Extending Healthspans: How Analyzing Hormone Replacement Discussion Can Improve Outcomes

Editor’s Note: In the fall of 2022, GreenBook’s IIEX Health event took place in Philadelphia, bringing both useful and inspiration content to insights and analytics professionals spanning the healthcare, pharmaceutical, medical, and wellness industries. Attendees found the content so valuable that we wanted to make much of it available to all who could not attend this in-person event. Don’t forget to register for the upcoming IIEX Health event now!

If you aren’t in those industries … how might you apply the learning within your own? At GreenBook, we believe that IIEX is more than a conference series. It’s a mindset. These are the forums in which the most important insights innovations are revealed, demonstrated, debated, and championed. What starts at the events drive change in our world. It is in that spirit that we bring you, directly, some of the poignant content we heard at IIEX Health.

---

Enjoy our On-Demand Video

Dive into the world of healthspans, what it means and how it impacts conversations on certain themes. Rauri Hadlington, of Relative Insights, leads the conversation on how analyzing hormone replacement discussion can improve outcomes. Join this conversation on the social conversations about hormone replacement therapies over a period of time Click to view the video (courtesy of Civicom).

 

 

View more 2023 IIEX Health content on-demand!

Whether you were able to attend, or you were not, join us online to see what was shared by some of the biggest brands, the newest startups, and expert-level researchers across healthcare, pharma, and consumer experience. Here’s just two of the amazing sessions you’ll find on-demand:

• Greg Hewitt and James Bauler speak to the innovation process behind Fuse Oncology, a spin-out of Cone Health, after a critical examination of the lag between a patient’s diagnosis and start of treatment.
• Zach Hebert speaks to how the Covid-19 pandemic has made it more important than ever to deliver the right message and how the right message can help ease the mind of vaccine hesitant parents.

Online you’ll find other fantastic sessions by speakers from Pierre Fabre Group, Novartis, Hinge Health, and more! If you want to stay on top of the trends in the healthcare industry — one of the largest spends in market research — you won’t want to miss IIEX Health On-Demand!

Not familiar with the Insight Innovation Exchange (IIEX)?

Ten years ago, GreenBook embarked on a simple idea: Could we create opportunities for market research leaders to share ideas and collaborate to define the future of insights?

If there was something new to our industry — a company, methodology, or platform — that didn’t exist 10 years ago and is now considered a “best practice” … well, you probably saw it first at an IIEX event.

What starts here will change our world!

---

Transcript

(Transcript courtesy of TranscriptWing)

Female: Now, I’d love to welcome up Rauri Hadlington from our sponsor, Relative Insights. He’s going to talk about health spans. I love the word “health spans” itself because, not just lifespan, but health span, and thinking about how the discussions around hormone replacement can improve outcomes. Thank you so much, Rauri.

Rauri Hadlington: Not to worry, and thanks very much everybody for being here. My name is Rauri, as she said. I’m an account manager at Relative Insight. We’re going to talk today about a little bit of research that we’ve done analyzing social conversations about hormone replacement therapies over time. What does Relative Insight do? Relative is an online AI and NLP based text analytics and visualization platform. I suppose the immediate question that anybody asked is why. I think that was touched on in the last session that there is so much unstructured text data out there and there’s so much value to draw from that, but it’s just not scalable. The only way that we’re going to be able to draw insights from the amount of text data that’s out there is by having a smart tool to be able to do that. The way that Relative Insight does that is by using comparison. The best way to talk about why that’s the perfect methodology is to tell you a little bit about your background.

Relative Insight was born out of law enforcement. What we did is we compared the way that people were talking in internet chat rooms, and by identifying the very subtle differences in the way that people spoke, we’re able to tell the difference between when somebody was a child or when somebody was an adult doing a very sophisticated impression of that child. We can all understand why that’s an important thing to be able to do. Should we now leverage that same comparative methodology and particularly we’re talking about the healthcare space today to help brands target their digital marketing, so make sure that they’re speaking in the language of their audiences, also to understand the differences between the way that the healthcare industry and healthcare professionals speak in order to bridge that disconnect that was touched on in particular with regards to the COVID-19 vaccination and the last slide, and then finally to understand patient experiences, and to start to understand exactly what a good patient experience looks like. What do people say when they’re having a good patient experience?

Is my clicker working? There we go. What’s really exciting is that what we’re now able to launch is Relative Health. Up until now, the healthcare industry has leveraged very general solutions against very specialist data sets in order to try to try and extract this kind of insight. So what we’ve done at Relative Health is that we re-categorized the English language specifically in that healthcare context so that you – so extracting these insights is suddenly that much more intuitive for users and adding this layer of context to analysis rather than leaving everything open for interpretation.

This is so important because, historically, qualitative patient research has been inherently biased. We rely on focus groups, we rely on surveys, we rely on in-depth interviews with self-selecting groups. Inherently, we need small populations to do that and there’s bias involved in the questions that we ask in the way that we interpret those answers. So what Relative Insight enables us to do is to look at a mass level of what people are saying organically out on online or in public conversation and start to analyze that and extract what’s most important. As I say in this particular use case, we focused on the conversations around hormone replacement therapy. There are three main reasons why firstly, we know it’s very topical so there’s a shortage of hormone replacement therapies on the market at the moment, partly driven by supply and partly driven by demand. We know that it’s a growing area and there’s lots of reasons for that. Were an aging population, people are living longer and people are expecting more, and this is where health spans come into it. People are expecting more later in life than they used to, and part of that is making them feel better despite that aging. Also, we’re seeing de-stigmatization of conditions of aging. We’re seeing de-stigmatization of people talking about their mental health, talking about their emotions, people talking about these symptoms and being more proactive in going out and seeking solutions. Finally, it’s universal. We all get older whether we like it or not, okay. Although we focused particularly on the menopausal conversation here, what we’re going to touch on at the end is that we’re seeing the same trends starting to come out in the male conversation when they talk about hypogonadism and testosterone replacement therapies.

What we’ve done here is we’ve taken language from social media and we’ve compared it over time, and the reason comparison is so powerful is because there’s been an explosion in people talking online about their symptoms. People are more likely to go online and talk about how they feel and talk about what they’re experiencing. with normal social listening, frequency-based metrics. What you see then is that everything goes north. If there’s more conversation happening, then everything’s being said more and it’s very difficult to extract what’s most important. So, by comparing year on year, what we’re able to do is normalize for the different sizes of data set and be able to tell you what’s being said more now than it was then proportionally.

Over time comparison helps us to keep a pulse on the evolution of patient conversations, helps us to remain up to date for too long. The healthcare industry has lagged behind what people are saying. We’re inherently reactive and so Relative Insight is going to enable us to be on top of that conversation to uncover new trends and perhaps most importantly to speak to people in their own language bridge that disconnect that we’ve spoken about in the last session about science and what people are actually saying.

Let’s dig into the research. Our baseline for this comparison is 2020. You don’t have to go back that far to find significant changes in the way that people are talking just for a little bit of orientation. What we’re looking at here are some topics that we’ve identified through our research, where you see the infinity symbol, what that tells us is that language feature that those words were used in this data set but weren’t seen at all across the rest of the conversation or the multiplier; so 8.7 times down on the right bottom right hand side of the slide there tells us that conversations about sex drive were 8.7 times more prevalent in 2020 than they were in subsequent years. As I say, is that proportional metric that makes this type of analysis so powerful.

What do we know about the conversation in 2020? It was pretty basic. People talked about their emotions in a very general way. They talked about their mental health. They talked about some of the basic parts of the practicalities of being on a treatment like hormone replacement therapies, taking pills or setting alarms to remind them to take their pills and so on. They start to talk on some of the side effects, risks of blood clots and, as I mentioned earlier, the impact that it can have on sex drive.

What’s so powerful about text analysis or good text analysis, at least, is that it doesn’t just give us that macro level. It’s not just telling us those words that are being used, but it’s enabling us to get right down into the context of that. So what we’re able to do is drill down into those words and understand the real context within which people are using those words. We can see on the far right there when people are talking about sex drive, it’s about the fact that they’re starting these medications and then they’re starting to experience that, and it’s so useful for the healthcare industry to understand the real problems that people are experiencing while they’re taking these medications.

Now we move forward into 2021. What we can see is that people are much more specific in terms of the way that they talk. Rather than just talking about symptoms, about mental health, about emotions in this general way. We see an over indexing of all of these words. So people are 28 times more likely to talk about being hot, 27 ½ times more likely to talk about their sleep, really identifying specific problems that are a challenge for them and starting to have conversations about how they might improve that and some phrases. So, as well as being able to dig into words, we can see particulars phrases which people are using as well.

I think this quote is such a powerful one. Somebody coming online to: A, talk about how terrible they felt as a result of their HRT symptoms, but then also start to talk about their improvements they’ve started to experience since being on HRT.

Along with this theme of people becoming more specific conversations becoming a bit more nuanced. Is women are also coming out to talk about learning. Using language around learning, using language around self-help strategies like diet and vitamins, and then this idea of gratitude and this is where the health spans thing comes into this. As people talk more, as people become more nuanced, as people become more open or vulnerable in their conversations, we also start to see them express gratitude and feel better for having come out and have had these talks.

Again, we can start to see some of the debate within that. So, a woman here talking about how glad she is to find a group of women that understand that that support network and that empowerment that comes from this kind of open dialogue.

Then the final part of the 2021 conversation was this idea of informed decision making will be coming online to inform themselves to understand what some of the risks and benefits of the treatments were and then making informed decisions and being proactive about going out and seeking those solutions.

Let’s bring it up to date and let’s talk about what people are saying now. As I touched on earlier, there’s been an explosion and conversation for this piece of work. We analyzed about two million words to put that into context. The whole Harry Potter series is a million words, so there’s a lot of words in this. But of that two million, about a million of them landed in 2022, so there’s been a huge explosion. What we see here is this emergence of the expert patient. People far more likely to give suggestions, make recommendations. People far more convicted in terms of what they’re going to do with that advice and information, having options, talking about whether they can and they will, and then sort of harks back to what we were saying about 2020 with taking pills and setting alarms, but again, much more nuanced than talking about access, talking about insurance, talking about money, talking about the real practicalities of getting the help that they need. Then finally around diet and vitamins, that conversation becomes more and more nuanced specific food groups that people are excluding, specific supplements that people are interested in taking.

Again, just to bring that home, some of those specific ratings that associate with those language features. So, what does all of this mean? Well, this is what we thought: As more people talk, as more people join the conversation, so it evolves. What were also able to see is that those underlying themes, those fundamentals from 2020 do persist throughout the conversation. We can see that through analyzing the similarities across these data sets. So, it brings the question: does this reflect that patient journey? Does this reflect that journey from somebody experiencing some symptoms seeking some help, but then ultimately becoming that advocate, becoming that patient, that that expert patient and then going out there to try and help others, and it’s impossible to do this research without looking a little bit to the future, and so this is where I mentioned at the beginning that we started to see this emergence of the conversation around TRT and hypogonadism.

Actually, one of the first things that’s really interesting about the language of this is that, although it is effectively a hormone replacement therapy, that language is almost inherently tied to the female and therefore much more likely to talk about TRT, and even the word hypogonadism isn’t really used and people talk much more about testosterone deficiencies. So, the conversation is already drifting away from that very medicalized language, but also the way that men and women speak is very different. Women much more likely to talk about the mental health, the mental side of things, and potentially the risks. Whereas men much more likely to talk about anti-aging and the divergence between performance enhancement and body building that has been tied to testosterone replacement for so long, and more about the anti-aging benefits of that. Then the male conversation much more technical in terms of numbers, in terms of hormone levels. People talking about, well, “At what point should I go and seek help? What dosages should I be taking in those sorts of things?”

That concludes everything that I’ve got to say. So, if anybody has any questions please fire away.

Moderator: Yes. There is a question in the app. That’s right, okay. Are you only analyzing English discussion? Do you also have Spanish language available?

Rauri Hadlington: We can analyze any Latin-based language. Some of the AI and NLP that Relative Insight is able to do is limited to English, German, and we’re working on Spanish at the moment but we can analyze that word and phrase language level in any European language.

Moderator: Very cool, very cool. That is all that’s in the app, Is there any other question in the room? Yes. Do you mind coming up to the mic so that we can hear you? It is an obstacle course. We can probably – hey, can I just do this? I don’t know where my team is. Pull it right up. Here you go.

Male 1: Can you sort the differences between American English and British English because there are nuances.

Rauri Hadlington: Yes. So the platform is sensitive to those kinds of differences, so it will identify differences in the way in the way that one of the big use cases that we have actually is exactly that. Companies that are successful on one side of the Atlantic  and want to refine their messaging for a new market. We spend a lot of time doing that work with our customers.

Female 1: What sources are you using to pull the data? If I’m assuming Twitter is probably a big one, I guess what are you looking into future thinking just knowing all of the events that are happening in that space right now?

Rauri Hadlington: Relative Insight is a software analysis platform so we can ingest any digital text data that you have. An amount of that at the moment is brands using social listening and ingesting that and that is what this predominantly was, but actually, use survey data is a big one for us. Review data, CX transcripts and customer calls, in the patient space, looking at transcripts from actual patient engagement so we can ingest any kind of digital text data that you have. So, fortunately, we don’t have to worry too much about what happens.

Male 2: This is a lot of data and it looks really great. In my mind as you’re going through that, I’m thinking about profiling and things of that nature from a consumer standpoint. How do you put this into action though? Like what do you do with it after you collect all of this stuff?

Rauri Hadlington: To use this conversation as a, as a for example, what you’re able to do is if you’ve got – if you know that you’ve got successful messaging in the HRT and in the menopausal space, you can track the way that that that is being received. You can understand how that’s changing and you can just refine your messaging as you move forward to make sure that you are responsive to patient groups. If you talk about segmentation for example and consumer profiling, something that a lot of our customers will do is that they will serve a whole group of their customers, of their audience and they will then break those down into profiles. They compare them against each other and they’ll start to understand that a really granular level, what the differences in the way that those different segments speakers so that they can do a better job of serving them be more targeted. So, I think the applications are endless. If you can understand the way that your audiences are speaking better, what you do with that is as I say, it’s infinite, really.

Moderator: Thank you. Oh, we have another question. All right, I think I can only take one more so I’ll come to you. Just give me a minute. Do I have – where’s Cara? Cara, I might need you doing the running? [Laughter]

Female 2: Hi. So, are you looking at a different range of trends that come out like findings or like the client has to let what they’re looking for or you come out with all the analysis of trends that are appearing in those conversations or dialogues?

Rauri Hadlington: In general relativity is a self-serve platform. So we – you will have your own logins, you’ll have your own access and your experts can go in there and run that analysis for you so they can balance between it surfacing to you what the most statistically significant differences are within the platform, and then the subject matter expert being able to go in there and extract from that what they think is information.

Female 2: But you would like to have an access directly to the software and you can just take the exact information that you’re looking for.

Rauri Hadlington: We do have a professional services division that can run analysis on people’s behalf. But, fundamentally it’s a self-serve software platform.

Moderator: I’m going to let you field one more question because we’re so grateful you’re trying to sponsor us.

Rauri Hadlington: I’ll try it on so quickly.

Male 3: Thank you so much. How are you determining connotation in terms? Like, if there was a population that was getting chills and a population that was losing chills. Are they combined together in the natural language processing or is there a delineation between this?

Rauri Hadlington: It’s good. I said I was going to answer this one quickly, didn’t I? [Laughter] There is – the platform is able to differentiate between the way that a word is used based on how it sits within a sentence. So, I’m going to regurgitate what I was told by our head of AI because it’s the quickest way that I could answer this question. if you think of the difference between a deer in a field that hops over a spring, that spring is being used as a noun and it will be identified as such, so it’ll come up as a geological feature or something like that in that topic analysis. But then if you think okay, the deer springs over the brook that’s now being used as a verb, and so that will therefore be tagged as a as a verb. It will be tagged as a movement within the platform. So, the platform is able to differentiate between the way that the word is used based on the type of word that it is. But then I suppose the other thing is the ability to drill down into the actual verbatim and for the user to then understand the context because AI is never going to be perfect. So it’s there – we’ve got that human level to really discern those differences.

Moderator: Thank you so much. This was very helpful.

Rauri Hadlington: That’s all right. Bye.

A special thank you to: