketing organization, I presume you know what goes to a chief marketing officer. I'm presuming the CMO doesn't need to approve every single tweet or LinkedIn message, otherwise you'd do nothing but doing that. Right. We need the same thing on the evidence side. Does it all need signing off from the business officer in case our, uh, claims get challenged on a call, or does it all need signing off by the chief medical officer? A regulatory or legal oversight may be needed if you have a medical device, for example. So it's different in every case, and part of our work is setting that up. But once we've got the bones of that, you know, the fastest thing I've ever got now was 4 hours total time to get something in the Journal of the American Medical Association. So can be super fast. It can take a little while. And we use things like preprints, we use things like targeting journals that we think are really interested in these submissions to make sure that we can land things as quickly as possible.

Speaker B: Mark, listening to what Paul just said, how does that look in terms of translating that into marketing programs, particularly in sort of content development?

Speaker C: So, uh, we use it for trade shows as material to hand out to people who visit our booth. We use it as sales tools. Sales Reps would have PDFs of this that they could share. We created blog articles that we not only published, but then had social media leveraging that. Again, just the numbers that we already knew because it was research that we had before when these numbers were used in any other document, whether it was marketing or a PowerPoint presentation that we will be using, we would just use this little asterisk and now it's of a published article. And suddenly you create much more credibility. Not necessarily creating new content, but even with the existing content, you create more credibility because it's published somewhere. I thought it was fascinating how people respond towards that. In years before I had used it, but wasn't really like understanding the value. Once I started realizing that we published a number of articles that we got out there, we were thinking about the press release. We didn't do that for some other reasons, but that I think if the timing is good, I would definitely recommend, hey, do that, a press release. And particularly when you're coming up to watch a trade show or something like that, for more information, come and visit us. It's just a great asset for some people. It speaks much more than for other people. And maybe that is the audience that we're communicating to. Or maybe that's also the titles of the article. There's a wide range. And, uh, maybe, Paul, that will be good to get your perspective. Also. There's a wide range of scientific publications out there, but the perceived value also and the ability to get published in there is varying a lot. But, um, how people responded towards it was very interesting.

Speaker A: And I can say this because many of my friends are scientists. Scientists are not the best communicators. Sorry. Too much time spent peering over a microscope and looking at cells does not necessarily prepare you for explaining your job in the real world. So you sit next to someone at a wedding and tell them about your interest in zebrafish genetics. And, you know, they, uh, might pour themselves a large glass of red wine. So one of the things I've noticed about myself is I've started making my titles of my articles get a little bit more straight to the point, uh, as I grow older. So my original ALS research was about the fact that the brain is affected in als. But my first paper was titled something like Neuropsychological Correlates of cerebral abnormalities using 11C positron emission topography. After, uh, a few years of no one listening to me, I started writing editorials called It's Time to Stop saying the Mind Is Unaffected in als. And I wanted to add, you idiots. Right, so you can get to the point. We wrote a paper two years ago in Nature Medicine that could have had a much more flowery title, but it just says large language models need to be regulated as medical devices. That now has been cited 200 times and has fed into conversations with regulators like the FDA and mhra. So I have learned. Get to the point. You've just heard me say lots of words. I can be quite verbose, but one of the original things I was taught was that, you know, if 100 people read the title, maybe 50 of them will read the abstract and maybe 20 of them will read the full paper and maybe two of them will read it all the way through. So if you don't get the point across at, uh, the top of that, I suppose you would think of it as a funnel, then your message is going to be buried. Right? And so very often what we are doing is often just rearranging the key point, you know, and, um, that is the insight that often when you're translating your information into a press release or, or talking to journalists, that's what they want, right? But us scientists, we get very excited about saying, oh, we got a grant from here, or we did it in 10,000 people, or we use this clever method. Okay, that's wonderful for your peers, but what is the point, you know, to prove? Wow, we made an app that means you don't need to have 12 hours of therapy to cure a phobia. That's amazing. If true, right? The asterisk, like you were saying, Mark, does the lifting of the if true, right? And it's important to say scientists spend their whole lives arguing with one another about everything. So just because you have one paper doesn't mean it's conclusively solved, but it at least shows that it shows what game you're in.

Speaker C: Right?

Speaker A: And you know, again, here's a section where I probably won't name names, but a reasonable chunk of this sector makes, uh, bold claims and then can't back them up. Right? And you sort of go, well, if they're telling porcupies about that, might they also be telling fibs about other things like how well qualified their team are or how, how safe their product is, or whether or not they're going to jack up the prices next year even though they pinky promise they won't? So it's an expensive signal, you know, of someone that wants to invest, someone that's playing a long term game with long term people, and that in itself can be a signal of hopefully of quality or of trust or integrity.

Speaker B: I feel I need to translate for our American listeners. Porcupise is London cockney rhyming slang for lies. So just final question for you, Paul, and then I'll come back to you, Mark. So if you could give one insight for listeners to this podcast to take away from this, something that would sit with them for a while, what would that be?

Speaker A: Probably the most startling thing you can do is open up an incognito window and start asking questions about either your company or your sector in Google Scholar. Just see what's in there. I will bet you a aforementioned pork pie that someone's already written something about you that's probably wrong, right? They might have said you're a large, uh, language model, when actually you're uh, you know, Bayesian reasoning algorithm or something like that. They might be talking about your competitors. They might be saying, oh well, you know, the top five companies you need to think about are 1, 2, 3, 4 and 5. And you're not on that list. Ah, that's a really eye opening thing to do. Right. Because you tend to live in your bubble, I would say nowadays you also need to run that on, um, Claude OpenAI and maybe ask a friend of yours or a relative or someone who's not in your tech space to pose the question, how they would pose it. Because I think sometimes one of the things we find is that if you know too much, your prompt will be too good. Right. But I can tell you from the procurement side of things, you know, some of your listeners, customers might be pharma people, for example. Pharma innovation people. Pharma innovation people are going to type in who does the best clinical trials or what's the best ecoa solution. Something like that. Right. You want to know where you score on that. And I think evidence is one of the ways that you can float to the top. But it's really important to know what's out of date, what's talking about a company that went out of business two years ago. Right. Or what's got a great surface area, but everyone knows beneath the surface they can't deliver. It's really important that you have that perspective and probably plan to refresh that every quarter or so. So the thing that can help you really be honest about it is turn it into a little five minute presentation for your team, get their reactions. You know, for one thing, it's actually pretty good for in group bonding to go and throw rocks at your enemies together, but also to kind of see how does the algorithm see us? Does it see us at all? And, you know, does it position us as the obvious choice for the people that we want to work with?

Speaker B: Thank you. Mark, what's one thing that listeners to this podcast should do in the next

Speaker C: week when you build your content plan, really think what matters? I think so frequently marketeers are recreating, reusing material that's already there. And in the past, that was awesome. But now with AI LLMs being out there and searching for unique content, different content, this whole game has changed. So go back towards your content plan and take a complete new look at it. It's what is unique here? What is the added value? How do we stand out versus the rest of the market?

Speaker B: Paul, it's been so interesting having you on the show. I don't know whether I told you this when we first met, but I actually have a degree in neurophysiology. Uh, and what I did coming out of university was I immediately got a job in advertising. So when you told your story there about the titles of your papers not resonating with regular human beings. It really hit home to me. So I love that. So Paul, where can listeners find out more about you and about proofstack and if they wanted to connect you, what should they do?

Speaker A: Yeah, so you can find out more about us at Proofstack Health we have a newsletter called Proof Points where every week we sort of drop war stories and advice for health tech operators such as your listeners. And you can find me on LinkedIn where I will be dropping dad jokes and observations on AI health tech, uh, on a near daily basis. So yeah, hope to connect to you there and find out more.

Speaker B: Well, look out for the dad jokes. Paul, thank you so much. And Mark, as ever, thank you so much.

Speaker C: Thank you.

Speaker A: Cheers.

Speaker B: I love this conversation. Paul is such a smart guy and I really learned a lot from what he said. I want to share with you some of the key takeaways. I've got five things I think are worth really thinking about. So takeaway number one LLMs already deciding who gets found in your category? Buyers are increasingly using AI chatbots to research vendors and those tools heavily weight peer reviewed scientific evidence. If you don't have published proof, you may simply not show up. Takeaway number two Evidence is a marketing asset, not just a scientific one. A published paper isn't the end of the process, it's the beginning. Used well, it fuels press releases, blog content, sales decks and trade show materials, and even more. Paul's phrase says it perfectly. Use every part of the buffalo. Takeaway 3 the fastest path to evidence starts with a conference abstract. You don't have to start with a full peer reviewed paper. Conference abstracts create a forcing function to align your team, sharpen your message and build the evidence muscle fast. Takeaway 4 Audit your AI presence and do it now. Open an incognito window. Ask ChatGPT clawed and perplexity what the top companies in your space are and what the evidence says about them. Uh, you may be surprised what's already out there. Then ask. Is that the story you want your buyers to find? Takeaway 5 open access matters. If your research is behind a paywall, both buyers and LLM crawlers can't reach it. Pay the open access fee and also post summaries, posters and content to sites like Research Gate and figshare. Make your evidence findable. A huge thank you to Paul Wicks. You can find him at Proofstack Health and you can subscribe to his ProofPoints newsletter. You should also connect with him on LinkedIn. And um, thank you as always to Mark Erwick, uh, for bringing the marketer's perspective into the room. If you found this episode valuable, please share it with a colleague and follow the show so you never miss what's coming next. Until then, keep pushing the boundaries of what health tech marketing can do. See you on the next one.