On June 24th, Digimind & ICON teamed up to bring a much-needed discussion on “Reinventing Healthcare Research With Social Intelligence” to attendees during an Expert Webinar session. The event was hosted by Michael Durwin, Director of Social Intelligence & Communities. The following is a summary of the webinar along with an abridged Q&A session that took place during the second half of the discussion. For the full recording follow the link here.
Healthcare, as an industry, progresses in terms of its practices, its laws, and its technological innovations. On that last front - and specifically in the marketplaces of pharmaceuticals and information, there’s becoming an ever-increasing reliance on data and social intelligence as a means to reinvent the game and upgrade old and outdated practices that don’t help anyone. This is rapidly changing, and many companies within healthcare are at the forefront of this digital revolution; one of our most recent speakers, Michael Durwin, Director of Social Intelligence & Communities with ICON.
Michael was kind enough to share his expertise with Digimind and an attentive audience of attendees last month regarding what this revolution in digital healthcare and social intelligence technologies looks like, and the findings were worth looking into…
Social Intelligence’s Vital Role In Healthcare
The theme throughout the talk was that Social Intelligence would play a larger and larger role in various aspects of data scientists and medical researcher’s toolset of emerging technologies. This extended to things like reports made to express a wide range of patient metrics, such as:
- RSV persona’s and patient insights
- Stroke prevention in AFib
- Chronic Hepatitis
- Alopecia
- Patient feelings about medication
- Patient sentiments surrounding their treatment
- Vaccine understanding
- And much more!
The role of social intelligence within the development of these reports focuses on social data derived from online sources, like active patient community pages on platforms like Facebook, or other related patient surveys. These sources were referred to by Michael, as “the voices of patients”, and they were focused on categories like:
- Emotions
- Sentiments
- Real-time conversations online
- Demographics,
- And medical network satisfaction
Not only that, but social intelligence that utilizes polls, blogs, and digital content that see’s reciprocal engagement gives researchers better insight into just where the public is when it comes to all sorts of vitally important realms of medicine, like drug effectiveness of treatment preferences.
It’s Not Just a Gimmick. It’s Science.
Michael covered a very informative real-life example of how social intelligence is more than just a one-time trick, but an incredibly reliable and supremely accurate tool for understanding the world around us through hard data. Using social intelligence, Michael and his team were able to complex dashboards back in March of 2020, having collected large reams of data that cataloged and followed the data coming out of China regarding the emerging COVID-19 virus, before it became the publicly acknowledged behemoth of a pandemic that it is today. No guesswork or hunches went into the real-time information that fed his team's projects, something that a normal research team within the vaccine medicine market wouldn’t have been able to replicate without similar technology.
The Q&A
What is the best way to filter out data from industry peers and media outlets to isolate patient voices and general social conversations?
The Important thing is to really know how to use these tools. Different tools have different methods: some are manual and some are automatic. One of the ways we do it is we filter based on professions - though we exclude journalists. This is determined by their self identification on the Twitter accounts and on Reddit. It can also be done by channel which is another way to do it. So you can look at the comments within blogs, or just the blog author. So there are automatic ways that are built into the tools that filter by profession and by channel, and we do try to exclude any posts that are repeated. We try to focus on retweets because that can drop false signals on you, and we try to keep it on original posts only. While people resharing a post can indicate that they read, it’s not set in stone, and depending on how they do it, it can be a way for some people to disagree, and also pull a quote that throws some false signals. So we try to avoid all of that and stick to original pieces of content.
Can you share examples of how you use social listening data to inform best practices on trial recruitment?
Yes, when you’re looking at a trial of patients - understanding that there’s a higher density of potential volunteers for the trial - you see that they’re patients and that they’re willing to participate in a trial. Many people aren’t understanding that the cohort that you’re hoping to pull from for clinical trials has access to public transportation. All of those things weigh in. Talking about public transportation for clinical trials for instance - if you understand that around a specific clinic, which hopefully you use in social intelligence to help choose around a specific clinic, there’s no public transportation. And understanding that your audience might live 30 or more minutes away with a disease like mesothelioma, which typically targets older men that used to be power workers, who had massive exposure to asbestos, they can’t drive themselves. So social intelligence can actually bubble up: do they have a car? It can inform you if you’re looking at decentralized trials which are basically done remotely so you can participate in the clinical trial from home. Even those people with a smartphone, where they can access the app that ties you to the trial so long as they know how to use said smartphone. So all of those different pieces can aid and help and inform us. So anyone who’s doing any kind of scientific research understands that limiting the variables allows you to focus the test. You can’t test everybody, but when we look at those profiles of potential patients from the protocol which can include women between the ages of 45 and 50 that have had head-shoulder problems, and who are not trying to have children, there’s a whole lot of variables that they try to correct for. It’s a small enough audience that you don’t have to have variables twisting the data around. We can use social intelligence to look at that group to see if there is a dense enough number of that group for there to be patients there of a particular disease. If anyone has ever looked at their medical forms, it doesn’t give a form about whether you have a car, or if you have a dog, or if you’ve had kids, or if you’re trying to have a baby. We can get that stuff from Social Intelligence gatherings. So it really finishes in forming the 360 degree persona that we like to get for a clinical trial outside just the medical data.
Do you integrate other data sources beyond social data, such as surveys, to build out the “voice of patient”?
We do run surveys, we run polls, we just ask questions for local responses. This helps our social intelligence to be not just another social platform. We do mix that data in. We do have our own medical and traumatic data, which is where people are getting treatment. What kinds of specialists they see. What kinds of insurance. We do poll all of that together to understand that 360-degree patient, and we do mix it in together. If you think back to the 1940s, and you see how surveys might’ve been done then, not everyone had a phone in the 1940s or even the 1950s. If you’re doing telephone surveys you’re automatically creating bias towards people who have telephones. If you’re thinking of Yacht Monthly, and you run a poll in that, you’re not going to be surprised on what national sentiment is regarding something like the new blue Oreos. It’s gonna tell you the sentiments of what wealthy yacht owners are going to be. So it can create that bias. We do run surveys, but then we mix it in and try and find the median so it can’t be skewed.
Do you ever get concerns about the issue of representation when it comes to social data, when compared to traditional surveys?
Statistics are based on samples. It doesn’t matter how you gather the data - whether it’s postcards or emails or telephones or you’re standing outside a grocery store with a clipboard. You’re only ever getting a sample size with some bias, that’s unavoidable - we’re not going to be at a point anytime in the near future where you’re going to be able to get a 100% sample of everybody. So while there’s definitely an argument that when you’re looking at social intelligence you’re looking at only people who have access to the internet, this is one of those best you can do. That sounds like one of those cop-outs, but when you look at how data has been gathered over time, this is the most robust data set, in the history of data sets because of social intelligence. And when you look at the idea of inherent bias, you’re only looking at the people who can access the internet. It’s a huge group of people - it’s a much larger data set than most people have ever had before. So when we look at how many people use the internet, you’re looking at access in terms of finances, access in terms of equipment, and whether they have high-speed access, if they live in a rural area where they don’t even have electricity - you’re going to miss a lot of people. When you consider that 4.66 billion people have internet access, that’s more than half the population of the planet. That’s probably the biggest sample size you’re ever going to get, and of course it comes with bias, but it’s also pretty representational except for those people who don’t have it. When you’re looking at social media, there’s 7.82 billion people in the world, and 64% of them use social networks. So it’s a pretty massive sample. So mostly everyone with internet access is using social media, and there’s kind of a wash between the two. But this number is also exploding, especially if you’re looking at Elon Musk’s microsatellite project.
What content should a pharma company produce to diagnose a certain diagnosis, via social intelligence?
If you’re looking at something like Lupus, which usually isn’t affecting people until they’re 40. They’re more likely to be online, so we can reach them that way. If you’re looking at people with Myethisliosum, as I mentioned earlier, it’s more of a blue-collar audience which can tend to be a little lower-income, less educated, and is a much older audience that you might want to look into with some more traditional methods, like radio, tv, print, billboards. So it’s not really a question that can be answered without doing the social intelligence research to understand that audience. You’ll need to know what language to reach them, what dialogue to use, what are the challenges, and how do you address them? If you’re trying to address people with headaches that’s going to be a lot different than folks that have a rash.
.svg){kind=small}
%201.svg){kind=icon}
