I'm so excited to welcome you all to this panel today um first of all I'll do some quick housekeeping um so I'm sure a lot of you already are used to these online panel events but as usual um if you're having any problems with audio please do let us know in the chat function um you won't be able you won't be visible or audible during the event otherwise but you can make your comments um using the chat so I'm really hoping that everyone will get stuck in and ask questions in the Q a function um there's also a transcriber doing love subtitling of the event and you can access these by clicking on the closed caption button at the bottom of your screen and if you want to tweet along please do so by using at Ada Lovelace Institute and all our panelists have hashtags and I encourage them to share their hashtags if they want to um so moving on to today's webinar um first of all my name is Mavis machirori I'm one of the senior researchers at the ETA Lovelace Institute um and the other lovers for those who don't know is an independent Research Institute with a mission to make data and AI work for people in society and were established by the nuffield foundation in 2018 and this is in collaboration with a few organizations such as the Alan Turing Institute the Royal Society British Academy Royal statistical Society welcome to us illuminate Tech UK and of the National Council and bioethics and our work covers many different strands including ethics public engagement public sector Technologies and more but today's webinar is brought to you by the health and covert Tech program um and I think just a quick thrill on why this webinar is really kind of important to all of us here um on the panel is that as everyone now knows Kobe that's the first pandemic of the algorithmic age and there have been so many solutions that we've seen being deployed to kind of deal with this crisis and a lot of them have Incorporated data-driven Technologies a lot of them at scale and quite quickly as well and so this has brought in new opportunities but also new risks examples of Slash technologies that we've probably all interacted with include Public Health identity systems contact tracing apps um and obviously the kind of linking of a lot of a lot of large Health Data stores but there's a real risk I think in all of this that their social inequalities or inequalities in health that are expanding some new ones some exacerbating um already existing um previous ones um and from for all different reasons and I don't think we can specifically say this was the cause but for a lot of different reasons and a lot of the kind of ways in which these technologies have been deployed actually exacerbating inequalities um to different scales for different communities um and so tools such as infant tracking like I said or digital contact tracing um apps have actually kind of marginalized some members of the community and one of the worries is as we start to deploy machine learning and AI systems on top of already existing kind of inequalities and the ways in which our Public Health structure currently is this is going to get worse but the whole area is also complex and messy what we can't do however sort of sit back and say well someone's going to fix it and this is why the health foundation and Ada Lovelace partner together um in 2020 2020 to start to look at what was going on in this space and what we've been doing is a partnership project that has been asking questions around well how have data-driven Technologies been interacting with um Social and Health inequality as well inequalities in health and some of these have a social determinant and this is only a small part this is only we've been doing work across four different work streams we started off with the data divide and we can put that in the in the chat if you haven't already come across it um we're doing a landscape review which is looking at the decision making that has happened within covid um we're looking at the actual experiences of members of the community working with an organization called Apple Collective um to really kind of map out the ways in which these technologies have interacted with people's everyday um lives and experiences so we can tell the story of what technology has been doing um and another work stream that we're also using to generate this evidence is one that is looking at a specific technology looking at what decisions go in what data goes in and what the impacts are on particular demographics but this is not alone enough we need to open up the discussion to other fields I guess and have a really good interdisciplinary discussion about what it is that we need to do if we're going to design better systems for future pandemics but also for being to take some of these Technologies and and see them becoming the norm and with that I've already spoken enough I am so delighted to welcome our panelists I'm going to ask each of them to come in turn uh to introduce themselves and give us a short provocation about what they see has been happening in the space and hopefully this will open up discussion about where we can go for the future and with that I'd like to start with my colleague Josh Keith from the health Foundation Josh would you like to come on and tell us a little bit about yourself and about data strategies and inequalities input space uh thanks Mavis and but thanks for inviting me to be part of today's question um I'll start with a few words about myself so as may have said my name's Josh I work at Mill Foundation uh where I'm currently uh interim assistant director of data and the sixth um for those of you who are not familiar with the work of Health foundation and we're an independent charity committed to Better Health and Care for people from the UK um we we try and attribute that mission to a number of ways and one of which is our work on based around the suits which I'm part of I've spent my career working in research analysis with a particular focus on the role of innovation and technology and in Health and Care and more recently especially looking at role data and basic Technologies in that context um and today's topic so data-driven responses to the coronavirus pandemic and particularly the question around the impact of these Technologies on existing social inequalities and health is something that's really important to our work at the health foundation and we've been focusing for a number of years on both improving the use of data in Health and Social care um and obviously said that the challenge of producing inequalities in health which we know from some of our work and and work with lots of others have been growing rods and narrowing in recent years have been high priority first last five years or so that's really the root of the research partnership performance theater of this industry on this topic and which made us briefly introduced um so Mavis asked me to talk um for a few moments around um the rapid deployment of data interventions and Technologies and their possible impacts on Health and Social inequalities and so I just thought I'd use the pandemic as a window to to look at that given that focus of today's discussion um so if we think back to before pandemic we'd actually already been witnessing an increased focus on the role of Digital Services later of approaches such as auspicious intelligence across Health and Social can't be on for a number of years we've seen the creation of an hfx we've seen the launch of the NHS AI lab with a substantial 250 million pack investment and a number of other things across the system that signaled the kind of direction of travel at policy level and then over the past um two two and a bit years now we've seen a much more rapid deployment of some of these Technologies as part of the Health and Care system's response to the pressures of the pandemic so from the rapid deployment of remote consultations in primary secondary care using pulse oximacies to help monitor patients with covert at home um the motion discussed the contact tracing app that made us touched on and then um kind of a step removed from that the central role that data is played in planning a lot of the response including the identifying members of the population who might want to consider the shielding because of the being clinically extremely vulnerable and to helping um deliver and prioritize the vaccination program so those data and digital and data driven technologies have been a really key feature throughout I guess one of the questions of today is what are we learning from this um one thing you're reflecting on is that the experiences has definitely showing a lot of positivity about the role that data for data driven Innovations can play in shaping Health and Social care but for all that obviously it's also really safe to highlight and maybe even reinforce some things that people have been thinking about before then the many questions that remain to be answered and charges that need to be overcome if that uh Innovation Based on data and data and Technologies is to have a positive impact and especially if that impact is to be equitable so so as an example some of those questions so how do they driven Innovation to Technologies if this with existing social inequalities in health so in some of our work and concluding um work we've been doing with Ada we've seen that the development of data driven tools can both be affected by but also resulting structural inequalities and biases and that certainly one huge barrier to um that likelihood of these tools being if being used at large in the Healthcare System producing Equitable um outcomes the questions about access to the questions about whether those most in need of improved Health and Care might be amongst those least likely to benefit from data driven Innovations part of this might be about um issues of access relates to digital exclusion especially if these information is delivered by digital channels um but also beyond the transfusion related to the way which data driven Innovations are developed and tested and evaluated um which might uh further into existing barriers to access for health services so who is not included in the data that is used to develop a data driven Innovation for example will have implications to who that Innovation will work um well forward who it won't work for um and and another thing that we've seen during the panda because the really rapid nature of innovation which is uh a real challenge when it comes to especially to new approaches using new technologies and it's one of the the questions that a lot of people are scrapping with at the moment is is how do we reconcile the speed of innovation and of deployment if there's Innovations with making sure that approaches to um evaluation and testing and deployment do include the time to build evidence and and be more certain of the impacts um impacts generally but particularly Focus for this discussion how technology how Technologies might work differently for different groups or in different contexts and balance that the risks of innovation but also was making sure that the opportunities to bring these benefits to everyone can be um to be realized in a I guess timely manner so not none of these things are easy or have a straightforward answer um definitely some of the things we've been working on with yet the place Institute um one thing to to reflect on is that um we need I think we need better evidence of the impact of some of these Technologies but also better understanding of the multiple different routes in which inequalities might be impacted by the rapid development deployment of Technologies and Christie what kind of um responses and approaches work to address and mitigate some of these um issues it's one thing to highlight the challenges it's a much different um uh for fish to them to out how those charges can be actually overcome in a practical way and then a final um thing I want to draw out was just around the timing of this discussion we're having today um so at the moment the policy focus is just think increasingly away from dealing with the coronavirus pandemic in response to that and focusing more on building on the use of data and data of innovation as the system moves forward so we saw in two weeks ago maybe three weeks ago now the final bet Russian Health and Social care published um I think there's a digital Health and Care plan coming at some point from from an HS transformation directorate um Health disparities white paper at some point in the summer so lots of strategies that and policies that are very relevant to today's discussion um and that make the conversation I think really relevant and it's I think really important that um as we move forward into kind of implementation phase there's lots of these strategies and plans that the lessons from the pandemic have already captured and embedded and that's part of the opportunity we have um so that's it for me uh thank you so much Josh that's a really great starting point and I think leads quite nicely to some of the work our next panelist um has really been trying to drive forward that issue of inequalities and how we address it and who comes in to help us address that I think is something that um I'd like to really hear from so I'm really delighted to pass the mic over to Tina Woods um and Tina is going to tell us a little bit also about herself and the work that you are doing and what you've sort of found within the pandemic and how we can think about some of these issues that Joshua's raised for instance Tina overt thank you thank you so much Mavis and and of course um it's been you know so wonderful working with you and and Josh um on some of the work um together so um as some of you know I wear quite a few hats but they all sort of coalesce around um really this need for a pretty bold system change um uh to kind of really to take what we have effect actively as a Sit care model that we have at the moment to try and to try and sort of initiate the changes that we need to see to really drive a more sort of preventable sort of uh sort of Equitable Equitable sustainable health model um so that's really where a lot of my my work has been focused on so um I run collider Health um I also run a social Venture called business for health um I just today um uh the announcement was made I'm going to be the uh working with the national Innovation Center for aging on uh on sort of healthy longevity and creating a whole new sort of focus around what that is and how we can really Drive change in that wholesale Arena um but I think really what we saw with the pandemic which is what I think we need to see now it's this the burning platform for the pandemic was we literally have to save lives you know and we all had to work together really really quickly and obviously saw a huge amount of energy around a shared purpose and we just kind of clipped together and just got done basically um we just had to do it and I would argue that's exactly what we need to be doing now and the burning platform for change is that we have a really ill population and that was one of the reasons why we were so hard hit from the pandemic we need to see that same urgency and that and see it as a platform for change if you look at some of the extraordinary figures I mean in 2000 you know uh you know we spent about um I think the Figures were uh yes 27 of day-to-day sort of um Public Service spending was on how but 2024 it's going to be 44 44 on NHS and care so that means less money available for actually all the areas that we really need to focus on to actually improve health because what we are seeing we have a chronic um ill health epidemic and and it's rooted in all and it's rooted in actually these wider systemic inequalities that have really sort of been exacerbated and it really sort of startled us into some action um so we need to really focus there it's the 80 outside 8 of NHS and care that we really need to be focused on yet are spending it's not it's going to be reaching 44 that has to be our burning platform for change so how are we going to do that so of course data is part of um the solution and we've seen that with with some with covet and we've seen amazing stuff happen with covert so we've got to bottle that up and sort of learn the lessons before we go back to the old old the old way of doing things we cannot do business as usual anymore we need far more bold Solutions so that's really what I'm really looking to do so the question that I was asked was you know how data can play a role uh for a more Equitable Public Health infrastructure and looking also the role of private sector um um Enterprise and health data use so um so for data to really work I mean I think the thing that comes up over and again we have to trust who is using our data so we act for the most part actually you know people do trust our NHS and care system with our data but I think um uh and so but we also we do need feed and this is absolute Paramount we needed the general public to feel comfortable and to feel that they can trust who they're sharing their data with for public benefit um and so the underpinning sort of paid environment and architecture has to be obviously interoperable we have to be able to share data really easily and that's still a real conundrum um to get to kind of you know sorted out but it has to be trustworthy so so this uh so some of the work that I've been doing has been all around that how do we create this trustworthy system and how do we actually tap into the vast amount of data that really isn't getting recognition actually and that's that 80 of the data responsible for the determinants of our health so um so looking at what we called private sector derived Health relevant data is a really really interesting area and of course combining that with public sector data and it's not just in NHS and care it's also other sorts of data it's also administrative data so people like Ian Buckner and Liverpool have done amazing work with their civic data cooperatives and CFA really harnessing local local intelligence Public Health bodies you know all the all the actors and including this you know this the the citizens you know who lived there they did really really well with with how they responded very quickly to the epidemic so we need to kind of harness those sorts of um uh experiences and also potentially data models to be able to do that sort of this Federated kind of data model so these are all the sorts of things that we started to look at and I'll mention one project which is the uh the open Life data framework project which is really sort of um initiated out of this recognition that there's this wider data ecosystem that we have to understand because it will also hold some of the clues to actually understanding a where the inequalities are but also more importantly the solutions for that so we published this framework we took some lessons from other sectors you know for example and we worked with Gavin stocks the open uh whose Pharmacy of the open data Institute but also the code was the co-architect of open Banking and of course some of you will know that open banking you know is driven by regulation of course we have on you know around the corner is a huge amount of energy that's going to be focused on after regulatory reform on data to kind of make the environment post-bread for this more agile and responsive to Opportunities Etc so this so in a parallel sort of way this is exactly what spurred open banking which and the whole intention around that was to open up um you know is looking at gdpr but also some PS2 Financial regulation to open up the data ecosystem make data more portable say listen Banks you don't own data it's actually belongs to the customer we want to make it easier for them to choose who they decide to bank with so it's spurred up this huge um in fintech ecosystem Challenger Banks and you know and so of course they have the same issues in many ways that we have in our health space in the sense that yes you know because you know consumers they don't want their banking data to sort of you know get abused you know in the same so some of the sensitivities are actually quite equal so that so the trustworthiness of the system I think there are many many sort of um lessons that we can take from that so and that is indeed what we did sort of looking at the Open live data framework which incidentally was published um uh in in November with the support of our George treatment Innovation science Minister who um you know is is incredibly sort of excited about some of the opportunities for us as a nation to be able to crack you know um leveling up missions which has seen a recommitment to healthy life expectancy but most importantly it's a reduction of health and well-being disparities so um so what we did is we looked at this framework and um uh looked at how we can create this trust Nino this trustworthy system you know consent based mechanism so we looked at all that it's been published and we're now hoping to take that into an other sort of a way to test out some of these ideas because in the end you need to test stuff out you need to see whether it really works in the real world so we've got a number of use cases we're looking at um but I think importantly one of the things that we want to do is and this is very much part of some of the the regulatory reforms you know potentially at play here in data for example is to take um as to look at initiatives like the government are looking at the moment with smart data for example how to make it easier for individuals to share their data held by private companies with trust of third parties um and how do we and how we can inform interoperability standards to make it easier to share data to support Innovation so one of the ideas that I'm looking at you know this is my work with the national Innovation Center for aging is actually to create sort of you know these sort of you know these healthy longevity scale boxes in the same way that we had in fintech is actually look at these cluster of organizations working with public sector working with you know the whole Myriad of actors you know that we need to be working with to come up with insights um to understand where the disparities are but but actually fundamentally is to develop the solutions to that um you know and potentially looking at um you know and of course AI how do we harness AI so um and I think there are some huge opportunities you know if we look at the national AI data strategy in the National Data strategy it's very very focused on NHS and Care data so there's a massive opportunity and this is some of the work that I'm going to be doing to really look and see how it can be much more bold and radical in terms of where how what health actually means and where health is outside NH just some care data so there's a lot of stuff that I think with the NHS and Care data stage live strategy which actually philosophically I think says All the Right Stuff implementation is the hard part I think implementation came up a couple of times so yeah so we have to look at how do we come up with new data sharing models where we can also we can also harness private sector data um you know for public benefit you know what are those commercial models going to look like so there's a whole bunch of work ahead where we have to test out some of these use cases um now I'm just going to say a word to some the other work that I'm doing which is very relevant which is the business for health work and we're taking um some of the thinking out of data to feed that into the development of what we're calling the business for health index we're working with the CVI we have support from um Chris Woody or CMO we're working with the office for health Improvement and disparities in fact we're sort of submitting evidence and case studies as we speak it's just about to be published and what we're looking at there is okay hang on we've got the business Community this is part of the massive solution that we need as part of the system change approach how do we get the business Community to do more to kind of reduce that NHS and Care demand we know that mental health muscular there are massive areas where business dramatically reduce demand so we're looking at the development of a framework working with the owners Health index team so we want to line it all around that health is an asset to invest in um we're looking at the role of business in Workforce Health massive area that CBI we're literally about to develop a diagnostic for launch in November that you know improved Workforce health and actually good quality work leads to good health health equals wealth how do we support local employers local businesses to do more for the local populations with good quality work with good access to jobs that's going to drive up health and local communities that's exactly where we need to be with place place based Health role of social Enterprises Charities working together solving problems in local communities so that's a fundamental part of what we're doing and of course products and services Food Systems you know ultra high processed food massive area we need to tack we need way too much junk food housing we need much better housing these are all ways in which the business Community can be incentivized and lastly I will just say the big idea behind that is um bringing Health really into ESG investment and Innovation so massive energy and capital existing in for example the Pension funds institutional firms their new regulatory um freedoms being given to Pension funds to invest in higher risk Innovation areas there are trillions of capital that could actually be invested into strategic products to deliver more on reduction of Health disparities and making it worth their while in the same way that we see in climate and ESG investment Net Zero let's do the same for health let's get the big big money to actually invest in reducing Health disparities because it will lead to health um equals wealth for all and we will improve you know our prospects as a society because we'll be in better health so I could go on forever as you can probably tell there's an enormous amount um that needs to be done but we need to be really bold and we really really need to look at the wider reasons other reasons why we are so ill it's not about NHS and Care Systems only it's a much bigger thing than that so I'll leave it there really great provocation I'm starting to think our webinar should have been maybe three hours long because there's definitely lots to talk about but it's a great point that you make right about we need to think more Beyond sort of the um the usual determinants of health and really think about what else people are needing um not just access to services but people need clothes and they need food and they need to be able to pay their gas prices I think these are all issues that come together um but one of the things that you said was about communities and being able to not lose that momentum and have you know system change one of the critiques that sometimes comes up is when we think about data driven systems is that the focus has really just kind of been on Innovation within the NHS for instance or is very UK Centric and sometimes I think we forget that actually the pandemic and some forms of data driven systems continue to be deployed outside of the UK system and you talked you talked about you know transparency see as one of the mechanisms in which we can think about doing this better I'd really like to hear from Kira as somebody who has a more global perspective about you know what you can reflect on what you've heard Josh and Tina say or more in terms of you know the research um expertise that you bring to your work and just tell us a little bit more about how you see coveted unfolded where the gaps are um and yeah I'll hand over to you thank you Mavis um and thank you to uh Tina and Josh for setting up this conversation nicely um so my name is Kyra Staunton I'm a senior researcher at The Institute for biomedicine and which is in Europe research and we're located in Bolzano in the north of Italy um and so for the last two years well generally we look a lot of the governance of Health Data supporting the scientists at our Institute but in the last number of years our group has been looking more on the ethical and Equitable use of that data another hat I wear is that I'm a consultant at the South African Institute National Institute for communicable diseases since about 2018. um and since then I've been working with them on the management of this stage for public health particularly in the context of TB and HIV and of course some conversations come up um now when we look at Kobe you know it was our first digital pandemic as with any pandemic we were heavily reliant on data but there was also this emphasis that Mavis mentioned on digital Technologies to help us understand control limit the spread of the virus politically these data and these Technologies they were used to very much limit our rights and freedoms um now we've seen a lot in the discourse in recent months and particularly in the last year and what went wrong and what we've learned and today when we're talking about designing more Equitable Health and Social care system I'm going to focus on data intensive research methods which is very much the area research of mines and the problems that are occurred that are inherent because of the up that they're inherent in our current Global health research systems so very much with this Framing and there's three points I think I'd like to make first the current system is not delivering the data necessary to ensure Equity benefits second a lot of the problems we're seeing is rising due to the open science and data sharing research agenda which is very much serving uh Western and particular communities within Western society and exclusively and third one of the conditions for equal use of this data must be having must be resting on equal access to the benefits of that data and as we saw during the pandemic current systems is really not meeting that so if I take it off with the first point in looking at our Global Health System historical Global health and Equity has meant that we we've heard a lot that research is being predominantly focused on diseases in high in some countries and because of exploitation discrimination a lack of trust and a host of other socio-historical reasons that there's many reluctance of some groups to participate in research now our research can only be as good as a data that we used to feed it because of these problems that we're seeing from from exploitation discrimination so on we're seeing that there are gaps and the spices in our data sets so there's an underrepresentation and at times misrepresentation of many population groups and therefore research on the outputs of research not targeted at these groups and again we saw this this happened again when it came to the rollout of covet um uh trials now if you look at the second point that I'd like to make and how this whole system with the current open science research agenda at the beginning of the pandemic we heard from governments funders scientists importance of share share share you know it's important for solidarity that we share data for research and we saw that the quick vaccine development uh has been very much credit in part to this global data sharing and while that is the outcome was good when we actually began to look at this data sharing and open science agenda it has been and always has been set by high income countries who have the access to Nursery skills infrastructure and Personnel to ensure that that enables its researchers are based there to benefit from the station sharing now we began to see particularly last year a pushback from to this open data sharing from many researchers particularly in Africa and this pushback was in part due to this asymmetry in infrastructure resources and capacity that we see between High income countries and low-income countries and data generation storage analysis and so on now these asymmetries very much have the roots and this exploitation that are mentioned in Equitable funding racism and other forms of disenfranchisement but it's what it means ultimately is that the current data sharing agenda and research is principally serving those who have their necessary resources in place to quickly use and exploit the data now this brings us to the third Point Equitable benefits um and when we're talking about access access to data but what about the benefit to the outputs of that data now the work that we were doing we're very much arguing that access to station must be tied in some way to a clear benefit um and so beneficial benefit sharing must be key in any data Exchange now I mentioned solidarity um and the calls for solidarity I think even one of the Kobe trials is called solar eidarity well solidarity very much has its roots in genomics research it's being urged that it should be a principal to guide genomics research and also now other forms of data intensive research and absolutely it's very much a loadable concept but covert demonstrated that to rely on this principle alone is not enough because solidarity will be continued to be exploited as it was in the pandemic for National and Commercial interests and if there is no and particularly if there's no clear and I would argue a legally binding process in place to ensure Equity of benefit because while we did have solidarity and people were happy to share their data for um this public Global benefit but when it came to the distribution of vaccines for example and also the pricing structure around this solidarity stopped and the national and economic interest very much took over now looking towards a more Equitable and Health and Social care system and how all of this kind of fits into like you know imagine the future I would argue that an equitable Health and Social care system must be supported by Equitable Global health research and if we have we look at this current very much neoliberalist some would argue approach to science does not change these issues are going to continue to arise during normal times and of course arise during any pandemic times now in the context of research we've had lots of solutions posed to remedy a lot of these issues that I've mentioned such as establish of academic research centers across the globe programs and data science training on Personnel ensuring that countries have the necessary infrastructure in place to support the science and also and what's key is that there's robust governance and in place to support this data use for research but a real issue relies with power and who holds the power in terms of decision making and funding now funding is critical to improve and address a lot of these disparities and issues that I mentioned but funding cannot and should not be dependent on the decisions of philot traffic organization or funding Industries or agencies that are in high income countries alone and equally the benefits from science and our data sharing Club exclusively led by commercial interests um I think what's important to remember is that protection of fundamental rights and interests and Commercial interests are not mutually exclusive what they do is require and when we're considering and conceptualizing this you know ideal future what we need to think about is how can we carefully balance these rights and considerations of all those involved in the data pipeline to ensure that there's benefit for all and that those um data donations and those who are data collectors are not exploited and so just to conclude what I would say is that when we are thinking about this future Equity must now be prioritizing at the heart of our data-driven responses and this will ensure that we're best placed to effectively respond to the next pandemic and this requires equity in resource allocation funding streams but finally and importantly equity in terms of power and in all power around decision making brilliant thank you so much here um so I'm hearing strands that connect so far all our panelists talks so solidarity public benefit Equity transparency whole benefits doesn't just stop at the data-driven technology but it also comes even before that in terms of the research and the data that goes into it and we can't just be insulin looking at um you know just local or national but you need to broaden this out because everybody benefits when we sort of have this Global agenda but I'm just wondering what this looks like and what regulatory mechanisms need to be in place or what standards need to be in place for those people who are trying to design these systems and this is where our next panelist Alison Gardner is going to really focus on so Allison I'm really delighted to have Unix on the panel and to tell us a little bit about what you think you're currently on it and well Allison is meeting um I really encourage you to put put some questions in the chat I know we're quickly running out of time but it'll be great if we can open the discussion of Alison avity sorry about that it would be me um hi I'm Alison Gardner um I like many of us where many many different hats I currently work in the multi-agency advisory service which is a cross-regulatory forum for the CQC mhra HRA and and nice um and you know a lot of positive work going on there I also uh am one of the co-founders and directors of women leading in Ai and we're doing a great project at the moment with equality now looking at uh um a universal chart for digital rights and um and a researcher with at Keel University and it's probably with the latter two hats I'm going to speak today but I also work on a variety of Standards um committees and developmental standards with the main focus on Healthcare and all the to do with AI and I was asked to talk about how you know standards could be used to help um you know well regulate and make sure that Innovation is done in an equitable fashion um and it is with that this I'm going to address a number of blockers that I offer meet when discussing an Ethics by Design approach for AI Solutions which occurs even in normal situations let alone the heightened situation of an emergency firstly the common um rebuttal um that any regulation standard standardization can do is that it will inhibit innovation it doesn't it does inhibit poor irresponsible and unethical Innovation as the introduction of the gdpr producer we lost quite a few apps well those apps therefore were not data you know privacy first apps then if they couldn't cope with the gdpr um but innovate in regulation standardization does promote good quality human rights protecting for Innovation so if you automatically default to regulation inhibits Innovation myth then I think some self-reflection is required there and in light of that I have to say that we need to move away from um the bro text style of move fast and break it um fashion of technological development which some might think it suits um emergency situations but actually I think it can be harmful and we need to move towards a collaborative robust reproducible and ethical methods of innovation and all we all know don't be that key to this is minimizing bias in order to reduce discriminatory outcomes to in all order to avoid hard coding and Equity into all our Technical Systems and I will say at this point that I'm not really going to comment too much about data because I think a lot of focus is on data but you know bias and harmful outcomes can be actually embedded throughout the entire life cycle of developing a model and you know even if you think you've got the most Equitable and fair data set ever going forward you can still actually in build and bias into a into a model if you're not careful and so one call method to ensure that we correctly identify buyers and mitigate for it it's of course to ensure that we have diverse and accessible multi-disciplinary and human-centric design which leads me to a second myth I come up against a lot which is that requiring meaningful and full life cycle input from diverse stakeholders will inhibit startups as it is difficult and expensive to do well I challenge this if you want to be Innovative then be Innovative work out how to make sure this happens if you can't have diverse teams that engage with diverse users Implement ethics board citizens juries Etc organizations such as either Lovelace have produced copious amounts of guidance and good examples and there are plenty of good examples that I come across within the health sector that embed human-centric design into their culture and a keynote actually as a side point for all startups if you're building your your team make sure it is diverse from the outset so that you build an inclusive company culture straight away because if you don't and you try and do this retrospectively it is very difficult to do and often meets resistance and have ineffective policies so talking about standards as we know technology tends to evolve much faster than standards uh set to ensure such emerging Technologies are effective safe and fair I always add Fair hence we often find Technical Solutions implemented at PACE Stevens outside Health emergencies before there are any appropriate standards and regulations being developed and you're playing capture so it is from bet but it is from this generation of technical innovations that we've experienced all the problems and harms that can occur and we now we can take learnings from them to develop the standards that will recognize and mitigate for future harms so I am a member as I mentioned of a number of Standards committees and I've noted a few interesting points firstly standard development often experiences similar problems as technological development in fact in that there is a lack of diverse voices and certainly from um you know citizen-centric organizations and representation from communities that will experience the impact of these Technologies more negatively than others and I often find myself having to speak quite assertively regarding the importance of diversity and inclusion as core principles and ideally always forget accessible as well and that I will mention that specifically separately as well and for such principles to be practically embedded into the standards not just a tick box oh yes we've thought about it you know big things but we have to have meaningful solutions to ensure that this occurs and that meets resistance and in one working committee and I won't name them we had to list the UN sustainable development goals relevant to the plan standards and of course sdg9 which is that industry Innovation and infrastructure and if you go deeper into it's about inclusive Innovation and sustainable Innovation but those won't have been the key buzzwords that stuck out when it was suggested but I had to argue quite strongly to include sdg5 which is gender equality and sdg 10 which is reduced inequalities into that standard and this standard was about you know how we produce and you know you know good safe effective and fair AI Solutions and unfortunately we had a good chair so I won that one and some standards are working very well towards this so IEEE p7003 which is considerations for algorithm algorithmic bias uh which holds at its core stakeholder engagement and an algorithmic risk and impact assessment so regarding stakeholder engagement it crucially identifies two types so obviously we've got the development team from the business leads to programmers and The Wider team around that but it also as a main group involves the impacted stakeholders I I will call them and the standard implicitly requires that engagement should occur and that monitoring for buyers should occur throughout the entire life cycle and there are other standards in developments so I'm involved with BSI Amy 34971 so you standard Geeks out there so that's guidance of the application of iso 14971 and you'll all love that risk management for medical devices aim software as medical devices so and that's an addendum to that one uh and the biases isn't written into that one there is a new standard coming up the BSI 30440 for AI in Health and Care and each and that also considers a bias on diversity and inclusion and accessibility you know and and I've and I've been member of these purely to push that agenda to make sure it occurs in the whole life cycle approach note in order to make any of these processes meaningful of any sort of ethical solution I'm aware of the time Mavis or going quick as I can there has to be a go no-go Clause so it's a risk of mitigations are not proportionate and they're unappropriate then the tech should not deploy there's some good examples of this the West Midlands police force refused to to implement a very expensively designed with um uh predict policing tool for serious knife crime and because they weren't happy with it there are so many examples of good practice and the and again within the you know the health sector that is being exam you know there's some examples of that so Ada Loveless is um AI um uh data sharing um assessments impact assessment is a good example so there's our other Suites coming up so ISO iecse 42 and the Sensen elect JTC 21 also that which supports the eua of AI euai regulation but the problem that we have with sponsors is that by very nature they tend to be a mandate and mandated so um although some might be designated and recommended and you know formally recognized as a gold standard to have and it'd be quite difficult to deploy without them there is a lack of very specific mandate and often this is because we have cow tying to promote Innovation demand demands and not wanting to impose upon develop her concerns such as IP and costs rather than consist citizens concerns and that brings me back to another bit of the blocker which is of course inhibiting Innovation myth that it actually any types of standardization or regulation or ethical governance requirements will slow down the response in ordinary development but you certainly an emergency and situation such as covid but I think Tina might have mentioned it already and here sorry apologies got it wrong but the vaccine examples of vaccines shows us that that's not the case you know the controls and developmental factors are really really strong they managed to to develop those vaccines by adhering to all the ethical governance and testing requirements due to the urgency focus on funding such developments and we're unable to occur to speed so there's no reason why you cannot you know do technological development ethically at speed in in in difficult circumstances finally um I will mention the learning advice put forward by a variety of superb researchers and the arguments about the contact tracing apps were amazing if you want and Twitter and and you know the fights um to do with that and out of that such many academic workers come through so I want to just name check um or you know um examples such as the UK pandemic ethics accelerator um The Observatory for monitoring data-driven approaches to covid-19 and the Bingham center for the rule of law who reached over out and Twitter and sponsored this webinar as you know and there was a lot of learning and key to many of these are requiring impact assessments be they algorithmic risk and impact assessment technological impact assessments extensions of dpias you know they all mean the same thing um and in those their requirements for built-in decommissioning of emergency tools and controls the scope creep and dual use and the capability for ongoing monitoring and key transparency and and so these are things that we must remember members of Civil Society in Academia are vital to ensuring that any technological response to a health emergency is defined time-defined Equitable accountable any benefits that occur from the developments for future ongoing use because we don't want to bin everything up pandemics ended let's get rid of all of this Tech now there's good stuff there but it should be re-scrutinized and assessed independently of the emergency response to make sure it remains um ethical and I believe technical technological response crisis is at its core very well meant and I am actually really confident that regulators and standard bodies are cognizant of the risks and are indeed responding accordingly and some examples have been named already and for me in closing 2020 in particular was a pivotal year when the issue of fair algorithms hit the mainstream including the Streets of London where people were chanting against algorithms and I'll just read the message off Quail was fine I'm going to defend them because they could shout to get us off because they knew it existed because they were transparent on their website so people knew to challenge and exercise their rights and so probably let's always not be horrible about off-paul um our developing regulations and standards are taking a learning from the recent pandemic on board and I am hopeful we can build the infrastructure for good governance for Rapid ethical responses so that we are ready for future emergencies and I shall end there yes jump in and say we need to open this up so thank you for that um I'd like to I think I'm not sure if everyone's on screen but um if we can get all our panelists on screen and we have time for maybe two questions um and so I see that we have um so just there's a comment in there um in the in the Q a and I'll just read it quickly um thank you um the panelist was wondering so this is from um I won't say the name in case we're not saying people's names uh but from JW um do the panelists do any of the panelists have thoughts about the application of data governance act or the data act and how it can help support some of the solutions proposed around interoperability standards creation more Equitable health and is there going an ongoing EU regulatory development um that can apply um or can be considered in the UK so we'll start with the first bit um really quickly if possible in a minute or less this is about standards and enforcing interoperability um and the application of data governance is there anyone who wants to take that uh Tina do you want to have a step in a minute or less yeah I mean I mean that's there's there's quite a lot in that and I think um Allison will probably be able to advise in some of the specific uh details of that so yeah I mean some of this has been dealt with with the bi standard that Allison had mentioned in Ai and Health and Care but I think um uh I I think going back to the principles these are absolutely core and I really really agree with Allison for example on the diversity inclusion criteria because I think it's it's core principles like that that has to underpin everything um because we don't get that right everything else will go out of kilter and so so I do think you know this whole question around you know data for what purpose and and and really understanding what are we trying to do here with the data so that that relates to principles that relate to what we as a society value it's it's it's all those sorts of questions that need to then you know be reflected in the standards so I'll just I'll stop there anything to add 30 seconds or less yeah I I I think the intention is certainly there um and I and the only thing is that I'd I'd be a little bit resistant to say anything at the moment because there's still lots of work particularly for the mhra just getting things mandated but there is good consciousness of of that issue and and monitoring of that so I I think maybe that there was less concern um because it's about getting the balance right between promoting Innovation and ensuring Equitable care and and human rights perspectives as well but I think the conversations are going in the right way encouraging and I'm wondering if we can hear from Josh and care about some of the potential barriers so we talked a lot about equity and transparency but what do you see as potential barriers to implementing an um kind of inequalities equities thinking in the ways in which we can design future pandemics either a barrier or you know an opportunity because we don't always ought to be negative on a Monday sorry Mavis do you mind just repeating that again I just lost you a bit yes so I'm just wondering if you could think of maybe a burial opportunity in the ways in which we can start to bring all these together in a way that embeds Equitable thinking or a ways a way to reduce inequalities in the ways that we transform the future use of um data or data driven Technologies I'll come to you first and then I'll go to Josh yeah I suppose just very quickly and briefly from a governance point of view which as a legal academic a lot of my work focuses in on that I think we actually need to start having Equity as embedded as a guiding principle in a lot of the work we do um so in in the in 2019 in July 2019 and um I was approached by the ACT accelerator to develop their governance station governance framework for the use of their data and digital Technologies and one of the key Focus for them and the discussions that we had with them and also the who ethics group who worked with us was in ensuring Equity throughout both equivalent the data but also Equity of benefits and that was very very much a guiding principle and for me I think that's key um in in governing Day show as we move forward biggest potential ways in which we can um harness these kinds of thinking in Technologies yeah thanks it's um I was going to say well one thing I really agree with that Allison was saying is that speed isn't an excuse for not doing things right and I know I talked a little bit about the speed in my opening remarks and if it's the right tools and structures are there then speed is not necessarily it's not the the issue um and it should be excuse for not doing these things I think one of the for me one of the big challenges is bringing is the breadth uh broadening out our understanding of Health inequalities for bringing that together in a way that means that things can be tackled so I guess as an example of that lots of there's lots of things happening at the moment that are looking a bit very specific inequalities so there's lots of really good work uh some were led by the NHS race and health Observatory looking at um ethnic inequalities in in health outcomes and they looked at data and the charges of ethnicity data as part of that and and a health Foundation we're partnered with the nhsa AI lab on on work in that space as well but that's just tackling you know looking at um in a a still broad range of inequalities but from one particular angle rather than taking I guess a more kind of intersectional approach to understanding the different routes to which inequalities might interact with um data produced Innovation and it's not easy to get that right but it's probably needed to take us to the next stage if that makes sense it makes sense to me and one of the things that I think has been drawing all of these discussions together is it's not just about the data it's not just about the teams that are looking at the data it's all of them together and the conversation that we can have around that so suddenly having a joint up conversation is is the way forward we can't just have academics we can't just have clinicians we can't just have um people creating policy but we all need to sort of be sitting together to to discuss these and something that I think you're mentioned earlier was the use of genomic data sets and we didn't really go into the different kinds of data that is being combined to create these data-driven systems but there's definitely I think a conversation to be had that deals with you know the quality of the data and the ways in which it's coded but also the sources of that data and where those inequalities lie and how each of those kinds of different areas come together um in a cumulative effects to then have these Downstream impacts on inequalities but that is a conversation for a whole all other webinar because we have two minutes left only um and so I don't think there is a burning question in the um in the Q a and I won't ask anymore for times but I really want to thank all our panelists and I'll give you a round of applause and hopefully everyone else on the webinar is giving you a round of applause um it's been really insightful short but sweet but it's been really useful I think as a thinking for next steps about how we embed all of these kinds of discussions into better designs um for Technologies because if we don't do it well we're only going to exacerbate the inequalities that we have seen emerge um in covid um and I think if you're interested in any of the work that we're spoken about today definitely get in touch with either any of the pan
2022-07-11