Data provide the raw material for analysis and decision-making, but it is only when contextualised that data or insights become truly actionable and valuable. Contextualisation refers to the process of understanding the circumstances and conditions in which the data were collected, and the broader social, economic, and cultural factors that may have influenced interpretation. Put simply, without contextualisation, data can be misleading and even harmful. Join our conversation with health economist and adjunct professor Dan O’Halloran as we discuss data storytelling and explore whether well-meaning policies are incentivising true value or value-for-money in Australian hospitals, day-cares, and aged-care homes and whether as decision makers we have the right to ask people to do things that they simply just don’t have the financial capacity or freedom to do.
Rhetta Chappell (host): Hi, and welcome to Show Me the Data, a podcast where we discuss evidence-based decision making and the ways in which our lives interact with and create data. I’m Rhetta, your host for today, and I’m a Data Scientist at Griffith University. Show Me the Data acknowledges the Jagera peoples who are the traditional custodians of the land on which we are recording today. And we pay respect to the elder’s past, present and emerging.
This is part two of our conversation with Dan. So, if you haven’t already listened to part one, we do encourage you to go and check it out exactly where you found this episode, and then come back to this section afterwards. This is what happens when we just have such fascinating guests, and we hope you enjoy.
Is with these with these diagnoses? And I guess, kind of modelling and simulating different things, do you think digital twins and healthcare are they cool or overrated? Do you think that so we at Griffith here we have a digital twin in one of our buildings, and I went to the presentation there, and everyone’s all gung-ho about digital twins and then they mentioned them in health care. And I had a bit of a Google and there’s obviously lots of different applications. But then I also kind of was left I was like, is this something that we’re actually doing or is it just cool to do?
Dan O’Halloran: Everyone likes the shiny thing.
Dan: And so, everything has a place. Right? Which goes to my point around, it depends. And so digital twin, are we talking about a digital twin of infrastructure? Right? Or are we talking about a digital twin of my genome?
Rhetta: Probably either or and or both, take it either way. Yeah.
Dan: Okay, so let’s talk about infrastructure, let’s start with. And so, you could do a digital twin of the infrastructure and look at workflows and alike. But the infrastructure isn’t the decision maker, the decision maker is the clinicians within that. And so, having an understanding of the digital twin of infrastructure flows, is it going to actually tell you the full journey of that patient? Probably not. Because the thing is, you’re only going to be looking at the flow of the person during that episode of care within that facility. And again, that situation is not centred around that individual citizen or patient who’s having to go to multiple providers. So in that case, I don’t think a digital twin actually adds value. But digital twin could be used to looking at efficiency, productivity, challenges, from an operational perspective of workflow. So the question then goes to all what is it? What is the problem that it’s trying to resolve? And what is the biggest problem? Now the genome, so that’s a more difficult situation, because we’re actually still learning a lot about the relationship between our genetic makeup, and how we then respond to particular medications. So, a digital twin of our genome, I’ve had a similar conversation with some leading clinicians in the country and the question then is, does the patient actually want to know?
Dr. Tom Verhelst: Mmh, what you’re at risk at.
Rhetta: Ignorance bliss, in some cases.
Tom: It’s a good point, if it’s like this 60% chance you’ll develop a cancer, we could cut off an organ, would you?
Dan: Yeah. But it goes to another point, though, is I was taught this in my undergrad, I did pharmacy down in Tasmania, we were told at the same time as a med student, never do a pathology result. And for the purpose of diagnosis, you do it to confirm a hypothesis or disprove a hypothesis. So, when you’re working a patient arm, you should be using diagnostic tools to support your intelligence of your assessment, not to be definitive that I just won’t assess you, I’d just do the pathology screen thanks. So, digital twin, there could be risks that we end up getting to a situation where it falls to the lowest common denominator, where you’re just pigeonholed. And it’s like, this is what’s going to happen. This is what’s actually happening to you with that innovation. That’s actually looking at you in context of everything else. Because digital twin can only capture so much. So is it a good thing? I don’t know. I’m still unsure.
Rhetta: That’s a really interesting part and especially about the one like if you find out something about yourself and does that then become like, destiny is the wrong word, but he kind of feel like there’s some sort of predetermined like, what’s the point? or should we do more? And I never thought of it from that perspective.
Tom: To that end there’s all these people who sort of talk about the augmented self. And sort of, you know, they start collecting data about themselves when they run and they walk, what they eat what they, you know. Some people, they get pretty extreme, you know, the measure, they’re stool quantity, stuff like that, pH, you could go as far as you want. I mean, there’s a toilet where you can measure eventually there’s a spectrum here that measures these things automatically.
Dan: Yeah, now we know why you took so long.
Tom: Exactly. I actually have a good question. Like, you’ll have, there’s two points to this question, if there’s a whole bunch of people who are sort of data literate, or data fan, who come in with this vast corpus of information about themselves outside of the clinical context. Does that help the clinician? Yes or no? And then if it were to really help, is that not creating, if it would be significant, in its helping outcomes? Would that not create massive inequality? Because some people will just not be collecting all that information? Because it’s opt in, right? If I mean, wearing one of those wearables is an optional thing, it’s not mandated?
Dan: Well, it’s sort of goes to choice, and opportunity. And not everyone can afford to wear a smart watch, or have these devices. And so by virtue of that is that if you only have the data, because your ability to buy a device and access it, then by that virtue alone, as that you’ve created some level of inequity. Now, does that mean it’s a bad thing that those that have that information? Maybe, maybe not, because you could run an argument that if those people are more proactive and actually managing their risks and identifying them earlier, then they may actually prevent the need to actually need urgent emergency care at a later point in time. Which then has an allocation efficiency to those that can’t afford the device, to then access the care. Because if the person who could afford the device didn’t know, then both of them would have been trying to access the care. And so there is a potential externality benefit to those that can’t afford it, by having those people that can afford it to have that information. Because you then start allocating more resources to areas of need quickly, more effective.
Having that information also potentially allows us to learn stuff that we didn’t know before. And so with more data, the more and you’re able to actually see things and patterns that you weren’t actually aware of. One of the challenges with sort of the genetic components is that it’s not just the impact that that result has on you, as the impact that it has on your children, the next generation. And so it then goes to well, if the analysis has been done on that, is that other people actually fully aware of that? Or is it actually a social good, that they’re aware of that? Don’t know? Right? And so there’s still a lot of questions to sort of be worked through. And then goes to well, how much information do you need to know? And why? And it goes towards what’s the purpose? And I think where this conversation started is, how do you then identify the insights? Where do you start? Where do you really start to find the needle in the haystack? The first question is, well, what is the problem that you’re trying to resolve? And so if you spend more time actually trying to understand the right question, then you can actually be more targeted in the analysis that you do to answer the right question. And so, as an example, given that we’ve all lived through the pandemic, often you would have people say, I want to know, I need to know how many COVID-19 cases there were. I need to know how many there were by local community. That knowledge alone without context, is actually meaningless. Because if you have those numbers lined up across local government areas, and then you had one that was really high, and one that was really low, in case numbers, if you then contextualise that information with the population statistics across those local government areas, and then contextualise it with the number of people of that population that individually came forward to get a test, you might actually introduce a different story where the risk is, that the risk may have been in the low case number community, because it actually had a really large population that wasn’t actually coming forward for testing. And so you didn’t actually know. It wasn’t in the numbers and so you then have an unconscious bias. And so that’s what I mean is around actually framing what is the question that you’re actually wanting to answer? Because just because you have the data, without contextualization, it’s meaningless.
Rhetta: I think that’s something that we spoke to Kevin from DSpark yesterday for the podcast and one of the things he talked about the value of their data is that often people come with one question, have a look at the data, and then realise they’ve been asking the wrong questions all along. And how the data really helps them then kind of move past that and get into the more interesting kind of juicy parts of what you can do with data and data analysis.
Tom: Or they buy the data to basically provide evidence to their preconceived hypotheses. That’s not a hypothesis, but it’s already more of belief. And then the data says the opposite.
Dan: That’s one of the biggest challenges, right? That’s, that’s the biggest challenge, in my view for data driven policy is that people have preconceived ideas, and misconceptions. They don’t think they have misconception, they…
Tom: Yeah, nobody thinks they have a misconception. In their heads it not the misconception, it’s the other people that had the misconception.
Dan: And so, but the fact is, and I think this is fair to say, is there a very, very few people that have worked in multiple healthcare systems across Australia, and very few people that have worked at all levels of those healthcare systems. And so until you actually have worked in multiple systems, and worked across different levels of the systems, the ability to actually, until you’ve actually had to do those jobs, then you don’t really know how they interplay. And so, if you’ve only coming from it from one position, from one, say, department and you’re then running an entire system, then your experience isn’t necessarily the same for the lot. And so that then means people will have predispositions on the data. And they will then go through Kubler Ross’s five stages of grief. Because the amount of time, I’d be retired now, if I got paid $1 for every time I was told the data was wrong.
Tom: Yeah, it’s funny when people say that their data is wrong. It’s kind of like that show that did on that nuclear power plant in Ukraine, I can’t think of the name…
Tom: Chernobyl and they took out the scanner, and they said, the number is wrong. Like, no, no, it’s probably not.
Dan: Right. And so, one of our biggest challenges is unconscious biases, or ingrained conceptions that, and on that point is that there are people that have the view, that if there was only one operator of the service, then all this would be fine. That there are significant examples across our country and across our federation, we have one provider, and they themselves don’t actually know what’s going on within their own system completely. And so, it’s not synonymous being a government structure to actually knowing, you actually need to have the data flowing to actually know.
Rhetta: We come across that with data custodians, and things like that quite a bit. People don’t know. Yeah. So Dan, if you could have access to any dataset in the world, and maybe we touched upon this a little bit earlier, but we’re kind of using this as a thought experiment, where parking our morality, practicality, ethics, costs kind of things to the side. And if you could have access to any data in the world to have a play around with and gain insights from, what would it be and why?
Dan: I actually think it’d be a lot more investment and access around labour market dynamics, and basically payroll data. And why do I say that? We have all these people come out and tell us that we have to do 30 minutes of exercise a day. They tell us that we should eat healthy, they tell us that we should go listen to music and do the like. But does everyone within our society have the choice to do that? And so, and that choice at the current date, so this recording has been done in February 23, and so we have record amounts of inflation in this country. And we have significant number of people in this country who took the advice of the RBA Governor who said interest rates would not rise until 2024. They went out and they got mortgages, they’ve got significant cost of living increases and not only have they had significant cost of living increases, their mortgages are increasing. That is not going to just have an impact on those individuals, that’s going to have an impact on their children, and will have an impact on next generation, and we will see that for years. And so, telling, just telling people what they need to do, if they don’t have the choice, is meaningless. And I’ve seen this a lot, particularly within the health department, is that actually, a lot of the time is that there’s not an appreciation that the financial freedom or financial constraints that people have, actually are leading to the situation that the person actually is in. And Angus Deaton has published on this around deaths of despair, and Martin Wolf. So Martin Wolf, from the UK, actually presented at the economics conference in Hobart last year, he says for the first time the UK is starting to see exactly the same thing. And that was before they had their 44 Day Prime Minister and had their currency basically devalue and have 15% inflation. And so, if we put that on in context is that we can go and we can get data about children and their experience, but the reality is, the outcomes that you’re going to see in those data sets will be driven primarily by the financial capacity of the parents, and the choice that those parents are in. And so, my view is, don’t waste our resources in actually identifying what’s been happening, identify where the cause is so we can actually make a difference to those children’s lives today, so we can actually prevent it. So those children don’t become a statistic in those other data sets.
Rhetta: That’s a really excellent answer and very different to anyone else. So points for originality.
Dan: Cheers, thank you, that’s good.
Rhetta: Very good. Awesome. Thank you so much, Dan, that was really interesting. I hope you enjoyed it.
Tom: It was fun. You made us think actually. Yeah.
Rhetta: To listen to more episodes of show me the data, head to your favourite podcast provider, or visit our website, RIDL.com.au, and look for the podcast. We hope that by sharing these conversations about data and evidence-based decision making, we can help to inform a more inclusive, ethical and forward-thinking future. Making data matter is what we’re all about. And we’d love to hear why data matters to you. To get in touch. You can tweet us @G_RIDL. Send us an email or if you prefer, just send us a letter by carrier pigeon. Thank you for listening, and that’s it till next time, take care and stay safe.