Podcast – Data the new frontiers in dementia research – Exeter Datathon

Voice Over:

Welcome to the Dementia Researcher Podcast, brought to you by dementiaresearcher.nihr.ac.uk, a network for early career researchers.

Adam Smith:

Hello, my name’s Adam Smith, and today it’s my turn to host this podcast recording for the NIHR Dementia Researcher website. Our podcast is just over a year old, so we’re rather excited at the moment as later this month, or early next month, we expect to reach 10,000 plays.

Adam Smith:

We’re currently working on a few ideas of best to how we might celebrate this milestone. I just wanted to say at the start of this podcast today to keep an eye on our website and our Twitter feed over the next couple of weeks to find out more. I should also add that all this has only been possible because of everybody out there listening.

Adam Smith:

Before I introduce our topic today, I just wanted to thank everybody for listening and all those that have contributed to the podcast over the last year. As I said, watch our Twitter feed in a couple of weeks and we’ll hopefully have maybe a bit of a giveaway and a special recording with some cool people who’ve got some advice for early career researchers. Thank you.

Adam Smith:

Okay, back to the main topic. I’m delighted today to be back at the University of Exeter. For those who remember, I was here a few months ago talking to some people who are living with dementia about what they wanted from research. Today I’m back in Exeter and I’m joined by a fantastic panel who are here to talk about data and how this is being used to push new frontiers in dementia research.

Adam Smith:

I should add that this is timed because the panel are all here in Exeter this few days to do a datathon, which is working on that data and how it’s used in dementia.

Adam Smith:

Let me introduce the panel. First of all, we have Professor Richard Everson, Sarah Bauermeister, and Dr. Hadi Modarres. Welcome, everybody.

Professor Richard Everson:

Thank you.

Dr Hadi Modarres:

Thank you.

Adam Smith:

Okay. Can we start by doing a little bit of a round table so you could all introduce yourselves and Maybe tell us a little bit about your work. Maybe if you could go first, Richard.

Professor Richard Everson:

Yeah. Hello. I’m a professor of machine learning and computer science at the University and we’re a member of the Alan Turing partnership. I’m the university lead for that. I’m here really today because I’ve been working with Professor David Llewellyn on diagnosis and understanding the relationship between profiles of different dementia subtypes.

Adam Smith:

Sarah. Let’s introduce you.

Dr Sarah Bauermeister:

Hi there. I’m Sarah Bauermeister. I’m from the University of Oxford. I’m a senior researcher there. I’m also a cognitive neuropsychologist and a psychometric analyst. I’m also a senior data manager for Dementia’s Platform UK. I’m here for the datathon as Dementia’s Platform UK are providing the infrastructure for the virtual desktop interface for the datathon.

Dr Sarah Bauermeister:

DPUK is a remote access platform. We are a data repository for over 47 population cohorts, which equates to over 3 million participant records.

Adam Smith:

That’s all run from Oxford?

Dr Sarah Bauermeister:

Yes. The academic headquarters is in the University of Oxford and our infrastructure, our servers are actually based at the University of Swansea.

Adam Smith:

Oh, fantastic. Hadi.

Dr Hadi Modarres:

Hi, I’m Hadi Modarres. I’m a data scientist at Cognitivity. So, at Cognitivity we’ve developed a cognitive test for early detection of Alzheimer’s and dementia. As part of my role, I’m analysing the unique patterns that distinguish between different subgroups of patients, from healthy, to those with mild cognitive impairment, and those with dementia. To do that I’m using machine learning models to differentiate between these groups. I’m here at the datathon as one of the participants analysing the data and trying to get insights from it.

Adam Smith:

Fantastic. Tell me more. Is Cognitivity, is that a spinoff from a university or is this a commercial?

Dr Hadi Modarres:

It’s a spin-out from Cambridge University. We were spun out about five years ago from Cambridge. Currently based in London.

Adam Smith:

Oh, fantastic. Well thank you very much and welcome everybody and thank you very much for taking time away from the datathon to record this podcast with us today. I should also mention, of course, you mentioned him before, Richard, there are two people who aren’t on the podcast today who do deserve a mention, which is Professor Llewellyn from here at the University of Exeter and John Gallagher who runs DPUK Oxford with you, Sarah. I know that they’re driving some of this work by bringing machine learning to bear on cohort data at a time when obviously drug trials have failed and are being cancelled, to try and bring these effective treatments into trials. We should obviously, of course, mention those. Are they both here at the datathon?

Professor Richard Everson:

David Llewyn, here. Yeah. He’s downstairs in the thick of it at the moment.

Adam Smith:

All right. Okay. So, as we mentioned, we’re all here at the datathon. Maybe we could start by, if I come to you, Sarah, could you tell us a little bit more about the datathon? I’ve heard of hackathons before. Is a datathon a bit like a hackathon but with data?

Dr Sarah Bauermeister:

Yes. Basically the idea of the datathon is really to bring together a group of data scientists from various different disciplinary backgrounds, not necessarily from a dementia background, and to bring the data scientists together in one room where they can collaborate, and come together, and interrogate all these large data sets utilising the DPUK infrastructure, and bring their methodologies.

Dr Sarah Bauermeister:

In this case, it is very much a machine learning approach, and come together and really bring solutions to a question which has been either proposed to them and see what arises over the three days. Really it’s a very exciting opportunity to bring these researchers in one room to see what arises.

Adam Smith:

These all come through. Are they from different disease areas as well? They’re not all, say, dementia or data people.

Dr Sarah Bauermeister:

Yeah, absolutely. We’ve got mathematicians in the room, we’ve got data scientists in the room, we’ve got biologists in the room, dementia researchers in the room. Absolutely.

Adam Smith:

The cohorts that DPUK brings together, they’re not dementia cohorts, are they?

Dr Sarah Bauermeister:

Absolutely. We’ve got birth cohorts. We’ve got cardiovascular-focused cohorts. We’ve got healthy population cohorts. We’ve got Parkinson’s disease-focused cohorts. Although we’re called Dementias Platform UK, by no means are these all dementia-focused cohorts. Absolutely.

Adam Smith:

Overall, this is a three-day event. Is this the first time you’ve done this here in Exeter?

Dr Sarah Bauermeister:

Yes. This is the first one of this year and we’re launching this program of five datathons in Exeter this year. We did hold one last year in October at the Turing Institute, which was almost like a pilot datathon, which was so successful we decided to hold five this year. This is the first one of five.

Adam Smith:

Five this year. So, work doesn’t stop at the end of the three days, one would assume. You bring people together, they’re really focused, and then they forget about this and go back to their day jobs. What is it you actually hope to achieve through these?

Dr Sarah Bauermeister:

Well, we’re really hoping that by attracting these young data scientists, and we’re really hoping that this pitch will go out to early career researchers specifically, that we’ll be able to break the back of these analyses and encourage them to take these through perhaps to publication level. We give them access continuing on to perhaps three months after the datathon, we’re setting up user groups to keep this collaboration going, and to nurture these through to output and outcomes. By no means does this stop at the end of three months. Very much focused on “Let’s see what we can get out of the data.”

Adam Smith:

That’s fantastic. Everybody else brings the ideas and you bring the data and the tools by which they can make use of it.

Dr Sarah Bauermeister:

Yeah, absolutely, and then nurture it forward out of the datathon as well.

Adam Smith:

That sounds interesting. What do you hope to achieve through this?

Professor Richard Everson:

Yeah, I really think the potentially exciting thing is the new groups and the new collaborations that are going to be formed. Realistically, you wouldn’t expect to crack dementia in three days in a datathon like we’ve got here.

Adam Smith:

Oh, really?

Professor Richard Everson:

Well, maybe the fourth would do, but we’ve only got three.

Adam Smith:

If only you know worked really long hours.

Professor Richard Everson:

If only. Yeah, yeah. Well, we will be working quite late tonight.

Adam Smith:

I hope there’s beer as well.

Professor Richard Everson:

There was plenty of beer. The real thing that is happening is that there are people down there from all sorts of different diverse backgrounds, with different skills, covering the whole range of expertise around dementia, around machine learning. Bringing those people together, putting them in a room, making them work together really. They really are working hard. They were tired at the end of yesterday.

Professor Richard Everson:

They’re making friendships, they’re making collaborations that I hope are going to persist not just for three months, but for three years, past their PhDs. Only it’s really going to lead to new initiatives in dementia research, I hope.

Dr Hadi Modarres:

I think that same principle about bringing people together. We’ve, we’ve recorded at least two or three podcasts over the last year. I remember a group from Yorkshire who did exactly the same thing where they’ve got people working in cardiovascular, somebody else who works in genetics, coming together because they realized that if we just stay in our little silos of looking at dementia here, that’s not going to solve the problem because probably the chances are dementia is a much bigger issue that’s affecting so many other disease and possibly [inaudible 00:10:23].

Dr Hadi Modarres:

By organising this event, you kind of force those collaborations to start because organisationally we don’t we don’t work like that, do we? Conferences don’t bring the right people together. If you don’t go away and make those connections and force them, they don’t happen naturally.

Professor Richard Everson:

Yeah. This is right at the beginning of the process where your conferences, which you mentioned, there’s a tendency to go to conferences and say, “Look! Look at this fantastic stuff I’ve done and keep off my patch. I’ve just done this and I’m going to do some more work on my next one proposal is going to be in that area. So stay away cause I’m doing it.” This is right at the other end. This is you’ve got the data and that’s it. That means that people can start to get together around the data. It has been really interesting over the past day or so. I was surprised midday today people were coming up with results. They’ve got graphs out of it already.

Adam Smith:

Oh, really?

Professor Richard Everson:

I was a bit of a sceptic, but I’m more of a convert now.

Adam Smith:

Is it a two-way street as well? Can the group come back and say, “Oh, if only we had this little piece of information that we might just be missing?” Does it matter? I mean, does it matter that data scientists aren’t lab scientists? Does it matter that they don’t understand the ins and outs or the importance of microglia?

Dr Hadi Modarres:

Personally I’m coming from an engineering background. My PhD was in nano science. I’ve been learning about the domain of dementia whilst I’ve been at Cognitivity. I think what’s great about this kind of event is that within the team that I’m working in there are people with significant domain knowledge, but there are also people with fantastic technical skills that they’ve gained from their research in astrophysics, computer science, mathematics. I think it’s the combination of these different skill sets together as part of a team that can really yield like new insights and new ways of looking at the problem.

Dr Hadi Modarres:

I think what’s been great so far is that it hasn’t been prescriptive at all in terms of pushing us to look at a certain angle. We’ve been given a lot of freedom to approach the problem in a new way. I think that can lead to new ways of looking for solutions.

Adam Smith:

Super. I have a fixed question here that hopefully we haven’t already addressed it I’ve got written down, which is probably to you, Richard, which is: How is longitudinal data creating this? How is this driving it? I mean, how is this driving this new frontier in dementia research?

Professor Richard Everson:

Well, I think that longitudinal data is important because if you’ve diagnosed dementia only when people have got dementia, it’s just too late. This large scale longitudinal dementia gives us an opportunity to look at some of the factors which indicate that a person might be susceptible to dementia, or getting dementia, and vice versa. That gives us the possibility of developing things. You know, behaviours, or possibly drugs, or something like that to mitigate those effects. It needs to be large scale because these are very weak signals.

Professor Richard Everson:

It seems to be that this effect that humans are very good at disguising and compensating for their dementia until perhaps it’s almost catastrophic and people go downhill very rapidly. That means that, in the very early stages, it’s likely to be a very weak signal cause it’s masked. Large scale longitudinal studies which then need things like machine learning and large amounts of computation to tackle that large amount of data, I think they’re important. They’re likely to have a really fruitful area of dementia research in the future.

Adam Smith:

Can we do that with the data we’ve got now? One assumes that to be able to teach a computer … No, you’re not talking about teaching. You’re talking about … Yeah, sorry. My naivety on the understanding of machine learning.

Professor Richard Everson:

It’s okay. Learn.

Adam Smith:

Well, for the community to learn and look for those patterns that might result, does it need at least to start with? Have a cohort that says these people went on to develop dementia. Is that in the cohort data? What is the high value areas in the cohorts? Are you able in the minute to say, “Look, these are the people to keep an eye on?”

Dr Sarah Bauermeister:

Yes. I mean, ideally you would want a cohort that does have an end point of a dementia diagnosis, but there is no ideal cohort.

Dr Sarah Bauermeister:

Some of our best cohorts have a very, very rich lead in data. They have multiple waves of data. They have perhaps 12 collection points with rich biomedical data, rich clinical history, but yet their end point, their dementia diagnosis, is perhaps a little bit sketchy. You’re not quite sure how they diagnosed dementia. It hasn’t got a clear classification. Then you have perhaps a different cohort that has a very clear clinical diagnosis of dementia, a very clear scale, but the lead in rich data is not so good.

Dr Sarah Bauermeister:

Sometimes this is not a very clear indication by just looking at the cohort and saying, “Oh, well this is an excellent cohort to look at for longitudinal work.” Sometimes it takes looking across multiple cohorts to work with these types of data sets.

Adam Smith:

Absolutely. You can see why it’s important to bring those together, but is their work aside from this going on to create that perfect longitudinal study even though it won’t really feed us what we want for another 40 years? Is that going on somewhere? You don’t know.

Professor Richard Everson:

I think there are proposals to develop that sort of thing, but what you’ve just said is 40 years.

Adam Smith:

We can’t wait that long, can we?

Professor Richard Everson:

We can’t. It’s just too long. We’ve got to try and do the best with what we’ve got, as Sarah says, bringing these datasets together.

Adam Smith:

Computers are smart enough to spot these patterns, and trends, and the things you are looking about and interpret the different cohorts?

Dr Hadi Modarres:

Yes, using machine learning techniques. Using deep neural networks. All these different types of modelling techniques. I think we are getting closer to the stage where we’re able to accumulate these datasets that we have available and to use them to more accurately and objectively classify patients and be able to detect the signs of cognitive impairment sooner.

Adam Smith:

I find it really hard to get my head around how that is even possible. How can you start to spot that from this, this cohort data? Can anybody answer that question? Put it into simple terms for me.

Adam Smith:

Maybe a lot of our podcast listeners won’t all work in this field. I think if there’s a way of trying to explain that, that would be useful. You have some big spreadsheets? No, not really. You’ve got these data sets with this information in there. How do you tell a computer to start to look what to look for?

Dr Hadi Modarres:

Yeah. The way I look at it is you can think of it as a doctor that is seeing lots and lots of examples of cases and it’s learning from them. The doctor’s kind of limited in the sense that they may only be able to see a certain number of incidents during their professional life, but you can train a machine learning model with thousands of tens of thousands of examples and every time it sees an example, it’s able to infer, and learn from that, and improve its way of classifying.

Dr Hadi Modarres:

The bigger the data set that we have of clean good data that we trust, then the better the algorithm can become and the more it can learn. Just as if you had a doctor that was seeing tens of thousands of examples and learning from each one.

Adam Smith:

That’s good, and there are very practical applications for that as well. Haven’t we? We’ve seen it in the work at Moorefield were they’re using it to look at eye disease and training communities to do that. Well, I’ve just given one there I guess, but other examples of where this technique and method of using this longitudinal data has been used before in other diseases in this same way or is this ground-breaking? Is this unique?

Professor Richard Everson:

Well, I think there are examples in, as you say, the famous Moorefield’s eye example. People have been very successful in image processing for skin lesions and detecting cancer like that. Although not really longitudinal data in that sense. I’m not aware of big longitudinal studies, but I’m a machine learner, so Sarah probably knows more about that.

Dr Sarah Bauermeister:

Yeah. I think it’s new. I think that is the bottom line. I think all three of us are sitting here thinking. We’re actually thinking hard to give you good examples and definitely in our field that we’re looking at the dementia research, this is new. I think this is ground-breaking what we’re trying to do, and I think it’s really exciting.

Dr Sarah Bauermeister:

I think also what we’re doing is we’re not trying to say, well machine learning, this is it. This is our Eureka moment. We’re now going to use machine learning. We have found the secret. We’re going to use machine learning in conjunction with existing methodologies. Multilevel modelling or structural equation modelling, which is actually the field I use to analyse these large cohort datasets, we’re going to use machine learning with these types of methodologies.

Dr Sarah Bauermeister:

The features that we select with machine learning we’re going to use with these methodologies. We’re not suddenly saying, “Well everyone, put down your tools.” We have found the moment. We’re going to use these methodologies together so we’re strengthening it. So it’s a multi methodology approach

Adam Smith:

I suppose it’s what you were hoping then that will emerge is what are back to those risk factors that we’re also common and familiar with in talking about these risk factors. It’s to be able to start to zero in on rather than some general risk factors, which we all know also cause cancer, and heart disease, and these other things is to zero in on the very specific risk factors. So we know whether this is a lifestyle change that people would need to make, or if it is something genetic, or sort of something different, the ultimate cause. Using that data.

Dr Sarah Bauermeister:

Yes, much earlier in the disease because by the time someone walks into the clinic and says, like Richard was saying, “Oh, well I feel that I have a memory impairment or I feel that there’s something going on here.” Sadly they do get diagnosed with dementia or Alzheimer’s. The disease has actually been present for 15 to 20 years before. If we can use these predictor variables and detect the disease so much earlier and when treatments are developed and preventions are in place, we can come at it at a much earlier stage of the disease.

Adam Smith:

Sarah, and I know Hadi, you touched on this earlier. What do you think collaboration across disciplines brings to this?

Dr Sarah Bauermeister:

I think by collaborating across disciplines what it brings is new ideas. Fresh ideas. I think sometimes we, we tend to look at things with tired eyes. You tend to think just within your own discipline a little bit blinkered. Suddenly someone else comes along with a mathematical head or a biology head on and they go, “But what about this?” Suddenly you go, “Oh, but I didn’t think about that.”

Dr Sarah Bauermeister:

I think if we’re all open to becoming a little bit more multidisciplinary, as long as we’re open to that, then suddenly we just start to look outside of the box. I think that’s what this is about. That’s what downstairs is about. You’ve got people from all these different backgrounds and you can hear them around the table as Richard was saying. They’re all saying, “Oh, I didn’t know that happened,” and it’s fantastic to hear.

Adam Smith:

When they break up into the groups, is there a machine learning person in every group who knows about?

Professor Richard Everson:

There is, but we didn’t organise it like that. It’s very interesting the way that these groups have coalesced I suppose into groups which have got interests and a range of expertise. I think there’s a tiny bit of rivalry between some of the groups, but there’s also …

Adam Smith:

Is there a surprise at the end of the day? Is there?

Professor Richard Everson:

Well, no, only the moral prize of contributing to [crosstalk 00:23:46] dementia.

Adam Smith:

Kudos.

Professor Richard Everson:

The other thing is that we don’t put them in little closed rooms in their groups. They walk over to the other groups. They talk to each other as well. They are in groups working on particular problems, but there’s a lot of collaboration across the whole data zone, which really is exciting and refreshing to see.

Dr Sarah Bauermeister:

Right on. Yes.

Adam Smith:

Yeah, absolutely. Are there, are there any risks to, to applying machine learning in this way? Is it all risk-free? Is it perfect?

Professor Richard Everson:

Well, one thing I do want to say is machine learning and AI, they’re not this magic dust that we can sprinkle over the problem and get the answer. You know, as Sarah says, there were lots and lots of other techniques there.

Professor Richard Everson:

The machine learners like me, we know damn all about dementia and so something like this is really useful because there are people with a whole range of skills and knowledge. I think, yeah, there are risks of blind application.

Adam Smith:

To Sarah as well. Sarah, obviously you are very focused on dementia. Do you have to keep these people focused? You know [inaudible 00:25:05] is somebody who’s very interested in other things. Suddenly you spot in there “Actually there’s something else really interesting here. Let’s go off down this rabbit hole of something completely unrelated because it’s interesting.”

Dr Sarah Bauermeister:

Indeed you do. You have to work together when you’re working with machine learning. You’ve got to have the theorists sitting there alongside the people who are excellent at machine learning.

Dr Sarah Bauermeister:

When they say, “Well look, these features have really popped out here and they look really significant.” Then the theorist comes along and says, “Well actually, are you sure that that isn’t just noise?”

Dr Sarah Bauermeister:

You do have to work together. In a way this is a methodology that naturally pulls different disciplines together. Otherwise you will end up with someone running along saying, “Oh, look what I’ve got. Look what I’ve got,” and actually it’s not making sense.

Adam Smith:

Yeah. Sorry, Hadi. Did you have something to add?

Dr Hadi Modarres:

Yeah, I was just going to say that even from a regulatory perspective now with the new GDPR regulations, we have to be able to explain why the model is coming up with the predictions that it is. If somebody is denied a loan, for example, based on an algorithm, then we should be able to interrogate what was the reasoning or how did the algorithm come to that decision? With some of these models it can be quite difficult to interrogate them and to understand where they’re coming from.

Dr Hadi Modarres:

I think machine learning researchers are working on ways of essentially being able to extract from the models which features are being used and how it’s coming to its decision. That’s something I think that we’ll need to have.

Adam Smith:

That interested me. That raises a question in my head because of course you’re using this data I assume in an anonymized way. This is just data.

Adam Smith:

Having discovered something, how do you make use of that? How do you test your hypothesis? Can you go back to people? Do you reconsent them? Do you know whose data it is to go back and check things that you come up with? Do you just design another study then to test what you think you’ve discovered?

Adam Smith:

I’m looking around the table for anybody who can answer that question.

Professor Richard Everson:

I think one of the ambitions that we have is that there are present large scale studies. The data that we’re looking at actually comes from the United States. Memory clinics in the United States.

Adam Smith:

Oh, okay.

Professor Richard Everson:

One of the ambitions that we have is that we’ll be able to apply the sorts of models that we’ve got to UK data and probably also some other United States data. That’s where I think they’ll find validation in that sense. Then, of course, the machine learning technique itself is that you divide your data up and you hold a bit of it out as test data. You just mustn’t touch that until you’ve developed the entire model.

Adam Smith:

Right.

Professor Richard Everson:

At that point you can try.

Adam Smith:

That sounds robust and then of course you can bring in other cohorts as well and apply that and test in that way.

Professor Richard Everson:

The really acid test, as you say, is can you transfer it to another cohort? That’s where lots of these techniques fall down. They work really well within the cohort that they were developed on, or one which is exactly statistically similar, but then as you collect the data in a slightly different way, or things were a little bit different, you measure slightly different variables or something like that, then whether you can transfer the method to that new cohort.

Adam Smith:

Yeah, it’s tricky isn’t it? I guess that was a question in here. What are the limitations of longitudinal data? Is then of course applying this, because by the nature of the cohorts you’re collecting, they’re all collecting different data in different ways.

Professor Richard Everson:

Absolutely and that is a hot-ish topic in machine learning at the moment. That it’s not something that just occurs in dementia. Everybody’s doing it. They’re working on one dataset, but of course what we want to be able to do is generalize from the data we’ve got to another practical data set.

Adam Smith:

Sorry Sarah, did you have something to say?

Dr Sarah Bauermeister:

Yes. As Richard said, each data set, each longitudinal data set, is so vastly different from another. If you’ve got one long attitudinal data set that’s collected over eight waves and one over six waves, each of those datasets, the distance between each longitudinal collection point is going to be different as well.

Dr Sarah Bauermeister:

One has collected it every 12 months. One has collected it every 18 months. Not only that, that you’ve got to take into consideration attrition. People are dropping out of those waves. Also some cohorts refresh their data. So you’ve got people collecting additional participants at different longitudinal points. Longitudinal data is very complex.

Adam Smith:

I suppose I feel like we’ve slightly taken a gloomy turn there, but to be really positive, I think what’s fantastic about this is that this is the start of a journey, isn’t it?

Dr Sarah Bauermeister:

Oh, yeah.

Adam Smith:

These data points, as we’ve alluded to, they’re new, they’re unique, they’re ground-breaking. We’re at the start of this process and I think what comes out of this and the work over the next few years, it should emerge to deliver and achieve something.

Adam Smith:

I think if any country’s best place to do that, it’s definitely the UK where we have this fantastic national health service and a single unique provider where all this data is collected.

Adam Smith:

Obviously getting access and making use of that can be challenging, but I think that in the UK it must be very well placed in the world to deliver on this.

Dr Sarah Bauermeister:

Indeed.

Adam Smith:

Just moving on because we’re about half an hour now, what are the next steps of this? Is this likely to lead to some grant applications? Are you still on the lookout for more cohorts to add to this? Can people bring their own data to this, to the DPUK for sharing?

Dr Sarah Bauermeister:

Yes indeed. So DPUK is ever-growing. We’re not closed. We’re not saying, “Well, we’ve got 47 cohorts, that’s it. We’re full. Also people are able to upload their own data. If you submit a proposal to use the data that’s on the platform, you can’t download the data. You put in a proposal and you can access the data on the platform. You can also upload your own data and analyse your own data set alongside the data sets of DPUK and then when you’ve finished your project you can remove your own data.

Adam Smith:

Can people choose to leave it in there as well to add to the greater body of knowledge?

Dr Sarah Bauermeister:

They can. Absolutely. If you have collected a cohort of your own, you’ve collected some clinical data and you want to keep it in a secure environment, you can upload it onto the DPUK data portal. Absolutely. Yes.

Adam Smith:

That’s really good cause that is a question about, “Is there any other opportunities out there for other early career researchers that are listening to podcast to get involved in future?” How might they get involved? Can, can they host data funds locally? Can they come apply to come to the next Exeter one or are you full now?

Dr Sarah Bauermeister:

Yes. Oh, we’re going to have another four datathons actually this year. The next one will be in September in Norwich and then there’ll be another three datathons this year. Actually later in the year. Another one in October, November, and December. Those you’ll be able to register your interest on the DPUK website. We actually have a tab and those will be held this year. So very much.

Adam Smith:

Fantastic. Really encourage all those early career researchers. I mean about a third of our listeners are overseas, so of course you’re welcome to come from overseas. Excuse to visit the UK in autumn to participate in one of these.

Professor Richard Everson:

We have participants in Exeter from overseas.

Adam Smith:

Oh, fantastic. Where have they come from?

Professor Richard Everson:

Iceland. Canada.

Dr Sarah Bauermeister:

Canada.

Professor Richard Everson:

Canada. Certainly.

Adam Smith:

Yeah. This is again a fantastic opportunity I think to come and participate in one of these future datathons. We’d encourage I think everybody to do it. I think we’re just about out of time now.

Adam Smith:

I’d like to thank our panellists, Richard, Sarah and Hadi. I will let you get back to the session. Just before you do, I know you’re only about halfway through the workshop and you’ve talked before about some nice charts and things coming. Is there anything emerging already that’s excited you in the last day and a half?

Adam Smith:

Nodding doesn’t work. Come on Hadi. This is a podcast.

Dr Hadi Modarres:

Our team is working on a temporal analysis. By that we mean we’re looking at the time it takes patients to go from healthy to mild cognitive impairment to dementia.

Dr Hadi Modarres:

We’re looking to see can we predict that time from their baseline test? Is there a way that we can know how long it would take this patient to progress through the disease? Can we segment people into those who deteriorate rapidly and those who have a slower decline? That’s been the main focus of our analysis.

Adam Smith:

That sounds interesting.

Adam Smith:

Sarah? Richard? Anything emerged for you yet?

Professor Richard Everson:

I’m going to punt on the technical aspects of it, but I think the really exciting thing that’s coming out of this for me is the collaborations that I can just see going on in the room. I think regardless of the technical bits that come out of this particular datathon, I think that is the most exciting and important part of it.

Adam Smith:

It sounds like it’s great that they’ll continue to be nurtured and, and come together after today.

Dr Sarah Bauermeister:

Yes. Agreed. I think many have already said, “How can we carry on working together? I think that is a success.

Adam Smith:

That’s wonderful. Just for, for the listeners that are really enthused and couldn’t be here today, are there any roundups shared from will there be something out in the next week or two sharing what came out of this? These workshops these days?

Dr Sarah Bauermeister:

Yes. We’ll be writing something up and we’ll definitely put a summary up on the DPUK website.

Adam Smith:

Fantastic. I should say we’re recording this podcast on Thursday, the 2nd of May. It’s going to be released on the 3rd of May. By the time you’re listening to this, this datathon will still be going on, but just coming to an end. Watch I think the DPUK website and Twitter next week to find out more about the outcomes.

Adam Smith:

This is usually where I say, and of course we’d encourage all of our listeners to get in touch with you if they’ve got any questions or anything that they want to discuss. careers or questions about how they might become involved and bring their own expertise. Can you be contacted by via Twitter? I know Sarah you can. You’re on Twitter. What’s your Twitter name?

Dr Sarah Bauermeister:

S underscore Bauermeister.

Adam Smith:

Hadi?

Dr Hadi Modarres:

I’m not on Twitter unfortunately, but I’m happy to be contacted on email. Hadi@cognitivity.com.

Adam Smith:

Great. Richard, can they contact you through the DPUK website or [inaudible 00:36:45].

Professor Richard Everson:

Through the extra website? Any search with your favourite search engine will find me instantly.

Adam Smith:

Fantastic. Normally you’re professional. Thank you very much. Okay, well thank you very much again for our listeners for taking time to download our podcast. Remember you can visit our website and look at profiles on all of our panellists and there will be details on how to contact them there. We’d also encourage you to post questions and comments in the comment section in iTunes, Spotify, and SoundCloud, and our website that go with this podcast.

Adam Smith:

Thank you very much again. We’d encourage everybody to share this using the hashtag ECR dementia. Again, finally, please remember to subscribe through SoundCloud, iTunes, Spotify. Share, post, and review and watch our website and Twitter feeds over the next few weeks to see how we’ll be celebrating 10,000 plays.

Adam Smith:

Thank you very much.

Voice Over:

This was a podcast brought to you by dementia researcher. Everything you need in one place. Register today DementiaResearcher.nihr.ac.uk

END