Home
Trending
New Release
Speaker A: This program is brought to you by Resonance 104.4 FM.
Speaker B: If you like what you hear and want to support our work, please make a donation at fundraiser.resonance.fm.
Speaker C: Hello, I’m Peter Warren, and in this month’s podcast, we’ll be talking about artificial intelligence in healthcare. I’ve been interviewing some of the top thinkers and policymakers involved in the technology, and I’ll be welcoming many more to our conference in London on the 3rd of October. More about that later. A lot of this new AI healthcare kit aims to prevent illness by checking your blood pressure and counting your calories, diagnosing illness early when it’s still easy to treat. The founder of the Docu-It app, Florian Bundrup, predicts that doctors will welcome such innovations and elderly patients will soon learn how to use them. Even Password’s Jane Wyatt finds it quite user-friendly, helping her to diagnose a friend who’s feeling unwell.
Speaker A: Okay, so we have fever, whatever, running nose, sore throat, and it actually will now ask you, have you additional symptoms? Okay. Is it sneeze, a red throat, pain while swallowing, impaired smell, itching of the nose or throat, or none of the above?
Speaker D: Ah, well, I haven’t looked into her throat, but she has been sneezing.
Speaker E: Sneeze, yes, for sure, and pain while swallowing, yes.
Speaker A: Okay, I will select those. But you already see that it will actually really go deep, and well, it obviously knows if you have certain symptoms, certain other symptoms might appear or might not appear. It will actually ask about both to include or disclude certain illnesses.
Speaker E: That was a very good demonstration of how how quick it is.
Speaker A: Yeah, actually.
Speaker D: And I guess if you have all the symptoms of a particular syndrome or illness, then it will tell you so right away.
Speaker A: Yeah, it will tell you. It typically needs between 10 and 30 questions, really depending on what kind of symptoms you have and what kind of illnesses it tries to assess and how severe those might be. And this whole engine is certified as a medical product, Class 1. Which is important if you want to use it in the public healthcare system, obviously, and to also provide that trust that is needed for patients and for insurance companies to really use this kind of technology. You can always improve. I mean, Class I is not the very, not the highest. You shouldn’t only rely on it, but it can give you a first estimation of what you might or what you shall do or what options you might have.
Speaker D: I can imagine though that some of the most worrying healthcare situations involve children or babies who can’t tell you how they’re feeling. How have you factored in any kind of safeguards against parents or childminders assessing things wrongly from people who, for whatever reason, because they’re too young, because they’re too old, maybe because they have communication difficulties or language problems?
Speaker A: Those are the most difficult ones because you cannot— you can access even less symptoms than you can for yourself, and even those are corrupted because you don’t notice everything about yourself as an external viewer, a doctor, would do. So you always have compromise here, but it will still ask you a couple of questions, still trying to assess the severeness of those symptoms that you at least mentioned. And as you have seen in the application, in the example, you are allowed to say, well, I don’t know. And for, for kids, obviously, for some situations, you will just won’t know and will then anyway determine whether it is severe or not. But if something is really wrong with your child, I guess not all parents will just stay home if an application will tell them, hey, stay home. So it is still allowed, and it should be allowed And that’s actually something we have to worry about. It still should be allowed to distrust technology and anyway consult with a doctor. And this is something where we have to have a debate about. Once these algorithms get better and maybe are publicly recognized as being valid all the time, if people that trust technology less than the average, are they still allowed to distrust it? Do they have to pay extra? It’s something that we don’t speak about. We speak about, oh, it will cost jobs and everything. I mean, in healthcare it won’t cost jobs because you still need, like, the pill. You still need to be treated, and we don’t have enough doctors anyway. So doctors, guys, you’re fine. We have to actually have a debate about how we use this technology. And how we deal as a society with it. Can we take it for granted? Do we force others to do so? And imagine a robot that operates you. Scary feeling in the first place. But now let’s think 5 years ahead. This robot— we have a study, an actual, actually independent study that validates this robot is right in 98% of the time. And the human operator is just right 93% of the time. People will change their minds about the robot now. They will ask to be operated by a robot, but some people will choose the human. But what about those 5% errors? So those 5% might have disability afterwards because these mistakes will injure them permanently. Who pays for that? These are interesting questions, not only about whether AI is reliable or safe. It will be safe at some point. It will get more reliable. Hopefully, at the time it is reliable and safe, we had a debate about how we use it, because we don’t do it right now.
Speaker C: There are many apps like Docu. Little startups and a few big players. Babylon, for example, and Benevolent AI. That’s a brand name, by the way, not a description for machines of loving grace. Microsoft’s InnerEye and Google’s DeepMind are already in action. Swift and accurate diagnosis is what they promise. And to achieve that, they need data. Our data. There’s an app called Tumor Trace. That could potentially save lives by spotting early signs of cancer and giving a rapid response, saving the patients from months of uncertainty and worry. Amir Butt, a serial entrepreneur in the healthcare sector, is raising the capital to launch Tumor Trace using a digital coin called a Medici. Amir believes it’s important to keep health data open and transparent. He foresees a future where people freely allow their health data to be used for diagnostics and improving medicines.
Speaker F: The people who run countries are just not geared up to understand the implications of artificial intelligence, and that is, you have— AI has got the capability of doing enormous good for mankind, to the point where it will help us become intergalactic species. All the way to something very, very dark and sinister of where you’re manipulating human behavior. I don’t mean à la Cambridge Analytics and manipulating people, you know, to help Trump get on to be president. I mean where you’re manipulating sort of movements and behaviors of people directly, because artificial intelligence spots underlying patterns. If you just make the assumption that human behavior is subject to underlying patterns, you can see what I mean. With the right kind of artificial intelligence, you can manipulate people very, very sort of finely without you even knowing, because all we’re doing is responding to underlying patterns, and we’re not aware of our bit of manipulation that’s occurring. And that, that is very, very sinister. And I think the solution for that is actually the education and understand legislation, understanding what the moral implications are But I think the poison in all of this is anonymity. As soon as you remove anonymity— so there’s been lots of psychological studies have been done about human behavior to do with, you know, sort of criminal activity or bad stuff. And it’s a function of how strongly you believe that you’ll get caught if you do something wrong. If you know that you won’t get caught, then more and more people actually move towards doing things which they wouldn’t do if they knew they were going to get caught. So anonymity in that sense is the dark side of the force, and I’m more and more in favor of there being zero anonymity in all systems, because that’s the only way you’ve got an absolute sort of track back to who the perpetrators of things that might be inappropriate, how to track them.
Speaker C: There’s a big move to make data anonymous within the NHS so that people can actually use that data in a very free way. So, but people want to have that data made anonymous. Where are you leaning? Are you saying— because in Belgium, for example, there is a move to make all data not anonymous, but, and for people to donate data. They’re, they’re being called upon to donate data in the same way that people donate blood. So that Belgium can achieve a preeminent position in AI. Is that what you’re talking about?
Speaker F: So the anonymity of medical data today is in Europe being governed by GDPR. The whole point of GDPR is to protect the individual. But if there was general non-anonymity, then there is no need to protect the individual because everybody’s in the same place. So I’m describing a situation which is many years from now where you have zero anonymity. All data is open and open source, as it were. But to go from here, where people feel very, very precious about anonymity and their own data and so on, to go from here to there is not an easy path because as soon as you make datasets open, people will start to abuse them. And the abusers have the privilege of the anonymity. So it’s not so much, this is my medical data and I need to protect it and I don’t want anybody else to see it. The lack of anonymity will mean that if you are an abuser of a system, then it’s very easy to find out who exactly you are. At the moment, there isn’t. This is one of the reasons why things like the cryptocurrencies have had a very bad name. Because you could do a cryptocurrency transaction where nobody need know who you are. And so money laundering is just trivially easy, which is why we’re having now legislation coming in where banks, for example, any cryptocurrency transactions must be backed by, you know, know your customer, anti-money laundering checks that you have to do.
Speaker C: But in a sense, what you seem to be calling for is some sort of blockchain that says this person is this person, this person changed this data, this person did this to the data. You want, you want an absolute record.
Speaker F: Well, so I mean, blockchain lends itself to that kind of data storage, but it’s not really about blockchain. It is about the principle of the— if there is no anonymity and if all data is open source, then by definition it means that criminal activity or abuse of data is easily traceable. Artificial intelligence can do extreme good for, for humans, for mankind, but also has the power to do extremely bad stuff as well.
Speaker C: In the last few days, the British government published the first draft of a code of conduct for data-driven technologies in health. And Health Secretary Matt Hancock said in a speech that the National Health Service’s IT systems are clunky, clunky, clunky. He moaned about the fact that the last attempt to harvest patients’ records to provide big data for training the AI algorithms ended in a fiasco. You might remember we reported it on Password in 2016. The Care.Data scheme had to be withdrawn by NHS England because of doubts raised by Fiona Coldicott, the National Social Care Guardian. Well, it’s back. Lord Clement-Jones, who chairs the UK House of Lords Artificial Intelligence Committee, has misgivings. His committee’s report published last year cited the Information Commissioner’s criticism of the London Royal Free Hospital for allowing Google DeepMind to use 1.6 million patient records without their informed consent.
Speaker G: It was really quite a frustrated speech in many ways, but he did go into a huge amount of detail about the procurement process, about interoperability and so on, and I thought that was hugely significant coming from the former DCMS Secretary of State. And you know, there are many of us who think that, my God, he’s chosen a pretty high mountain to climb, in these circumstances. And it’s not just artificial intelligence, it’s the use of blockchain, it’s the use of Internet of Things, cloud, and so on within the health service. But the thing that I disagreed with him about, and I, and I found that he wasn’t quite on point about, was the fact that it’s not just about equipment, it’s not just about, if you like, the IT aspect of it, it’s about the datasets and having the data in the right order, right across the board in a comprehensive way, is absolutely vital in the health service. And until they get that right, then they won’t be able to move much further. Because as you and I know, artificial intelligence lives off data. You know, it’s a cliché to talk about, you know, data being the oil and the fuel of the future. But it is true. And, you know, without the right kind of NHS data, organized in the right kind of way, we’re not going to make advances in artificial intelligence. And it, you know, when you have to use external people like DeepMind to identify a fairly small subset, you know, of the data which they can use, and then they use it, that’s fine. But that isn’t going to happen at scale across the health service unless we get the data collection and organization right.
Speaker B: Why do we need to do this? Why do we need to adopt AI, you know, Matt Hancock said clunky, clunky, clunky about the NHS systems. He— why do they need it? Well, what is, what is the imperative politically?
Speaker G: Well, let me draw you a parallel. For instance, Iceland has got the most fantastic data set of genomic information about the DNA of their inhabitants, if you like, 250,000 people. And it is an absolute, you know, it’s a fantastic— it’s a, it’s a It’s a treasury of information which is usable by people who are researchers in the DNA and the genomics field, and that will lead eventually to breakthroughs in all sorts of health areas. Well, the same is absolutely true of data use in the NHS, and you know, you can’t improve care unless you’ve got the right data. Now, the data may be, you know, photographs of particular cancer treatments, cancer conditions, and so on and so forth, It may be anonymized data about particular types of patients, demographics, and so on and so forth. I mean, it’s going to be incredibly varied, that data, but nevertheless, medical researchers and people who are developing new applications and new treatments need to be able to organize that data and basically use that for the algorithms and the AI that they’re developing.
Speaker B: There’s a lot of suspicion about, though, isn’t there? A lot of people are thinking, hang on, what’s happening is that this government is trying to free up this data so that it can create an industry and use that to essentially privatise the NHS. Is that something that concerns you?
Speaker G: Yeah, well, I don’t think it’s going to happen. I mean, quite frankly, we had the DeepMind situation as far as the roll-free was concerned, and I think it’s been established pretty much that You know, we’re not suddenly talking about using NHS datasets to enrich, in quotes, the private sector. The private sector may help develop treatments within the NHS, but the benefit of those treatments and the benefit of that AI that’s developed will come back into the NHS. So, I mean, I think the ground rules— and we know, we saw that the Information Commissioner was brought in, she basically censured DeepMind for what it had done. DeepMind, as a result, has set up a separate independent review board, which has now produced two reports. And, you know, those reports have to some extent been critical about DeepMind’s activities. So, you know, it’s led to actually quite a lot of new developments in terms of ethical governance, but also in the way that the NHS safeguards its own data for the future.
Speaker C: And the technological clunkiness of the NHS was highlighted last year when the WannaCry ransomware let hackers immobilize whole hospitals, mainly because the hospitals were using an obsolete Windows operating system known as XP. Lois Rogers writes about healthcare in the Sunday Times, and she’s skeptical about the AI revolution.
Speaker D: I think whatever way you cut it, you have got to accept that we will have that kind of hack attempt, and that will probably happen more. I think I would almost put money on that happening more, and that will trigger a whole cascade of alarm about how we’re going to cope with attempts to take out chunks of the internet and steal data. I think that if we get our minds around the idea that everything that you put out there can potentially be got at by people that you don’t— it wasn’t intended to be seen by, or you don’t want to see it, then we can proceed in a slightly less paranoid way. I think all the time there is a kind of double-edged sword. It has huge benefits, but it also has huge risks.
Speaker C: Lois Rogers does not believe that an app a day keeps the doctor away, for the simple reason that the people who need most care are the very ones who are least able to manage the technology.
Speaker D: We know that we have an ageing population, an increasingly unhealthy population beset by the side effects of obesity, heart disease, dementia. Everyone’s living longer, so they are a burden, an increasing burden on the health service. For a longer period of their lives than was the case in the past.
Speaker C: So what does that mean then? What does the health service or health services across the developed world have to do to try to deal with that?
Speaker D: Well, it just means that there is ever greater pressure on doctors to see patients, and the way that they’re trying to deal with that is using new forms of technology—
Speaker C: robotics, AI, artificial intelligence of one sort and another—
Speaker D: to monitor patients. It might be devices that will talk to people in their homes or do blood pressure checks and other basic bodily function tests in their own home. But the problem with that is that most of the people who are chronically sick tend to be elderly, tend to be less technologically minded than the rest of us, and therefore they’re going to be people who have much more difficulty adapting to this type of technology, this type of kit.
Speaker C: Kit that looks like an app and lives in a smartphone might be hard to handle for a person who’s elderly and confused, but an intelligent assistant that looks and feels like a radio, well, that’s different. Research in Japan and the Accompany project developed by Birmingham University have found that most lonely older people appreciate having a Siri or Alexa-type companion that they can talk to. Peter Derleck represents Nuance, the creator of Dragon natural language and Florence voice-activated healthcare software.
Speaker H: Voice is the interface because that’s how people communicate, right? I mean, we’re not inventing someone new, something new. We’re trying to make systems help communicate the way people have always, well, at least in recent history, always communicate it. So yes, we believe that the voice front end, the speech recognition, the virtual assistant is the front end access point to this wealth of AI capabilities because that’s how humans communicate. It’s not the only thing and it requires other technical advancements, but if you’ve got to type everything in or punch on a keyboard on a small mobile phone, it doesn’t work as well. It’s not as effective. There’s parts of the population that won’t do it. We’ve already seen the data. I mean, look at the pickup of even simple things like smart speakers in homes. So there’s a lot of proof points today that if you can put a good conversational user interface as the front end for someone, whether that’s a physician using Dragon Medical to talk to the Epic electronic health record, or that’s someone ordering something through an Amazon Echo in the home, there’s a lot of data points. I would say if you can create that front-end interface, the adoption goes way up and then you drive the outcomes that you care about.
Speaker C: But that also throws up another point, and a very interesting one, and one that many people are saying that is going to take off now over the next 10 years, which is that there’s a growing recognition among people that they’re providing a lot of information and perhaps that they may not be getting the best return for the amount of information that they’re giving. Within the last few days, for example, it’s been revealed that Google tracks you even if you say that you don’t want the technology to track you because there’s some value in that for them. Do you think that people are going to start to say, hang on a minute, there’s some value in you knowing where I’m walking around in this building? There’s some value in you knowing what my symptoms are, you know, to do with my rare condition that that you’re then overlaying on all of these other people. Do you think that people will start to say, I want to get a bit more back out of all of this?
Speaker H: Oh yeah, for sure. I think you have to distinguish between a consumer who’s using their smartphone and having Google behind the scenes— not to pick on Google— but behind the scenes monitoring everything for them to make money from, especially if you’ve said No to. That’s like a big no-no. That’s very different than, for example, I’ve got an 85-year-old mom at home and I’m trying to make sure she doesn’t feel lonely and if something bad happens we can take care of her and therefore I need a real-time data feed to her blood pressure and how she’s moving around the house. Because as you said, the value deal is completely different there. I’m opting in, or my mom’s opting in and I’m getting something out of it. So I think it really does come to what’s the deal that you’ve signed up for. These data security or privacy things blow up generally because there’s no agreed-to deal for what you’re giving and getting, or you think you’ve agreed to something but what’s actually happening is something different. That’s where you’ve got all this problem, whether it’s the Facebook stuff or Google stuff or whatever. You’ve got to have an explicit understanding of what the deal is that you’re giving up to be monitored. And if you don’t understand that or haven’t agreed to it, that’s a huge problem. But if you understand that, you’ve agreed to it, and there’s value, then everybody can make their own decision.
Speaker C: I mean, that’s going to be an interesting point, isn’t it, particularly with the elderly, because lots and lots of elderly people can get quite grumpy. They can resent their children taking over control of of their lives, and they may say, I don’t want my child to actually be given any information about me. They’re meddling. I don’t, I don’t want this. So that is something that people do have to be quite clear on.
Speaker H: Yeah, for sure. And again, in, in that case too, I mean, there’s nothing that stops an elderly person from saying that I want the people monitoring to be my clinical team, not my family, if that’s the case. So I think at the end of the day that, you know, you’re getting into this, the personal dynamics of that family and how they want to do it. But technically, I think the issue is that you have technology, AI technology today, both on the speech side and the textual side, that can really make a difference in the quality of care or the quality of life that someone has when applied appropriately. That’s really, I think, our our message. And you have the different technologies are at different stages of maturity depending on what technology you’re talking about and what the application of the technology are. Behind the scenes, you’ve always got to worry about privacy rights, data rights, who can see what, you know, that’s kind of a universal thing that you got to worry about in all of this stuff.
Speaker C: Let’s just pick up on another issue. You mentioned at the beginning that the relationship really should be that this is technology that augments the medical community, it helps doctors. And you also said that there should be a minimum amount of, to use your plane analogy, landings that people make. Is really the reason for that because of legal issues? Is that because at some point there has to be somebody who has to take overall responsibility in all of this. And if you start to argue about the responsibility of a particular technology company, you’re going to have a huge, huge legal case going on.
Speaker H: Well, yeah, I know it may vary by country too, but I think, I mean, obviously there’s governmental, you know, concerns around this stuff. I think also at the end of the day, we all have responsibilities.
Speaker C: And as Lois Rogers said, There are huge risks, but there are huge potential benefits too. Even perfectly healthy people can benefit from AI-driven apps. For example, if you’re a woman who’s trying hard to conceive a baby, or trying hard not to, then Ida Tinn has a technology. She launched her app called Clue at last year’s Ada Lovelace Festival. And she tried to convince password Charlotte Roundtree that logging your periods with Clue is okay and they can be trusted.
Speaker E: So I would say that we actually are very, very conservative at Clue when it comes to selling data. We have never sold any data until date. I think the only reason why I’m not very kind of black and white when it comes to saying I would never ever sell data is because I think the key question is what is is in the benefit of the users, right? And you can imagine scenarios where it’s helpful to provide, for instance, manufacturers of the pill with information about side effects that users might experience so that they can build better formulas. So, but generally, I would say we care a lot about honoring the trust that people give us by sharing their intimate data with us. Well, I think that’s fantastic. In the UK, there’s, there’s something coming into play called the General Data Data Protection Act, which is stopping certain corporations from either leaving their data underprotected or from selling it off without people knowing. Yeah. Does— do you show the same kind of, I want to say intensity, but care in making sure all the data is protected and very encrypted? Oh yeah, absolutely. That’s like one of our, if not like the top priority. I mean, I think we live and die by the trust of our users. So for us, it’s absolutely key that we do everything we can to keep this data secure and safe. Both from a kind of technology standpoint, but also as we just spoke about, the more kind of ethical and kind of human, like what— it’s a contract that we have with users, right? They have to understand what they give and what they get, and that has to be very transparent. I think that’s a really good way of thinking about it. A lot of companies aren’t doing that, obviously. No, and I think it’s very unfortunate because, you know, I, I wrote about kind of, you know, data can be used for good because we almost forget right, because people are misusing it, and I think it holds so much potential for bettering our lives and our health outcomes that we really have to make sure that we can show good examples. This, this kind of technology is giving us sort of an option to find out more data and more scientific results than we ever could have before. Absolutely. Because obviously not everyone is able to go to a trial, but nearly everyone is able to use an app. Absolutely. And I would I would say even with where we are today, just collecting data is raising people’s body awareness a lot so that now we can kind of apply our own learning and to see if something is off and we know when to go see a doctor because we know our bodies well. Yeah, that’s a huge, I think, empowering thing, particularly for women. I think a lot of female health problems are often either underrepresented, under-researched, There’s something I was speaking about the other day was that female heart attacks are completely different to male heart attacks are. And so that is very under-researched and it’s not shown in like the media. There was this new legislation that came out in America where it says that you have to include women in trials because usually you leave women out because they have these annoying hormonal cycles that change everything. So that’s, you know, that’s a really big step forward. I think that’s fantastic. Well, speaking of data, obviously, you spoke, you said it could be used wrongly. How could you imagine that Clue data could be used wrongly if you weren’t trying so hard to obviously keep it safe and keep it protected? So I think if we, for instance, bombarded people with ads that were very targeted because we know a lot about their health story, like, I would find that inappropriate. I think you can ask people if they would like to have recommendations for services and products that might be truly helpful, right? If you have something going on, you might actually want to purchase whatever service, but that has to be kind of like on consent base, I would say. Have companies shown an interest in buying the sort of the data that you produce from Clue? People are very interested in having the contact to the users. I think that’s the biggest thing, because a lot of people that sell products to women, like the pill manufacturers or people selling tampons or whatever, they They don’t necessarily know so much about how their users use their products. But again, we’re very conservative. We have not made any partnerships with these. I mean, we’re conservative when it comes to protecting our users’ interests. We’re not conservative when it comes to thinking about being a progressive voice in female health. But yeah, people are very interested, but we are— no, we have not been so interested in that.
Speaker C: Intimate data like this could be a goldmine for all sorts of commercial companies. But it could also leave women and girls open to abuse and coercion from controlling partners, disapproving parents, pro-life campaigners, and so on. It’s not hard to imagine how statistics about menstruation patterns could be used in ways that are not ethical. Did you hear what Ida Tinn said there? In clinical trials, they usually leave women out because they have these annoying hormonal cycles that change everything. Gender bias seems to be built in there. And what about people with special needs? Because as Dr. Beatrice de la Inglesia from the University of East Anglia points out, The use of data by the health service is massive and even now involves filtering Twitter messages to find out about problems such as air pollution.
Speaker D: So at the moment, for example, some of the projects that we are dealing with here are to do with social media data that is becoming important in the context of looking at disease and what we call syndromic surveillance. So we work in a project with Public Health England, and they have systems which look at the development of, of like outbreaks of diseases and, and things that might, might require, for example, some response. So they look at things like flu and anything like mumps or asthma, air pollution, and so on. And they do that looking at things like the 111 calls and the attendance to hospitals. And what we’re trying to do with them is looking at whether something like social media data, for example Twitter, which is quite readily available, may be usable in this context. Because sometimes people might not maybe call a doctor if they are feeling unwell, but they might discuss their symptoms or how they feel in Twitter. And if you can detect that signal, then that might help them, and it might be more timely perhaps than the signal they might get from the other systems. So that’s what we’re trying to understand, is whether we get a good signal, whether we get a more timely signal from something like social media.
Speaker C: You’re essentially looking for dialogues that are going on in social media and where somebody’s complaining to a friend that they’ve got a sore throat, or they’re saying that they can’t do something because they’ve got a sore throat or they don’t feel particularly well, and you’re using that as an early indicator because what, that they don’t go to a doctor?
Speaker D: Uh, yes. So the idea is that if, if you, if you can capture that population, it might be different to the population that go to the doctor. So you might be tapping into a different set of people, or they might have gone to the doctor but also discuss it, in which case you might see the same signal but strengthened by the fact that you’ve seen it in different places. So in a way The idea is that these kind of things will help them to, to better understand how things develop. And also, if you think about the third world, the developing countries, there is a lot of use of mobile phones there. Mobile phones are becoming quite widespread, and there might not be access to very good healthcare facilities. So if you can detect, for example, disease using things like social media, that might be very advantageous to, to countries where the access to health services is poor, but maybe the access to mobile telephony and therefore things like social media might be better. So there are advantages in— if you can detect things like that or use it to influence how, you know, how you transmit information about disease and so on, then you might be able to do much better in terms of syndromic surveillance in countries where you don’t have very good healthcare facilities.
Speaker C: Now, there are challenges in this, aren’t there? Because people are very, very sensitive about their health data, and they’re also sensitive to whether they feel that they are being surveilled. Is that an issue?
Speaker D: It is a very big issue, and I think that the ethics of data collection, the ethics of how you use the data, is a very big theme And you do have to be particularly careful because people, as you say, they don’t like being surveyed. They’re worried about how that data is going to be used. And even in the context of Twitter, where you might think something that is put in Twitter is quite public, people still have concerns about how that data is being used. And also, there is an issue of trust. They might not trust very much, for example, a signal that comes out of something like Twitter because they don’t think it’s reliable. So there are issues, there are a lot of issues that we have to deal with as we go on with this kind of like artificial intelligence theme and data and data analysis, which, which are to do with the ethics of the process, making sure that we use the data adequately, that you inform people of how the data is going to be used, etc. So yes, that’s a big challenge.
Speaker F: Exactly.
Speaker C: I mean, informing, we’ve really got to up that, haven’t we? Because essentially people have got to be very, very clear about what is going on with data. I mean, that transparency is going to be an issue because there’s got to be a temptation for people to think, oh, we’ve got this data and we can use it for anything that we want.
Speaker D: Yes, and I think some of the new regulations that have come up are trying to address those kind of concerns, because we have seen the use of data in context that people might not agree with. For example, Facebook used for to influence political decisions, etc. And I think there is a fear that— and that comes with any good technology— there is something that is a technological advantage can normally be used for good purposes but also for bad purposes. And in a way, there is this, this issue of how do we make sure that it only gets used for good purposes. And so I think we have to develop good processes and good ethical kind of like systems that work in the right way and use the data in the right way. There might be very good outcomes from analyzing, for example, health data. So if you think about the data that gets routinely collected in a hospital, that if analyzed Well, may uncover disease interactions that we don’t know about, because currently a lot of the clinical research is to do with clinical trials, which are difficult to run, expensive, and much of the data that we routinely collect— that’s another part of the work that I do— may be used for clinical research and may unlock things that we don’t know about diseases, about interactions between diseases, etc. But the problem is how we do that. Without breaking ethical boundaries. How do we do it? How do we convince people, or how do we make sure that all of this analysis that could be done is going to be in their interest and they might want to give consent and participate? And I think it’s a matter of, you know, convincing people of this.
Speaker C: That was Beatriz de la Inglesia. Just underlining the problems that we’re going to have to deal with when we start using big data. Vinette Taylor at Telefonica told me that she’s working with companies that can make personalized healthcare that corrects biases by collecting whole medical histories.
Speaker I: You know, there’s a lot of historical data that would help you make decisions quicker. If we start from scratch today, We’re probably going to be 5 or 10 years down the line before we’ve got the data that we can then make some smart decisions on. So if there is a way of taking some of that historical data and the new data that we find, we get from GPs, from hospitals, and that we volunteer ourselves in— a lot of us walk around with wearable devices today that’s tracking our health, but what we’re doing with that, we tend to keep it ourselves and not share that data. So if there was a way that we could share and reuse that data, then, you know, it becomes interesting. I’ll give you an example of something. There’s a startup company that I’ve been mentoring and doing some work with recently, and they have an application that allows you to store information on people with special needs. So if there’s a patient that’s got MS or that’s got autism and they’re on special medication, what tends to happen is every time there’s a new carer comes to see them, they’ve got to go through what medication they’re on, what treatment they’ve been having, and you know, sometimes they don’t remember that information or they don’t pronounce the medication correctly. And the same with Alzheimer’s and dementia. If you had an application that stored that information for them but actually learned on, you know, what medications gave them, you know, good results, what didn’t, and you could learn on that data, then that’s a live data pattern that’s been produced. And then the algorithms from that can then help you make smarter decisions And I think that we’re going to see more of those applications. And I know that, you know, there was an article by Matt Hancock recently where he said that for NHS, you know, he definitely wanted to see more health applications becoming available and more apps being used on the wards. And I think that’s when it becomes really interesting, when we stop talking about it and start delivering some of this stuff.
Speaker C: But then one of the issues is going to be how do you get that historical data if you, as you say, it is on paper? That’s a huge challenge to be able to get hold of that information and upload it onto computer systems.
Speaker I: Yeah, and you know, there is quite a lot. We did quite— not so much in the NHS, but in quite a few other vertical industries where we worked with customers on doing exactly that, digitalization of their paper-based systems, and especially where it’s forms-based. We have a product called O2 Smart Compliance that does exactly that. It takes task-based systems and says, you know, this is a form that you’ve got, and what you do is you tick boxes, you write things in boxes, or you take pictures of things, and it takes that from being a paper-based and transforms it into a digital format. But it doesn’t stop the way that people work. So it keeps the same format as what they would have in a paper-based, but it just turns it into a digital form that allows them to capture that information. But it also allows you to scan in data. The problem that we’ve got is that a lot of this data is handwritten data. And, you know, although we like to think that character recognition is there when it comes to handwriting, there’s still going to be human error that’s, you know, there that we need to correct on that side as well.
Speaker C: You’re listening to Password on Resonance FM with me, Peter Warren. After this, you can hear A World in London with DJ Ritty. So far, we’ve heard that the National Health Service has clunky technology and that the government does too. In the AI-powered future, I might live to be 100, along with half a million other people living in the UK in 2060. Dr. George Leeson of the Institute of Population Ageing has been telling me that this is going to be a huge problem.
Speaker B: I suppose this is one of the biggest challenges that we’ve been facing for decades now because we’ve gone probably within half a century gone from a situation where the majority of people would be retiring with a state pension to a situation where the state pension in no way can support a person in retirement and therefore more and more people of course have been encouraged or forced to have other strands if you like of pension, pension schemes, private savings. But even so, even so, it’s all linked. If we’re going to keep retiring people at 66 or 67 or 68 or 70 and they have to finance a retirement of potentially 30 years, then it’s difficult to get the maths to add up on that.
Speaker C: So in those terms then, Well, does that mean that we will have to be fitter, that we will have to lead healthier lives if we’re actually going to live something of an enjoyable retirement?
Speaker B: Yeah, I think obviously when we talk about these life expectancies, we’re talking at population level, and there is a huge variation across our population in those life expectancies. With significant differences in life expectancy from the most disadvantaged areas of the country to the least disadvantaged areas of the country. So if we’re looking at— and in a way that’s one of the biggest issues of inequality that really needs to be tackled. So obviously our lifestyles and our behaviours across our life course impact on our health. Medical science is— has been at least, whether that will continue in the future, has been able to enable even people with relatively poor health who’ve maybe had negative lifestyles to live relatively long lives., but they will spend a larger proportion of that life in ill health compared with those who’ve lived healthy lives and had healthy lifestyles and have had the resources to be able to do that. So do we want to remain healthy and do we need to be active and do we need to look after ourselves?, then, you know, the obvious answer is yes if we want to have a healthy and attractive retirement.
Speaker C: Now, this is a situation that’s already confronting the Japanese, isn’t it? And the Japanese are responding to this because they say that— I think in Japan that there are 2 young people for every 5 old people. They’re saying they don’t have enough enough people to look after their rapidly aging population. So they’re going to, as you say, rely on technology, but that’s why they’re building robots. They’re building robot carers. Is that something that we’re going to be doing?
Speaker B: Well, I’m not sure robot carers alone are the solution to this, but you’re right. I mean, Japan, the demography of Japan is more severe than in most other countries around the world, but you know, even in Britain even in Britain, the demography is such that it is a huge challenge in terms of health and social care. And obviously technologies could be just one of a number of ways in which that need for health and social care could be addressed. And then again, we have to look at what are the pros and cons of that. So do we want to be looked after by robots, which one could argue would lead to even greater isolation and decline in well-being of older people than we would otherwise see?
Speaker C: Although, I mean, old people say that one of their biggest worries— there was some recent research that was done by a building company. Old people who say that their biggest concern is being lonely. If an artificial intelligence agent, if a robot ensured that they didn’t feel as though they were lonely, wouldn’t that be a good thing?
Speaker B: Yeah, I’m sure for some people. So we have— have you come across Paro the seal? Paro the therapeutic robot? Yes, yes, I have, yeah. Yeah, so we’ve got two of those here and we do some research with those with older people. And in fact, yesterday I was visiting a community centre near Oxford for older people and with Paro. And I think, I think, you know, as I’m saying, for some people Paro could be a very useful companion. And several of the older people at that community centre yesterday were giving voice to that feeling. And saying, gosh, you know, I’d love one of these, keep me company. And this is a seal, this is, you know, I mean, and obviously as technology has developed, who knows, we may even have humanoid to make us feel that we are not on our own. But I still think, I still think there’s an issue there about we as human beings looking at ourselves and saying, is that really how we want to provide companionship for people who are on their own.
Speaker C: Good question. Only last night I was stopped in the pub car park by an elderly lady who said she was lonely and would I come and have a drink with her. Of course, you can’t compare a cuddly seal pup with a night out in the village pub, but is it better than nothing? And how do we protect an old lady’s personal health choices? Health data when she can’t even remember her own name? Mark Deane, a senior lawyer at Cooley, is working on ways to balance the loss of privacy with the gains to society from sharing our health data.
Speaker J: I think ethics is very much more of a mindset rather than anything that can be hardwired into the way an organisation works. I think we have to recognise that ethics can be, to a certain extent, incentivised within a business context, but we equally have to understand that not all businesses are operating on precisely the same trajectory, and it may well be that companies are in it perhaps for a shorter-term gain rather than anything more long-term. So we have to be very careful as to how we go about implementing ethics or seeking to implement ethics. As I say, it’s something that’s very hard and very difficult to hardwire. But in many ways, at the heart of this is a concept of transparency, and I think it’s almost Newtonian in its outlook. That is to say that the desire of privacy on behalf of the data subject should be equal and opposite to the desire for the lack of transparency by those people that are seeking the use of that data. And so therefore, if we can set it up as that balance like a Newtonian equal and opposite, then we can actually find a way of driving the individuals to lower their standards and their interest in the security or the privacy, if you like, but only in circumstances where those people that are seeking that data are equally transparent in the uses to which it’s going to be put.
Speaker B: I see. So what you’re saying is that essentially If the government wants to have our data because it wants to kickstart an AI industry, then it’s got to be extremely clear about the data it wants, why it wants it, and what it’s doing. You’re saying transparency on both sides.
Speaker J: Transparency on both sides has to exist because only when you have that degree of transparency that you can genuinely seek to get a form of consent for that data being used. Now, we all know and we all expect that those people that are seeking to get hold of data at this moment in time may not be fully aware of the uses to which that data may be put a month from now, 6 months from now, or even a year from now, but that doesn’t necessarily excuse them from not being transparent about their aims and their aspirations for seeking that data in the first place.
Speaker B: But it’s interesting, isn’t it? Because what you’re doing is you’re saying that there’s a need for a behavioral change in a company. What you’re saying is, yeah, the company’s got all of this data, it suddenly decides it wants to do something. What you’re saying is that somebody, that company should then say, I’m going to go back to all of the people I’ve got data from and I’m going to say, this is what I’m doing, are you okay with that? That’s the sort of behavioural change you’re asking for, isn’t it?
Speaker J: That would be one solution, certainly. And another way of looking at it would be for those people that are holding the data to say, well, there are some genuinely very personal aspects of this data that we don’t need. Is it really necessary for us to keep the, the name, the address, the precise date of birth of this individual for us to necessarily get some value from the data that we’re holding. Strip away some of those core data fields, and it may well be that we can get to a situation where the data offends the individual a lot less than if it was kept in a more anonymized format.
Speaker B: I mean, it’s interesting, isn’t it, because this whole notion of anonymization— the NHS are just putting in a system which allows them to match from your mobile phone number and your date of birth, your medical record. So when you ring up on a mobile phone to make an appointment, all of that information is instantly there. Now, what they’re saying is that’s anonymized because the data is held in some other place, and that record and that phone, that phone number and your date of birth are— the computer system just scrapes through the database until it gets, gets to that and unites those two. They’re claiming that that’s anonymous because one bit isn’t really seeing the other bit. But you could quite easily reverse out that process. So this issue of anonymization is going to be something that’s going to be very, very difficult to implement.
Speaker J: It will be. And therefore, we have to make sure that we have in place the necessary security protection so so that data, if it is obtained and is collated and is stored, is done so in an appropriately secure way so that it cannot be interrogated or vulnerable to third-party threat vectors or threat actors. What we also have to do though is we have to try to explain to the public where the value is in third parties or other businesses holding on to this data. I think over Over the past few years, we have ended up in a world where data has been exploited by a number of different companies for commercial gain, and it’s made people very skeptical about the uses to which their data is going to be put. If, however, we can get into a situation where people can understand the very real benefits that they might experience in circumstances where healthcare professionals are able to access that data in real time and the data is kept in a secure environment and up-to-date and accurate, then I think that we might start to see a sea change of people’s tolerances to that data becoming more widely available and shared within the healthcare sector.
Speaker C: That’s all for this month’s Password, so if you’ve just joined us, you’ve missed it. The show is repeated on Sunday at 2:20 in You can download the podcast at Mixcloud. I’ve been talking about artificial intelligence and health and meeting some of the main players. The debates continue. How to protect patient confidentiality? Is it right that we place our well-being in the hands of automated gadgets run by private companies? What does this mean for our National Health Service, for the doctors and nurses and the taxpayers who fund it. These live issues will be very much center stage at our Donald Mickey AI and Health Conference in London on Wednesday the 3rd of October at the Institution of Engineering Technology in London. You can join in the discussions and make your voice heard. You can find full details and register to attend at the website theiet.org.uk. Password is produced by Blue Buffery and scripted by Jane Wyatt with me, Peter Warren.
Speaker B: Thanks for listening. Goodbye. This program has been brought to you by Resonance 104.4 FM.
Speaker C: If you liked what you heard and want to support our work, please make a donation at fundraiser.resonance.fm.
