.jpg)
Make An Impact Podcast
Make An Impact Podcast
Harnessing Social Data to Revolutionise Medical Outcomes with Miranda Mapleton
Can social data revolutionise healthcare? Join us as we uncover the incredible journey of Miranda Mapleton, CEO of White Swan, a charity that's leveraging anonymised social data and AI to transform patient care. Discover how a personal story involving the a friend's sister's rare condition ignited a movement to improve health outcomes through innovative technology. Miranda walks us through her transition from a marketing executive to leading a team of 130 passionate volunteers, all driven by the mission to make healthcare more effective and accessible.
We'll explore the ground-breaking work White Swan is doing in collaboration with universities, charities, and commercial organisations, and how their bespoke projects and reports are making a real impact. Learn about the rapid advancements in AI, specifically Large Language Models, and how they're enhancing data processes to benefit patient care. Miranda also shares White Swan's future ambitions to democratise their data for academic research and expand partnerships, emphasising the critical role of client feedback and returning clients. Tune in to be inspired by how data and technology are paving the way for a healthier future.
Hi, I’m Heidi Fisher, the host of the Make an Impact Podcast. I'm an impact measurement expert, passionate about helping you make a bigger impact in the world by maximising the impact your services have.
I can help you to measure, manage and communicate the impact you have better to funders, investors, commissioners and other stakeholders, and to systemise your data collection and analysis so that it frees up time and doesn’t become an additional burden.
I love helping you to measure social and economic impacts, including Social Return on Investment or value for money assessments, as part of understanding the change you make to peoples’ lives.
You can get in touch via LinkedIn or the website makeanimpactcic.co.uk if you’d like to find out more about working with me.
so welcome to the make an impact podcast. I'm your host, heidi fisher, and today I'm joined by miranda mapleton from white swan. I'm delighted to have her as my first guest for season eight of the podcast series. Um so over to you, miranda, to do an introduction and tell us a little bit about yourself, if that's all right, please.
Speaker 2:Hello and thank you for having me. So my name is Miranda Mapleton. I am the CEO of White Swan, which is a registered charity specialising in social analytics. That's, taking large volumes of social data and using AI and natural language processing to make sense of that and in the service of improving health. So we focus on health as a sector, so that can be accelerating diagnosis, improving the effectiveness of treatment, treatments or ultimately, preventing ill health.
Speaker 1:Well, it sounds fascinating. When I heard you speaking a few months ago at Digital Health, I was like, wow, this, this sounds amazing. Um, I'm very interested in in the use of technology in health services. How did it come about that you started doing this work in the first place?
Speaker 2:So my background's actually in marketing, so I was a director of PepsiCo Mars, so very unrelated. And then I went to e-com and a friend of mine, steve King, who's the CEO and founder of Black Swan Data, asked me if I'd found White Swan. And then he told me a story which I found incredibly inspiring, and that was that his sister had been undiagnosed, living very, very. She was a very poorly girl for about 10 years in a wheelchair, and doctors were at a loss to explain what was wrong with her. She'd get progressively worse and they said to him and his family you need to prepare yourself to lose her. And he couldn't accept that prognosis and wanted to use Baxon's technology to see if they could work out what was wrong with her. So he did that with his team and surfed millions of social conversations looking to match her symptoms with those of others. From that work came to two rarer conditions she hadn't been tested for, one of which was Parkinson's dystonia. She got tested for it with her doctors. That's what she had and once on the correct medication she was out of her wheelchair and living a normal life. She's now doing CrossFit Championships, I think most lately in the states.
Speaker 2:So, uh, just an absolute transformation. And what he said is look, this is the power of this data, but it's currently hugely underutilized in healthcare. Would you found a charity for me to to use this for good? So that's what I started with white swan, initially with just a group of volunteers. About 30 volunteers joined me and now we've got 130 volunteers, a trustee board, a full-time you know team of five, and it's just completely transformed as an organization over that that last seven or eight years. What we can achieve in healthcare.
Speaker 1:It sounds incredible, I mean, I I love her story. Um, it's amazing, um, and I think for me it's the fact that this data was there, but you just weren't. It just isn't being used. So how, how does it work in terms of of that, the use of that data and getting that data?
Speaker 2:so we uh. So we have access to about 565 000 sources of this data and this data is all anonymized, so we don't deal with any personally identifiable data. Everything's anonymous and shared on public forums. So this is publicly available information, but the challenge comes in collating it, curating it and making sense of it, because it's very noisy and messy as an instructor data set. So we have our own proprietary platform called Million Minds that we've developed over the last five or six years, which allows us to pull in about 400 million conversations across this 565,000 sources, and then we take a subset of those conversations dependent on the project we're working on, and we utilize still Black Swan's leading edge technology, so we use some of their AI algorithms, things like that. So that might be to clean the data, get rid of noisy, irrelevant data, etc. And then we've developed our own proprietary tools as well a knowledge graph in order to make sense of this.
Speaker 2:So what we do is we have a huge wealth of data available to us, but we take a section of that. Depending on the challenge, we curate it and cleanse it, and then we analyse it to answer key research questions for our partners and we have partners across lots of different sectors now. So you know, from the university sector, people like University of Exeter, university of Birmingham. From the university sector, people like university of exeter, university of birmingham, uh, the charity sector, people like the bishart foundation, alzheimer's society, um, right through to commercial organizations like bear and and and other partners who work with many other health organizations, like the kinetic group. So it's um it. We really work with anybody who wants to, who shares our goal of improving health outcomes.
Speaker 1:Wow, and I love the fact that it's four million conversations. I think that figure just scared me thinking about trying to actually analyse and sort through that much data. You definitely need technology to help with that.
Speaker 2:Yeah, it's a big number. I mean, some of our the projects vary, if you know. You can have recently we did one with 14 million conversations in the area of rumination and worry and that was quite a messy data set. You know it's quite complex because you have to understand the context of conversations to understand whether it's a relevant conversation or not, and that's where NLP or natural language processing can help. But it another project might involve 5,000 conversations and you know you've still got to have the same attention to detail, analytics capability, but the starting volume is very, very different.
Speaker 2:So it's it's it's full of it's, full of challenges and interest, but certainly the flexibility of this data is incredible.
Speaker 2:So if you think that anything that touches the patient, whether that's in patient care delivery so someone we work with rural miles and hospital, for example or whether it's in design and development and reviewing medications, or whether it's understanding why people don't stick an idea to patient you know medications or maybe why they don't get tested for particular cancers Really in this data there's a wealth of insight because people share real time, in their own words, without a filter on it, without being asked the question.
Speaker 2:They just share what they want to share and by doing so you can get the deep whys behind behavior and why things happen and why they don't happen and what the unmet needs are.
Speaker 2:And it's because of that that you can really draw insights that you would never get through a more traditional methodology and and the healthcare industry right now is pretty traditional in most of its healthcare research and as a result, it doesn't necessarily get to these deep whys, because you either have impersonal kind of quantitative methods where you cannot, you know, set the structured questionnaire once the questions are set, you know people don't also remember what they did or how they thought or what they felt in that journey, necessarily if you're asking the months on, or you have a few, a very small number of people but highly engaged through a patient group, which again has got its value.
Speaker 2:But you need to understand that that's not necessarily representative of all patients and it won't help you get to, for example, seldom heard groups or people who can't go and attend a focus group at 10 o'clock in the morning on a Tuesday, for example. So it's got a lot of potential and that's really why, as a charity, we're on a mission to help the patient's voice be heard by bringing that to life and making sense of it for organisations.
Speaker 1:Yeah, to me it makes sense to use this data because it's like you say, it's the real data that's telling you what's really happening. Have you ever found anything completely unexpected from having done this analysis?
Speaker 2:there's lots of stories I could tell you actually. I'll pick a couple of things that I found really interesting. So one was with the royal marston and this incredible hospital. We worked directly with the chief nurse and they want to understand how they could improve their hospital service further. Now Royal Marsden, because it's so renowned.
Speaker 2:Most of the patients that go there feel very lucky to be treated to the Royal Marsden and they treat their consultants in a bit of a godlike way in terms of you know, you're going to hopefully cure me because, understandably, they're you know poorly and they are worried about their futures. And so in their traditional service they really didn't get any any improvement feedback, with the exception of possibly parked car parking because of where they are in London, and they were desperate to say but you know, I can't believe everything's all brilliant, what could we do better? So we dug into data and, because of this relationship that the patients have with their consultants at Royal Marsden, they didn't ask the Royal. And what could we do better? So we dug into data and, because of this relationship that the patients have with their consultants at Royal Marsden, they didn't ask the Royal Marsden consultant or didn't feel comfortable asking them about clinical trials happening in other hospitals, and so that was one of the insights we surfaced say it's because of this relationship. They feel that being almost unfaithful to their consultant by asking about what was happening outside of the role models and that might be relevant to them. So you need to signpost that better for your website experience. You know, lead your consultants, have that proactive conversation with their patients, and that was just a really insightful recommendation from us that they that they raised.
Speaker 2:Another one that I found really interesting was we work with University Hospital Birmingham on a rare cardiac condition called hypertrophic cardiomyopathy. Now this is a it's a tragic condition because it often strikes young, fit and active people. It's the it's often you see those stories about a young footballer or a marathon runner that mid-marathon and they have a heart attack and you know, just literally out of the blue. This condition is often behind those sorts of stories because it's very rarely diagnosed before. You have those incidences and they wanted us to help them improve the diagnosis rate. So that that involved us and in fact, that we came to the project because one of our volunteers lost his own brother to it and he said to us could we do a piece of work on it pro bono.
Speaker 2:So we did and what we found during that work is we used social data to identify the very earliest symptoms that were often missed. So this came out of conversations, often from you know, loved ones. Obviously there are people who survive this condition, who do get diagnosed and who are able to have the right mechanisms in place should they suffer a cardiac arrest. But these earliest symptoms we quantitatively map those together and that had never been done before. But we also identified that sports teachers were a key and sports coaches in particular were a key audience that if they were better educated to spot these very small differences then they could help drive an earlier referral for diagnosis, because they see multiple youths going through the same level of stress in terms of physical stress. To spot the small differences that happen in their reaction to that stress level, that could have helped to accelerate that diagnosis through that stress level that could have helped to accelerate that diagnosis.
Speaker 1:Wow, I'm just in awe of the capabilities of this information and what it can actually do and change. Obviously, you're working on lots of projects and doing lots of great work, but I'm just like why't more more health services use this?
Speaker 2:Well, I guess I mean, first of all, I think people need to be aware that we exist and that this data has this capability. Some organizations have had a fairly poor experience of what I call traditional social listening in the past, which has been fairly basic. Often you know pre-done reports where you can just learn a certain number of things and they just roll out the report time after time. Everything we do is bespoke, every project we do is bespoke, every report we do and piece of work is bespoke. So I think the depth we go into and the use of our analytical tools analytical tools is is second to none, and I think if people have had a bad experience of social listening in the past, they don't necessarily realize that this, this is actually the power of the data, if you use the data in the right way.
Speaker 2:And obviously you know, as a charity, we do as much as we can pro bono, but we also have to, you know, fund the team and so we have to charge for services, which we do for most of the organizations, certainly any commercial organization, most of the organizations, any, certainly any commercial organization. And as a result of that, we, um, we need to uh, apologies if you heard a little minor woof in the background, then that would be the labrador, um. So, uh, yeah, we have to, you know, charge those services. Obviously it needs to. They need to um see that the they get the value which we're absolutely sure they do out of the work yeah, definitely, definitely.
Speaker 1:Um see, you mentioned earlier you've got a team of 130 volunteers. Now that that sounds like more than a full-time job managing a team of volunteers, but what brings people to to want to volunteer with you?
Speaker 2:so we have a. We have a real range of people who volunteer. They they tend to want to use their skills for good. So obviously there's lots of different ways to volunteer, but these a lot of people we work with as volunteers are specialists in their area and what they love the idea of using they like to use their specialist skills for good. So we might you know, you might want to volunteer helping a local food bank or maybe, you know, supporting a local charity event, but these people love the thought of using their brains and their capabilities in a way that does good. So often they come to us through either the black swan network, because we're still we still benefit from their technology so that anyone who joins their organization is automatically told about the charity. So it's about half our volunteers 40% of them come via that route and the others are people that I meet when I'm presenting. I recently met a wonderful, you know, trainee doctor who saw me speak at the same conference. He did, and then came up to me and said, can we have a coffee? I'm really interested in your charity, and then he's now become a volunteer for us. So then that's that personal network, or, you know, contacts, of contacts that hear about us and want to get involved, and we have a.
Speaker 2:We have a very much um, I guess the way that people volunteer. It can be quite ad hoc, it could be ongoing. It depends on their capacity and capability at any one time, and we accept that life happens and people's availability flows. So all we ask is that if you're going to commit to something, please see it through, because that's the only fair way for us to run efficiently as an organization and we are very probably led by my standards we deliver everything we say. We're going to deliver always. So therefore I expect the same with the people that volunteer for us too.
Speaker 2:But we accept that life ebbs and flows and therefore there's times that they can give more and and sometimes give less. So we get some people who use their work volunteering days to do, to do pieces of work. Other people just take on a project and we'll give them, you know, a certain time frame to do it that they could they agree with, and then we check in on a regular basis. Others just take specific tasks. That might be. You know, can you research this topic area for us, or could you clean this particular data set or tag it up for us. So we, we, we find that they come to us because they want to use their skills for good and I think, like every volunteering organization, like that, people have to feel valued and respected and wanted, and we do our best to make everyone feel like that in the charity that gets involved with us too brilliant I was.
Speaker 1:I was just curious. There's like 130 volunteers sometimes when I've had like five. It's been. It's been a lot to to manage um. But um, one thing I want to to touch on obviously you're using um data technology, um ai is now the the hottest topic in the world.
Speaker 2:How does AI impact and influence what you're doing in terms of your work? So it's changing so fast, as you say, and we found even in the last 12 months, the way we can use AI is fundamentally changed with LLMs coming in into mainstream. You know availability. So we what we have found with that as let's use LLMs as a good example of it is that they've been fantastic for augmenting some of our processes and speeding up some of our processes. But they don't but they only are as good as the data they're sitting on. So we've spent a lot of time making sure we've got a really good data set. And then, where we've used LLMs, it's in order for us to make better use of the data set we've got. So we use it for cleansing our data. So that's been really helpful. I mean they've become so much more sophisticated with the ability to understand context in conversation that it's helped us speed up that process of getting from, say, 14 million conversations to the four that really matter and particularly to answer certain questions over like we want to answer.
Speaker 2:We want to know the conversations that are appropriate for answering this question versus the conversation you know. The ones that are appropriate for answering question B, all might be within the same project we're looking at. So we use it a lot for cleansing and curating. We've also used it for clustering analysis, for example. So how do we cluster data sets and create and look for themes in conversation? So we might be able to say here's the top eight things people are talking about for this particular condition or this care pathway.
Speaker 2:So we've been, we're experimenting all the way and what we found that, however the technology is, you've got to have very intelligent querying capability of the data as well. So can you ask the right questions? Can you and not just ask the right questions can you phrase the question in precisely the right way to get the right outputs that you need as well? So, uh, and like I said, we're still learning, we're still experimenting. It's a constant, you know, move forward as the technology changes. So that's just a couple of the ways that we use it now yeah, sounds good.
Speaker 1:It sounds like you. You still got the the human um quality check in there yeah, absolutely.
Speaker 2:What you can't do is just say just tell me what the answer is, because that wouldn't work and, I think, because we have we we use it to clean data, but we don't use it to do the analysis of the data as to what is the messaging, what do people actually think? And also, because our data set is based on true, real people. It's not semi-synthetic profiles or synthetic profiles, it's actually real people's posts. That also helps because then we can, you know, we can look at all of that data and uh, and also share that data, uh, in terms of examples of it, with our, with our partners.
Speaker 1:It's and like I keep saying it just so. It excites me so much how you're doing this because, um, I think that there's a huge gap in for me doing impact measurement and evaluations. This is like a piece of data that's very often missing in terms of the analysis, because people don't actually think that they can get the quality of data from this type of approach.
Speaker 2:Yeah, and I think that's where we have to educate people on the possibilities and power of this data, but also be honest about the limitations of it too. So it won't answer every question for you, and I think, as a charity, we don't mind saying that we give a free volume check at the start of any project to any partner. So the first thing we do is we say let's check the volumes on that and give you and make sure that there's sufficient data there to answer that particular question you've got. And sometimes volumes can be lower for all sorts of reasons, but it might be.
Speaker 2:If you're very, very specific, okay, now I want to know what young people who live in a seaside town which is five miles from a city, for example, then you're probably going to find that it's quite hard to identify those very particular conversations.
Speaker 2:But if you said, well, actually what we want to know is what you know minority groups in these three biggest categories, think you know about a big topic? I don't know, diabetes, for example, or cardiovascular care or something significant then we'd say do you know what that might be possible? Let's have a look at it. So it's uh, it's really understanding that the benefits of it is you get rich, powerful, depth of insight across a much broader spectrum than you would have everything else. What you won't get with it is is you know, we can't tell you. 10 percent of people are definitely of, the responses are definitely like this and 15 are definitely like that, because that's just not realistic. But you will get far more uh, you know, depth and insightful. Insightful uh, comments because they because they come in real time when people are really experiencing things, and it's the stuff you don't think about or talk about or remember after the event.
Speaker 1:Necessarily you're down the line yeah, that's always the way when you try to look back and remember how you felt. Um, it's very grayed over. It's like now.
Speaker 2:If I think back to the, the pandemic and and um, yeah, my my feelings about it are very different to how I felt during that moment, absolutely absolutely, and that's human nature and I think that's the benefit of catching the insight in the moment, that you don't get you know after the event, and it's often why a lot of people, when they go to their doctor with a problem, they struggle to remember everything that they felt at various times over the period of running up to having that appointment is why the social data was really powerful in diagnosing Julie, because actually that allowed her to capture and she kept a little diary of how she felt along the journey and when we were looking at conversations between the other people were having. You're also looking for connections for their in the moment insights and feelings and what helped them and things as well.
Speaker 1:So so, um moving on um, one of the questions I do like to ask is um how do you measure your own impact? Um so as, as someone that is impact obsessed, I'm always curious about how organizations go about measuring their own impact so we, we have a number of metrics I guess we use.
Speaker 2:We do measure the number of projects we deliver, so we use that as one metric of like. Where have we delivered sufficient, uh, sufficient volume of work really? Ie, you know, because every project, although they could be bigger or smaller, the fact is you're you're impacting a population of a patient, a patient population, in the context of that. So we have a number of those per year that we say we have to do. We obviously have some financial impacts, you know measures just to make sure that we can continue the team as well in terms of um, continue our good work. We have a pro bono only um metric.
Speaker 2:So do we do at least two pro bonos projects a year? So we, we do it in that way and then we rely on like no, we rely on feedback um as well. So do we get right testimonials from our clients or do our clients want to work with us again if we've done a really good job for them? Because we always think, well, if we've had the right impact with them and they're really happy with the insights we've delivered, then that will lead them to work with us again. I'm very happy that we have people coming back to us, sometimes several years when they've moved to a different organization, and then say, oh, we'd love to work with you again and we remember what you did five years ago for us. Could you work with us again?
Speaker 1:yeah, I think that's always the best way to to have people coming back and working with you again. It definitely shows that what you're doing is working. So my final couple of questions, and the next one is just around what does the future hold for White Swan? Really Anything you want to share?
Speaker 2:Nothing confidential, please anything you want to share. Nothing confidential, please. That's. I know what we, as I mentioned, we feel that this data is currently hugely underutilized in healthcare, so we want to champion, empower and deliver patient-centric insight using social data. That's our ambition, and to do that because we everything is about impact in the charity sector, like that is what we're here to do. We want to try and give more people access to the data sets to enable more value out of it. So right now, the vast majority of our work is we do a piece of work for somebody and then they take that forward.
Speaker 2:What we'd like to do and we think this could be particularly powerful in places like the academic sector is open our open the platform a bit further to give an interface for them to query that data themselves. Now, it will have to be a querying function because we can't hand over the raw data to to those partners, but we would like to increase our impact by increasing accessibility in key sectors like academia to enable that to happen. So that's one thing on our on our radar. Um, we've also got ambitions to extend our partnership significantly in order to get more. Because the three partnerships, you can often reach a much bigger audience and well, because we're here to create impact, we want to expand, expand that impact through the touch of other people too. We've got several partners now more in the agency sphere, because they then have lots of other clients that can then have an impact with. So we'd like to increase our our focus on on those sorts of partnerships to help us have more impact in the future.
Speaker 1:Sounds good. I can already think of at least one university that would love to partner with you on this. I'm also like, yes, please let me. I think for some of the health and social care organisations that we're working with, there's potentially some really valuable data that you've got um. So, yeah, um, most definitely, um would like to chat to you further about that. So, um, when you um you're talking about partners, you mentioned um academic, um academia. Are there any other types of partners that you want to kind of shout out to? If you're doing this kind of role, get in touch with us, because now's your opportunity to do it absolutely so.
Speaker 2:I mean the other, any organization that works with lots of other organizations in the healthcare sector we consider. So the other partnerships we have now are people like synetic, the kinetics group over communications. We have partnerships with them because we know they deal with lots of other clients and and stakeholders in the in the health sphere. So any organizations like that too, I mean we. The only thing we hold true to is they need to have the same goal of improving health outcomes. So if you're an organization that wants to improve health outcomes, you're interested in this type of insight and you want to come and talk to us, please do. Please reach out. It's easy to get hold of us hello at whitespondcouk or just miranda at whitespondcouk. Send us an email, message me on linkedin and we can set up a chat and see how we can we can help you perfect.
Speaker 1:I was just gonna ask you to share your um socials, but you you've done that. Um. So, miranda, it it's um been amazing talking to you. Um. I love how you're you're using data and technology and AI to to actually um improve the the lives of patients. You know, like you say, that's what it's about. It's about making a difference. Um.
Speaker 2:Thank you so much for joining me today thank you very much for asking me, inviting me.
Speaker 1:It's been a pleasure so thank you for joining me on today's podcast episode. I'm heidi fisher. I'm an impact measurement expert and I'm passionate about helping you make a bigger impact in the world by maximizing the impact your services have. I can help you measure, manage and communicate the impact you have better to funders, investors, commissioners and other stakeholders. If you'd like to find out more, please do get in touch via LinkedIn or via the website make an impact ciccouk and I look forward to seeing you on the next episode of the make an impact podcast.