DeepMind Health: Revolutionizing Patient Safety through Technology and Collaboration

The role of technology in healthcare has undergone significant transformations in recent years, with DeepMind Health being at the forefront of this change. The company's approach to working with clinicians is centered around collaboration and co-designing interventions in the system. This approach has been instrumental in developing innovative solutions that improve patient safety and outcomes.

One of the key principles guiding DeepMind Health's work is the importance of iterative testing and learning. By continuously testing and refining their solutions, they have been able to create a robust framework for identifying areas where technology can make a significant impact. This approach allows them to gather valuable insights into what works and what doesn't, enabling them to refine their solutions over time.

The design of DeepMind Health's apps is carefully considered to ensure that they are user-friendly and support the work of clinicians. The screens are clean and intuitive, with minimal information presented at any one time. This approach enables clinicians to quickly access the information they need without feeling overwhelmed by too much data. The technology should support the clinician's work, rather than the other way around.

The use of technology in healthcare is not just about automating tasks or making things more efficient; it's also about improving patient outcomes and experience. By leveraging machine learning and analytics, DeepMind Health is able to identify patterns and trends that might otherwise go unnoticed. This enables them to provide clinicians with real-time insights into patient data, helping them make more informed decisions.

One of the most significant challenges in healthcare is managing the intervention on once a patient has been identified as deteriorating. According to ARAs research, up to 50% of patients who are identified as deteriorating receive inadequate care. This highlights the need for effective task management systems that can help clinicians prioritize and manage their work more efficiently.

To address this challenge, DeepMind Health has developed an innovative solution called Hark. Hark is a system designed to support clinicians in delivering high-quality, patient-centered care. By leveraging machine learning and analytics, Hark is able to identify areas where patients are most likely to deteriorate and provide real-time insights into their condition.

Harks effectiveness was demonstrated in a study by ARAs Dominic, which showed that clinicians responded 37% faster with Hark than they did with traditional alert systems like Pages. This highlights the potential of technology to streamline clinical workflows and improve patient outcomes.

The launch of Hark marks an important milestone in DeepMind Health's mission to revolutionize patient safety through technology and collaboration. By integrating detection and task management into a single platform, clinicians will be able to access the information they need quickly and easily, without having to switch between multiple apps or systems.

One of the key objectives of the Streams app is to deliver the right data to the right clinician at exactly the right time. This requires careful consideration of how data should be presented, how it should be prioritized, and who should receive it. By leveraging machine learning and analytics, Streams will be able to identify patterns and trends that might otherwise go unnoticed.

Ultimately, the goal of the Streams app is to shift some of the 97% of activity in hospitals that is currently reactive towards activity that is proactive and ultimately preventative. This requires a fundamental change in how we approach healthcare, one that prioritizes prevention over treatment. By leveraging technology and collaboration, DeepMind Health believes that it is possible to achieve this vision.

The development of Streams marks an important step towards achieving this vision. By partnering with clinicians and other stakeholders, DeepMind Health is able to co-design solutions that meet the needs of patients, clinicians, and healthcare systems as a whole. This approach has the potential to revolutionize patient safety and improve outcomes across the healthcare system.

For anyone interested in learning more about DeepMind Health's work or sharing their own ideas for improving healthcare technology, the company invites you to reach out and explore its partnership opportunities. By working together, we can create a better future for patients and clinicians alike.

"WEBVTTKind: captionsLanguage: enwell good evening ladies and gentlemen and it's my privilege and pleasure to welcome you here to the Royal Society of medicine for the second name so name d lecture how digital Innovation can improve health care I want to start by expressing a particular welcome to the Dango Family David Michael Robert and Ellie and their partners and all of their family who are here tonight I also want to specifically welcome some other people kice Ellison who's been such a uh a benefactor to the RSM over many years together with Brian Whitt who's been a long-standing supporter here I'd like to welcome staff and students from the Westminster Academy an academy that's supported by the Dango foundation and which uh I just heard from David Dango earlier has achieved 19th Place nationally in the sixth form rankings which is a fantastic achievement uh in the last few years and we're also pleased to welcome representation from patients Association for this evening's lecture in a minute uh David danger Will introduce our two guest speakers Lord Dary of denim and Mr Mustafa suan so I will not say any more about them except to point out that Lord Dary has I think the singular distinction of having resuscitated some one of his fellow peers in the House of Lords so if any of you not feeling so well uh here's the here's the man to to to to help out you can say it's a heart surgeon telling you that as well anyway this uh lecture this evening's lecture is named in honor of s name Dango who sadly passed away last year at the age of 101 he was born in Baghdad in 1914 when Iraq was under ottoman Rule and he subsequently traveled to London to study engineering and on his return to Iraq went into business he built up a diverse portfolio of activities including property development and letting a match and Furniture Factory and he won the first contract to bottle Coca-Cola in Iraq which is also particularly Advocate as the headquarters of Coca-Cola right next door to us at the RSM in 1947 he married Renee Dango and they had four children following the birth of Israel and Arab nationalism things became increasingly difficult in Iraq and the Dango Nam Dango moved his family uh to Britain and for a few years he traveled between Britain and Iraq but uh subsequently events became very difficult in Iraq and he he permanently relocated to Britain settling in London he vowed that if he made another fortune in the wonderful country that had taken him in he would plow his gains back into philanthropic causes so he set up a commercial property business which his sons joined him in and and it soon began to thrive and with that success he was soon able to make good on his promises he started to give back to many organizations and Charities focusing on education and health in 2014 he made the largest philanthropic gift to the Royal Society of medicine and some of you will have seen the lecture theaters uh that is named after him as you came into the entrance hall this evening but he's also he also donated to the Francis Crick Institute and cancer research UK he funded over a thousand scholarships for students with no family history of tertiary education and gave £4 million for students from low-income back sorry from low-income backgrounds to study STM subjects so he had a massive philanthropic uh donation to many many people in many many different walks of life and these as I've indicated earlier include the Westminster Academy in London whose pupils hail from over 60 countries and as I said we're particularly welcome to to many of them this evening so name was made OB in 2006 CB in 2012 and in June 2015 became the second oldest person to receive a Knighthood he had seven grandchildren and two great-grandchildren who survive him in addition to his four Sons the theme of tonight's talk is medical Innovations and many of you will know that the RSM has a flourishing program in medical Innovations which has now extended to over 180 presentations many of our speakers have traveled from long distances to be here to speak about their ideas which will benefit patients others have traveled from not such large distances including Hammer Smith and King's cross in April this year we'll be hosting our 13th sum our 12th Summit with 13 speakers a large faculty from around the world but interestingly including a 15-year-old from Britain who won an innovation prize for Research into the early diagnosis of Alzheimer's so here at the RSM we're delighted to be able to encourage youthful Enterprise and energy and Innovation over 250 people have already registered to attend the April Summit and if any of you still uh in the audience who haven't registered wish to attend I would suggest you register fairly soon uh otherwise you may have difficulty in getting in so I'm going to invite David dangle to introduce our speakers Lord Dai will be speaking first followed by Mustafa selan there will be an opportunity for questions uh and discussions but I would ask you when we come to question time to please keep your questions uh brief and to the point otherwise I shall have to shut you up the closing remarks and vote of thanks will be given by one of our RSM members and also Junior doctor Dr KY MCG gretton so welcome to the RSM and David can I invite you to come up and say a few words good evening Professor Lord Dary of denim holds the Paul Hamlin chair of surgery at Imperial College London the Royal Marsen hospital and The Institute of cancer research he is director of The Institute of global Health Innovation at Imperial College and chair of Imperial College Health Partners he is also an honory consultant surgeon at Imperial College Hospital NHS Trust research by Professor Dai is directed towards achieving best surgical practice through innovation in surgery and enhancing patient safety and the quality of healthcare his contribution Within These research Fields has been outstanding publishing over 800 peer-reviewed research papers to date in recognition of his achievements in the research and development of surgical Technologies Professor Dari has been elected as an honory fellow of the Royal Academy of engineering a fellow of the Academy of Medical sciences and in 2013 was elected as a fellow of the Royal Society he was kned for his services in medicine and surgery in 2002 in 2007 he was elevated to the United Kingdom's House of Lords and Appo appointed parliamentary under Secretary of State at the Department of Health upon relinquishing this role within central government in 2009 professor sat as the United Kingdom's Global Ambassador for health and Life Sciences until March 2013 during his appointment and Beyond Professor Dai has developed his status as a leading voice in the field of global Health policy and Innovation Professor Dai was appointed and remains a member of Her Majesty's privy Council since June 2009 in this year's New Year's honors it was announced that he would be admitted to the order of Merit this award is the personal is in the personal gift of the queen and is limited to 24 living recipients like my late father s Naim dangu Lord Dari was born in Baghdad and he confined in me just now that he went to the same school as my father and then moved to the school that I went to in Baghdad and he has given much to this country so I am particularly pleased the Ed that he has agreed to give this year's Sim Deno lecture Mr Mustafa suan who will follow Lord Dari is co-found founder and chief product officer of Deep Mind Technologies a leading artificial intelligence company backed by the founders fund Lee kashing Elon Musk and David bundman among others the company was bought by Google in 2014 in their largest EUR European acquisition to date he is now head of Applied AI at Google Deep Mind responsible for integrating the company technology across a wide range of Google products at 19 he dropped out of Oxford University to help set up a telephone Counseling Service building it to become one of the largest mental health support services of its kind in the UK he then worked as a policy officer for the then mayor of London Ken Livingston Mustafa went on to help start Rios Partners a boutique consultancy with eight offices across four continents specializing in designing and facilitating large scale multi-stakeholder change Labs aimed at navigating complex problems as a skill negotiator and facilitator Mustafa has worked all over the world for a wide range of clients such as the UN the Dutch government and WWF may I please introduce Lord d uh Mr President uh David thank you very much for that warm introduction I have the challenge now to make the case or at least show you that the school that I went to wasn't as good as 1914 uh but we did go to the same school and it's amazing coincidence to see that and uh and subsequently you probably will also see uh when we share the platform for Mustafa that I haven't yet cracked the artificial intelligence side so I'm going to use whatever I have up there so the whole purpose of today is to share with you Innovations in patient safety patient safety is something has been very close to my heart and I know many of my clinical colleagues here will know not just something about it but it's the type of thing that always stresses any clinician involved in patient care let alone the patients that we have the privilege of treating I'm also very pleased that I'm co-chairing uh co-presenting this with Mustafa who I've got to know over the last two or three months uh an amazing person uh in ter terms of his intellect and as also an interest in what we're doing so before I start I think it's worthwhile remembering the late uh surname who you've heard a lot about him from the president but also acknowledge the tremendous contributions he's made in philanthropy and research and Life Sciences uh we're all very grateful for him and his family for their generosity in supporting research and development so moving into the arena of patient safety most of us will know who Florence Nightingale is and the contribution she made and she survives in our imagination as the most inspired nurse this country produced but the lady with the lamp some of you may know this was also a pioneering and a passionate statistician following the international statistic Congress which was in London in 1860 and I wouldn't say that far away from this building the London hospitals decided for the first time ever to publish the inpatient mortality rates over a period of five years and this publication by William guy from King's College looked at the average as you can see on the right hand side over a period of five years I think David slowman is not here but his chair is here Dominic he'll be delighted to see that nearly a century ago or more the all free was the best reforming hospital and I have no doubt it is still the same now within that a century or within that Century later we've all seen the tremendous contributions that s science and technology and innovation has made to our life expectancy now who would have predicted in the last 100 years that life expectancy would be doubled you tell me any other discipline or any other sector that has such an impact on Mankind in the first part of that Century most of these interventions were Public Health interventions in the latter part of that Century some of the technological innovations some of which are up there MRI CT scans and what's unique about that last century 40% of the major disruptive Technologies came from the UK from Great Britain that's an amazing Legacy to have so the investment the dangor family have made in the Life Sciences industry will show its fruits despite the statistical Publications that we've seen earlier despite the life expectancy being doubled we are still challenged those of us working in health care those of us in this room who have received care from our health care System have a major challenge facing us and that is patient safety most of you will remember when all of that started in bris Bristol and the cardic inquiry there and more recent recently the mid staff inquiry and you don't have to go far will you meet what it was three or four weeks ago and his tragic death from a misdiagnosis of sepsis if you look at the scale of the harm it's about 10% one in 10 patients who go to a hospital to get better they leave that hospital with an injury one of three of these injuries are serious enough and more than 50% of that harm is preventable if you look at the mortality rates the predictions now this is a publication from uh alen Hogan from the London School of hygiene and tropical medicine in 2012 which was updated recently with the Bruce Kio review more than 10,000 deaths in hospital settings in the United Kingdom are preventable and the most common causes are poor monit ing diagnostic errors and medication errors the famous saying error is human and the we've been looking at this we we are very very fortunate to be funded by the uh National Institute of Health research and I happen to have the privilege of leading the patient safety translation Research Center and we've been looking at health healthare and why is it become so risky despite all the technological advances that I've showed you earlier firstly Healthcare delivery is become a team sport we all know that we all respect that but coordination of it members of the same team let alone coordinating number of different teams around the needs of Fairly complex patients with multiple morbidities is become a challenge at the same time multiple complex interventions which could be on different site with centralization and decentralization drivers in the health system and the fragmentation of healthcare one of the insightful things that patients told me when I was had the privilege of leading a major review of the NHS is that if you see the pathway of care from a patient perspective is very fragmented we create these boundaries because we can cope with it my boundary is the surgical patient who's under my care the day they arrive and the day they leave but the patient sees the whole pathway of care so the fragmentation all of these different factors where technology has enhanced our life but in many ways we've created a system that is too complex and too dangerous in many ways in today's talk I'm going to focus on two very important core safety challenges that I think are facing us today and in the last two or three years it is becoming more and more acute firstly how do we detect a patient is deteriorating in the hospital setting this patient and their family thinks that they are in a safety zone They are in a hospital but the challenge we've had is how do we actually monitor the patients and to ensure that we pick up the their deterioration and as a result of that intervene at an earlier stage than then when it's too late the data out there is very clear including nice and I could see the previous chair of nice here the National Institute of Health and clinical Excellence have recognized that substantial problems are there in identifying a deteriorating patient and you may say why that happens to be the case we actually have a huge data that we measure uh in a hospital setting I don't know if the patients or members of the public but certainly most of the clinicians here are fully aware on the volume of data we collect and this is 2016 we still have them on these flimsy pieces of paper half the time it's not even beside the bed measuring all sorts of things physiological data biochemical data including stool charts urinary output so on and so forth the problem with this type of technology is a subtle change in in any way is rarely recognized until such time as there is an acute drop in the blood pressure or a real uh increase in the pulse that will May alert the nurse who might be available to detect that and once that detection happens obviously she or he has to inform someone who knows how to deal and how to manage that case the second issue in the pathway of care of someone who's deteriorated is to intervene start firstly by getting the right person to see the right patient in the right time you might think that is very easy but in actual fact in about half the patients that escalation of care if you go back retrospectively and look at that information is delayed and this was the publication of a member of our team PhD student with us Max Johnson who looked at a syst atic review I Mayan I just say this is not a unique problem in the United Kingdom it is as you can see the US Australia and Japan about a 20 to 50% delays in identifying these patients and the Box on the right hand side show shows you the correlation between the delay in picking that up and its impact on morbidity and mortality now let me tell you how we communicate when someone is deteriorating what's the first line of communication to someone to come at least deal with that emergency in a ward environment if you go up to and this picture was taken from uh one of our Wards yesterday because we wanted a picture to show you what the technology we're currently use the bleep system is about 50 years old I'm sure there some of you here will swear by these but I'm just about to show you something better uh and uh that is what the house officer or the senior officer will wear and at the same time on the right hand side one of the commonest Handover notes that is given between two doctors leaving a shift and handing over the notes to the person taking on the shift and what's interesting something else we really as we studied this on average a junior doctor on a shift of eight hours will receive about 100 different calls or jobs in samar's hospital the acute medical service just the acute medical service forget about surgery and all the other disciplines they receive about 1,000 jobs and tasks a day you go and ask anyone what are these tasks no one has a record of them you get a bleep this patient needs a uh some Bloods done this patient needs a catheter done this patient needs that and that is the type of messaging that the the individual receives and they haven't finished that job they write this sheet on the right hand side to hand over the patient so it's a fairly cumbersome thing the nurse goes up to the phone after she detects a deteriorating patient calls the operator the operator will page the doctor and this poor doctor is receiving a significant number of messages on his pager and he or she has to make a decision where to go and without any information in terms of prior priority I izing the patient that they need to see next so the main core and the hypothesis of our work has been how do we have a better detection systems and how do we have better system to intervene the first one is early warning signs and how we can pick those uh signs of deterioration and the only way we could do this is what we are currently using in all aspects of our life whether it's shopping Banking and that is the digital technologies that are available to us and also a better use of data I famously said in 2008 you can only improve things if you measure them and that's the reason I remember working with many colleagues in in this room I remember giving a talk in which we talked about really having quality accounts and measuring quality because if you can measure it you can actually improve it now other sectors are very smart they've done this before just look at the number of serious and fatal accidents the mortality from road traffic accidents over a period of 30 years now how come they could achieve this and all the Investments we have in the Life Sciences all the technological advances that I showed you earlier could not be translated in improving patient safety in a hospital environment and how did they do it in reducing road traffic accident a fairly ranging Technologies over 30 years from seat belts to break pads and finally the last technology I don't have a very new car but I've seen and I've driven one which is the what they call the assist uh uh assist driving technology when actual fact it alarms you as a driver it tickles you I think when it drives in the seat that you're about to hit the curb or you're about to hit the car beside you I love to have one of these which might actually push me to recognize someone who's deteriorating so most of our effort has been how do you actually translate technology in enhancing patient safety information technology is the way to do it but the current systems we have are cumbersome they've taken years to install and I don't want to go to the fiascos of the cost of the information technology systems that were installed installed within the NHS I'm going to share with you a couple of slides the end user view ultimately I know this because I'm a doctor doctors only use technology if it's going to make their life easy look at some of the uh some of the enduser uh uh views that we've captured it took one hour to teach a doctor how to prescribe paracetamol through the it system I don't think I'm stupid but it could be more complicated and so on and so forth in actual fact you don't have to ask the doctors and the nurses who using these systems ask the regulator CQC I'm not just some of the reports from their inspections data quality issues have also been highlighted of total of 460 records that could have been completed 245 were valid 280 were invalid and 155 were missing so despite the investment if you don't have an intuitive system that the doctor or the nurse feels that is actually it's helping them to deliver a better care to their patients they're not going to engage with these tools and the end of the day they become an administrative tools in which the Finance director could calculate how many patients came in the door how long did they stay there rather than really looking at the huge potential of information technology which has transformed everything else in our life but not necessarily in healthcare so we've been looking at how could we engage clinicians at the front line identifying the gaps and using more simplified Technologies in developing these Technologies and and this whole idea started with some collaboration uh as you know I come from Imperial uh which is probably the best endowed institution when it comes to the power of engineering science software writing and beside Imperial there's another wonderful organization called The Royal College of Arts I don't know how many of you have visited that but design was something else that I recognized as I matured will have a quite an impact on delivery of healthcare so this is the Helix Center which is a joint venture between Imperial College and the Royal College of Arts so this is the pod which was this was funded by hefy it's actually in the middle of the hospital and the reason we put it in this wonderful peeron between the medical school and the old building uh is because that was the best way of really getting that staff coming into there and really coming up with their clinical problems and at the same time patients who came in there really sharing some of the challenges in terms of the wars they were in and we have a number of designers there you can see Maya who is in there and uh number of others there who essentially designers Engineers software writers who really try to identify these different gaps and develop the different apps as you can see here which is uh which having an impact on a number of different Pathways of care but the one that has been very unique which has really been championed by one of our lecturer Dominic King who's there somewhere there he is uh is a lecturer in surgery back to some of the challenges that I was telling you earlier how could we capture this information better how could we develop an early warning uh systems in which a doctor could prioritize some of the task that they've been given and then intervene and start the treatment so hark is essentially bringing all of these different inputs blood test physiological monitoring and combine those together and really deliver it then to the individual clinician looking after a patient in a ward environment or even in a primary care environment so I'm just going to go through some of the description of it this is a task being ordered it's very simple task you can order it within 30 seconds and I think what's unique about this the person who's ordering that task will actually identify the person of the right seniority in ordering that task the task is created and then the task is received by the individual who needs to ask on to who needs to respond and act on that command the prioritization are significantly easier obviously with certain algorithms you can see here the actual doctor will have a priority of different jobs so at least you could go and see the patient who's sickest in the short shortest period of time rather than having a list in front of them and trying to make a judgment call without real the necessary information so if this is some of the work which was the output of evaluating this at Imperial College NHS trust mainly on the samar's site you can see from a functional perspective there were 11 supplementary communication functions with the device information transfer 22 out of the 24 CL critical factors were better communicated using the app and digital app and finally the response time as you can see had significantly improved by by 30 37% less time taken to respond to a task request so we've developed this you know it's amazing because you go around you think NHS is a basket case but the NHS is not you can actually create Innovations in the NHS NHS should be leading Innovations we've reached a stage after the develop that Dominic and a number of other colleagues in the department recognized that really to scale this up and trial it even within the large Imperial College Trust on the three sites we need some help and we were delighted to meet Mustafa and his colleagues and deep mind you know some of the I I was there in the office yesterday some of the smartest people I've come across and the idea of developing a partnership which has already worked extremely well at roal free as you may know and really taking this up and scaling this up and I know the NHS trust leadership are very excited about this and they've been absolutely brilliant in facilitating this Innovation coming through so I think I want to share with you an example a very simple what I call a Frugal Innovation that could have a huge impact in improving communication identifying sick patients preventing deterioration and escalating care and thanks to the group which is the uh this is the launch of our patient safety and translation Research Center nearly three years ago in the House of Lords and that Dame Sally Davis there who is the director of uh nihr and I'd like to thank finally obviously Imperial College for facilitating this the trust who really with open arms have embraced this Innovation andr but also Imperial Innovation so on that note I'm going to now hand over to the smart guy who's going to tell you how we just take this over and really have an impact on patients and their safety thank you thank you very much so thank you um ARA for a wonderful talk and for framing I think the patient safety challenge that we have taken on so my name is masafa I'm one of the co-founders of Google Deep Mind before I talk to you about the patient safety challenges I want to start with a little bit of a background on Deep Mind who we are and what's really motivating us I'd like to put it to you today that one of the defining characteristics of the Modern Age I believe is that we are overwhelmed by the complexity of the of the systems that we've created all around us we have the smartest experts people who trained for decades and decades struggling to respond to uh some of our most complex problems whether they be climate change or building sustainable Food Systems of fair production or whether it's taming our complex financial markets in some sense what we're struggling to cope with is to make sense of these vast streams of data these high dimensional temporal abstract streams of data that are coming in from all of the senses that we've created around us that somehow describe the nature and structure of the world today 5 and a half years ago I found a deep mind with two of my colleagues and in some sense it was a response to exactly this problem we recognized that we were overwhelmed by complexity and overwhelmed by information and we posited that over the coming decades the real benefits and Innovations and productivity gains that will come in our civilization will be those that we derive from better understanding the vast streams of data that we're creating around us if we could somehow learn the nature and structure of this data well enough to be able to predict its course over time then that would better help us to intervene in this complex world around us and at its heart Deep Mind is very much a British research organization so we have something like 250 staff 180 of the very best AI researchers postdocs phds and professors in the world over the last 12 months we've been lucky enough to score two nature front covers first uh this time last year with the nature front cover for our Atari dqn work where we trained an algorithm to learn to play multiple Atari games old school uh Atari games from the 80s purely from raw pixel inputs just in the last couple of weeks we've announced that we we've become the first ever system Alpha go to beat a human professional um at the game of Go in the next couple of weeks we're actually going to be playing the world champion in Korea for a million dollars in a 5-day Live Match and what we've tried to create at Deep Mind is a kind of hybrid organization we've tried to bring the very best that a focus on long-term research has to bear and learn all of the best bits from Academia and bring those Into the Heart of our organization while simultaneously combining those things with the pace and the scale and the agility and dynamism that we get from working as a company at scale and all the while thinking what can we do that would be meaningful what really is our purpose what are we really focused on and that's where Deep Mind has social impact at the very heart of our core mission in some sense I think this makes us uniquely placed to work within the NHS we have the resources and Technical expertise of one of the best technology companies in the world and yet we also have the agility of a startup structured as we are in these small cells that have a great deal of autonomy and can work really really fast but all the while driven by the kind of meaning and purpose of a social impact organization so let me start with what I think is a very brief diagnosis on why healthtech has underd delivered over the last couple of decades in some sense it's sort of struggle to do what nurses and doctors want what they know that they need one of those reasons I think is because it's consistently delivered I guess what you could call a kind of one siiz fits-all or top down system rather than something which has been co-designed in collaboration with clinicians directly typically the characteristic of these deployments is that they've been very large scale step change introductions where there's really only one version you buy a product off the shelf you deploy it to a thousand person organization and it and it updates and improves very slowly and this isn't the the characteristic of the sorts of apps that you'll be using on your phone every day and that you're also familiar with the other thing that I think has has really slowed thing down things down is that typically the companies have stuck with proprietary standards which tend to create closed systems and that means that they've defined their own database schemas they've defined and hand curated data in a particular way which is the reason why a lot of your technology systems don't talk to one another and I think that's a really big mistake that prevents both academic research CL clinical epidemiology s and also startups or doctor hackers who have a great idea from getting access to that data and being able to experiment and deploy their own prototype app we've already covered some of the common complaints but I'm sure you'll be all familiar with the questions of why on Earth are we using pages in 2016 where else in the world do we have pages in production in such an important environment there aren't enough desktop computers on a ward I've heard this over and over again in the last nine months and it sort of amazes me when you do find one the batteries probably died or the smart card reader doesn't work or there's a key missing in the keyboard and there some really fundamental infrastructure that doesn't seem to be quite right quite remarkably I've actually experienced faxes in operation I I genuinely didn't know that these things were still in use until and and until until I started collaborating with my friends at the Royal free and and at Imperial I've also heard the complaint that we wait for hours sometimes even days for x-rays to be transferred between hospitals clearly this is something that should be digitized now look this is obviously an incredibly intimidating arena for technologists in some sense the history of the last two or three decades represents a kind of daunting Prospect this hasn't been a particularly successful place for Technologies to work but I want to argue that if the NHS needs new technology as ARA has so rightly put then the onus is really on us um technology needs to find a new way to work with the NHS let me try and present you a few of those ideas today this is actually incredibly simple ask the question what does the user need and when you've asked that question ask it again and again and again and let that be the Guiding Light shaping everything that you do as a technologist doctors and nurses already know what the problems are they work on the front line day and day out they're very smart people they're very creative and often they even know what the solutions are what I've heard is that doctors and nurses have been demanding one mobile first why can't I have data in my hand at the right time I need the right piece of information right in my pocket so that I can make the appropriate diagnosis or take the appropriate intervention this needs to be sharable across teams why can't I communicate this information in my own views on on on the current state of the condition of my patient to my colleagues and this should be super easy to learn to use no one's ever had to go on a training course to learn to use an app that you've downloaded from the App Store and the truth is that these Innovative workarounds are already in use you know it's sort of a taboo topic but the vast majority of doctor doctors and nurses today are using WhatsApp to communicate patient identifiable data to share images and they do their best to keep it uh anonymized but clearly there are risks to the current system they don't comply with the most upto-date NHS standards and that's something that we've got to tackle but frankly we really can't blame clinicians they're motivated to really make a difference and they've committed their careers to helping patients to have the best possible care why wouldn't they take advantage of what they obviously see to be a tool that makes their lives easier and I think the challenge for us is how can we deploy some of the best user designers some of the best Engineers some of the best machine learning researchers to work on some of the problems that really matter so let's tackle the first patient safety challenge that ARA put to us how do we identify which patients are at risk so what's actually going wrong here why aren't we why aren't we detecting them on time well part of the problem is that we're not recognizing their physiological deterioration and that creates a breakdown in reporting and then we don't get access to the data in real time largely because we're dependent on these sorts of hand uh amended charts they're really difficult to share they're very difficult to interrogate you only get information from them when you go to seek it that's an important Dynamic you only get information from them when you go to see it it's exactly the opposite of the way that your smartphone works your smartphone tells you when there's an important piece of information that you might need to pay attention to so let's take the example of acute kidney injury that we've been collaborating with the RO fre on contributes to about 40,000 deaths annually and quite remarkably it's been estimated by NHS England that around 25 % of these cases could be preventable and this cost something on the order of a billion or so pounds in direct NHS spending it's a national priority and and in the last couple years the NHS have issued a patient safety alert to try to frame um how uh clinicians respond to this but essentially this has been a PDF with The Cutting Edge best practice emailed to everybody and stuck up on posters and I think that that can be improved so I'm going to tell you about a collaboration that we kicked off with Dr Chris Lang um and members of his team at the Royal free over the last six or seven months or so Chris is a consultant nephologist and kidney expert and he's also the associate um medical director responsible for patient safety at the free so we set about co-designing directly hand inhand with clinicians at the free the first step is how can we understand the user who is the user what do they do how do they work on a day-to-day basis we did scores of focus groups and onetoone interviews we try to empathize with their day-to-day experience by shadowing them and trying to learn from what happens minute by- minute hour by hour during their day and it was great fun for us we use that to try and identify what their unmet needs are today and we use those needs to then prioritize our own requirements as you see here in some of the images immediately as soon as we've got a sense of what their needs are we start to build things we we prototype we experiment we sketch things we do wireframe and then immediately we want to test them the answer is rinse and repeat this process actually works we go over it every 48 Hours as soon as we've heard that there's a new insight that somebody's had we go and put it to wireframe we sketch something out and then we take it back to the nurses and doctors and we say how does this look does this make sense how would you like to change the buttons the color the choice architecture the menu options does this flow make sense we learn we iterate we test and we constantly improve the other component to our codesign has been to try and actually map out the complexity of this pathway so it's been estimated there's something like 900 to a thousand different patient experienced Pathways in the hospital system so what does it look like for us to actually try and map this complexity out well from the patients perspective it's it's pretty painful lots of different things are happening to this patient throughout the process but if we zoom in a little bit we can see that this red part pathway here is where the vast majority of the activity takes place and in some sense this is when the patient has already developed some kind of life-threatening complication of Aki so what we decided is how can we go and look backwards what happens before that part of the process how can we do better risk assessment by looking at the underlying comorbidities how can we do a better job of real-time monitoring of the data in order to try and proactively intervene in this patient's care in order to drive their experience down a parallel pathway that is more stable so obviously the best way to do this is to build something and test it here is the um app that we co-designed with our colleagues and friends at the RF free I'm going to run you through what it might look like so we log in immediately you see uh the patient list and so this could be sorted by Ward or by specialty or by consultant and so if we go ahead and just tap on um the first patient we see that this patient actually has a pretty serious acute kidney injury level three we're able to browse through the results of the retrospective Trend analysis and we can see that there's been about a tripling against Baseline of this person's creatin they're in a pretty bad way and we also want to take a look at the um potassium we see that it's very highly raised at at seven so this could indicate hyperemia we may want to zoom in and take a look at the CRP which is also extremely raised and compare the trend analysis against our creatin zooming out again we see that lactate is at a 5.1 and this might give us an indication um that this patient's cause of AI has been um a sepsis if we want to see in a little bit more detail we can see the retrospective blood test results exactly who ordered it when what was the time and what was the value in more detail we can also zoom in and see a little bit more about the patient details you can see here that this patient has come in with uh type 2 diabetes and hypertension and also if we look down a little bit further we see that they originally presented with abdominal pain and were subsequently diagnosed with acute appendicitis we also have the contact details at hand so that we can contact both the NEX ofkin and the general practitioner so let me show you a little video now of I think clinical teams like to be able to Define problems and team up with people who can help solve them it's been very interesting to have this collaboration between Hospital in a company like deep mind we're very interested in the idea of streaming at speed clinical data to mobile platforms because most of our teams spend a lot of their day on the move because of doing their clinical jobs in my work I'm quite involved in implementation science which is a way of testing clinical changes in practice working with deep mind health has followed that model completely that way of testing iteratively testing so planning something doing it studying whether we thought it was good or not is extremely valuable what's nice about helping to design something is that we have a say in how it will be useful what I've seen about the design of the app is that the screen is clean and it makes sense and there isn't too much information on it it's not all in your face technology should support your work should support what's going on with that patient so if you can communicate that easily to people it's going to save time and it's going to help you do your job and it's going to help you get to your patients quicker and so getting to your patients quicker is of course the part of detection which is only half the story to respond to ARA's second patient safety challenge the real question is how do we actually manage the intervention on once we've identified which patients we need to go and see at the bedside and this is where I think the delivery of effective care has been delayed according to ARA and Dominic and their research and up to 50% of deteriorating patients and their response was really quite incredible they went off and created hark to support nurses and doctors to deliver care much more effectively clinicians responded 37% faster with hark than with pages and so tonight I'm really proud to be able to announce that Hawk's going to be joining forces with deep mind health and becoming an integral part of the service that we go on to provide let me take you through an example of what it might look like to manage tasks with Hawk so let's take the the case of requesting a medication order the patient clearly needs a prescription the nurse goes on to bleep a doctor the doctor hopefully calls back in good time the doctor then will go to the ward and have a face-to-face conversation with the nurse decides that yes it's right that we need a prescription and then go on and inform the nurse that the request has actually been completed and then finally the patient uh is able to receive the medication of course this assumes that you get through on the first bleep and that the poor nurse isn't just bleeping constantly because the doctor that that he or she is trying to get hold of is actually on half term this week or is busy in theater all day this is what should happen really doctors should be able to prioritize and manage their tasks on their phone they should be able to accept a task from a nurse and tick to say they're handling it they should be able to have a conversation with a nurse or a fellow doctor in messaging in their app about that specific task and then they should be able to send a confirmation message when it's been done completely remotely saving a vast amount of time so this evening I'm really pleased to be able to announce that we're going to be launching streams by deepmind health and our objective is for the first time ever I think to integrate both the detection and task management into a single platform the plan is for it to be a beautiful and intuitive and most importantly really easy tolearn experience that anybody can download in the in the clinical setting and start using streams will deliver the right data to the right clinician at exactly the right time and really the objective here is for us to try and shift some of the 97% or so of activity in the hospital today which is reactive further towards activity which is proactive and ultimately preventative and this of course is where our Cutting Edge analytics and machine learning comes in how do you prioritize the series of alerts that go to a patient how do you ident go to a doctor or a nurse how do you identify which person on the clinical team should be receiving the right task and how do you ensure that they've been followed up in good time and of course the real question here is ultimately how will we be able to provide patients with access to this data how we be able to empower patients to play an active role in their own treatment and in their own experience and really play an accountable a role in holding accountable the treatment that they receive whilst they're in hospital so I really want to leave you with the idea that this time I think it's really important that we have to do things in a fundamentally different way and for us this is about partnering and co-designing with clinicians clinician Le technology is actually a new way of thinking about how we co-design interventions in the system I truly believe that together we can revolutionize patient safety and this evening I'd like to invite anybody and everybody who's interested to ping me an email personally and come and hang out in our office and come and talk to us about the apps that you have in your mind the things that don't work for you and your visions of how this might be able to work come and join our partnership thank youwell good evening ladies and gentlemen and it's my privilege and pleasure to welcome you here to the Royal Society of medicine for the second name so name d lecture how digital Innovation can improve health care I want to start by expressing a particular welcome to the Dango Family David Michael Robert and Ellie and their partners and all of their family who are here tonight I also want to specifically welcome some other people kice Ellison who's been such a uh a benefactor to the RSM over many years together with Brian Whitt who's been a long-standing supporter here I'd like to welcome staff and students from the Westminster Academy an academy that's supported by the Dango foundation and which uh I just heard from David Dango earlier has achieved 19th Place nationally in the sixth form rankings which is a fantastic achievement uh in the last few years and we're also pleased to welcome representation from patients Association for this evening's lecture in a minute uh David danger Will introduce our two guest speakers Lord Dary of denim and Mr Mustafa suan so I will not say any more about them except to point out that Lord Dary has I think the singular distinction of having resuscitated some one of his fellow peers in the House of Lords so if any of you not feeling so well uh here's the here's the man to to to to help out you can say it's a heart surgeon telling you that as well anyway this uh lecture this evening's lecture is named in honor of s name Dango who sadly passed away last year at the age of 101 he was born in Baghdad in 1914 when Iraq was under ottoman Rule and he subsequently traveled to London to study engineering and on his return to Iraq went into business he built up a diverse portfolio of activities including property development and letting a match and Furniture Factory and he won the first contract to bottle Coca-Cola in Iraq which is also particularly Advocate as the headquarters of Coca-Cola right next door to us at the RSM in 1947 he married Renee Dango and they had four children following the birth of Israel and Arab nationalism things became increasingly difficult in Iraq and the Dango Nam Dango moved his family uh to Britain and for a few years he traveled between Britain and Iraq but uh subsequently events became very difficult in Iraq and he he permanently relocated to Britain settling in London he vowed that if he made another fortune in the wonderful country that had taken him in he would plow his gains back into philanthropic causes so he set up a commercial property business which his sons joined him in and and it soon began to thrive and with that success he was soon able to make good on his promises he started to give back to many organizations and Charities focusing on education and health in 2014 he made the largest philanthropic gift to the Royal Society of medicine and some of you will have seen the lecture theaters uh that is named after him as you came into the entrance hall this evening but he's also he also donated to the Francis Crick Institute and cancer research UK he funded over a thousand scholarships for students with no family history of tertiary education and gave £4 million for students from low-income back sorry from low-income backgrounds to study STM subjects so he had a massive philanthropic uh donation to many many people in many many different walks of life and these as I've indicated earlier include the Westminster Academy in London whose pupils hail from over 60 countries and as I said we're particularly welcome to to many of them this evening so name was made OB in 2006 CB in 2012 and in June 2015 became the second oldest person to receive a Knighthood he had seven grandchildren and two great-grandchildren who survive him in addition to his four Sons the theme of tonight's talk is medical Innovations and many of you will know that the RSM has a flourishing program in medical Innovations which has now extended to over 180 presentations many of our speakers have traveled from long distances to be here to speak about their ideas which will benefit patients others have traveled from not such large distances including Hammer Smith and King's cross in April this year we'll be hosting our 13th sum our 12th Summit with 13 speakers a large faculty from around the world but interestingly including a 15-year-old from Britain who won an innovation prize for Research into the early diagnosis of Alzheimer's so here at the RSM we're delighted to be able to encourage youthful Enterprise and energy and Innovation over 250 people have already registered to attend the April Summit and if any of you still uh in the audience who haven't registered wish to attend I would suggest you register fairly soon uh otherwise you may have difficulty in getting in so I'm going to invite David dangle to introduce our speakers Lord Dai will be speaking first followed by Mustafa selan there will be an opportunity for questions uh and discussions but I would ask you when we come to question time to please keep your questions uh brief and to the point otherwise I shall have to shut you up the closing remarks and vote of thanks will be given by one of our RSM members and also Junior doctor Dr KY MCG gretton so welcome to the RSM and David can I invite you to come up and say a few words good evening Professor Lord Dary of denim holds the Paul Hamlin chair of surgery at Imperial College London the Royal Marsen hospital and The Institute of cancer research he is director of The Institute of global Health Innovation at Imperial College and chair of Imperial College Health Partners he is also an honory consultant surgeon at Imperial College Hospital NHS Trust research by Professor Dai is directed towards achieving best surgical practice through innovation in surgery and enhancing patient safety and the quality of healthcare his contribution Within These research Fields has been outstanding publishing over 800 peer-reviewed research papers to date in recognition of his achievements in the research and development of surgical Technologies Professor Dari has been elected as an honory fellow of the Royal Academy of engineering a fellow of the Academy of Medical sciences and in 2013 was elected as a fellow of the Royal Society he was kned for his services in medicine and surgery in 2002 in 2007 he was elevated to the United Kingdom's House of Lords and Appo appointed parliamentary under Secretary of State at the Department of Health upon relinquishing this role within central government in 2009 professor sat as the United Kingdom's Global Ambassador for health and Life Sciences until March 2013 during his appointment and Beyond Professor Dai has developed his status as a leading voice in the field of global Health policy and Innovation Professor Dai was appointed and remains a member of Her Majesty's privy Council since June 2009 in this year's New Year's honors it was announced that he would be admitted to the order of Merit this award is the personal is in the personal gift of the queen and is limited to 24 living recipients like my late father s Naim dangu Lord Dari was born in Baghdad and he confined in me just now that he went to the same school as my father and then moved to the school that I went to in Baghdad and he has given much to this country so I am particularly pleased the Ed that he has agreed to give this year's Sim Deno lecture Mr Mustafa suan who will follow Lord Dari is co-found founder and chief product officer of Deep Mind Technologies a leading artificial intelligence company backed by the founders fund Lee kashing Elon Musk and David bundman among others the company was bought by Google in 2014 in their largest EUR European acquisition to date he is now head of Applied AI at Google Deep Mind responsible for integrating the company technology across a wide range of Google products at 19 he dropped out of Oxford University to help set up a telephone Counseling Service building it to become one of the largest mental health support services of its kind in the UK he then worked as a policy officer for the then mayor of London Ken Livingston Mustafa went on to help start Rios Partners a boutique consultancy with eight offices across four continents specializing in designing and facilitating large scale multi-stakeholder change Labs aimed at navigating complex problems as a skill negotiator and facilitator Mustafa has worked all over the world for a wide range of clients such as the UN the Dutch government and WWF may I please introduce Lord d uh Mr President uh David thank you very much for that warm introduction I have the challenge now to make the case or at least show you that the school that I went to wasn't as good as 1914 uh but we did go to the same school and it's amazing coincidence to see that and uh and subsequently you probably will also see uh when we share the platform for Mustafa that I haven't yet cracked the artificial intelligence side so I'm going to use whatever I have up there so the whole purpose of today is to share with you Innovations in patient safety patient safety is something has been very close to my heart and I know many of my clinical colleagues here will know not just something about it but it's the type of thing that always stresses any clinician involved in patient care let alone the patients that we have the privilege of treating I'm also very pleased that I'm co-chairing uh co-presenting this with Mustafa who I've got to know over the last two or three months uh an amazing person uh in ter terms of his intellect and as also an interest in what we're doing so before I start I think it's worthwhile remembering the late uh surname who you've heard a lot about him from the president but also acknowledge the tremendous contributions he's made in philanthropy and research and Life Sciences uh we're all very grateful for him and his family for their generosity in supporting research and development so moving into the arena of patient safety most of us will know who Florence Nightingale is and the contribution she made and she survives in our imagination as the most inspired nurse this country produced but the lady with the lamp some of you may know this was also a pioneering and a passionate statistician following the international statistic Congress which was in London in 1860 and I wouldn't say that far away from this building the London hospitals decided for the first time ever to publish the inpatient mortality rates over a period of five years and this publication by William guy from King's College looked at the average as you can see on the right hand side over a period of five years I think David slowman is not here but his chair is here Dominic he'll be delighted to see that nearly a century ago or more the all free was the best reforming hospital and I have no doubt it is still the same now within that a century or within that Century later we've all seen the tremendous contributions that s science and technology and innovation has made to our life expectancy now who would have predicted in the last 100 years that life expectancy would be doubled you tell me any other discipline or any other sector that has such an impact on Mankind in the first part of that Century most of these interventions were Public Health interventions in the latter part of that Century some of the technological innovations some of which are up there MRI CT scans and what's unique about that last century 40% of the major disruptive Technologies came from the UK from Great Britain that's an amazing Legacy to have so the investment the dangor family have made in the Life Sciences industry will show its fruits despite the statistical Publications that we've seen earlier despite the life expectancy being doubled we are still challenged those of us working in health care those of us in this room who have received care from our health care System have a major challenge facing us and that is patient safety most of you will remember when all of that started in bris Bristol and the cardic inquiry there and more recent recently the mid staff inquiry and you don't have to go far will you meet what it was three or four weeks ago and his tragic death from a misdiagnosis of sepsis if you look at the scale of the harm it's about 10% one in 10 patients who go to a hospital to get better they leave that hospital with an injury one of three of these injuries are serious enough and more than 50% of that harm is preventable if you look at the mortality rates the predictions now this is a publication from uh alen Hogan from the London School of hygiene and tropical medicine in 2012 which was updated recently with the Bruce Kio review more than 10,000 deaths in hospital settings in the United Kingdom are preventable and the most common causes are poor monit ing diagnostic errors and medication errors the famous saying error is human and the we've been looking at this we we are very very fortunate to be funded by the uh National Institute of Health research and I happen to have the privilege of leading the patient safety translation Research Center and we've been looking at health healthare and why is it become so risky despite all the technological advances that I've showed you earlier firstly Healthcare delivery is become a team sport we all know that we all respect that but coordination of it members of the same team let alone coordinating number of different teams around the needs of Fairly complex patients with multiple morbidities is become a challenge at the same time multiple complex interventions which could be on different site with centralization and decentralization drivers in the health system and the fragmentation of healthcare one of the insightful things that patients told me when I was had the privilege of leading a major review of the NHS is that if you see the pathway of care from a patient perspective is very fragmented we create these boundaries because we can cope with it my boundary is the surgical patient who's under my care the day they arrive and the day they leave but the patient sees the whole pathway of care so the fragmentation all of these different factors where technology has enhanced our life but in many ways we've created a system that is too complex and too dangerous in many ways in today's talk I'm going to focus on two very important core safety challenges that I think are facing us today and in the last two or three years it is becoming more and more acute firstly how do we detect a patient is deteriorating in the hospital setting this patient and their family thinks that they are in a safety zone They are in a hospital but the challenge we've had is how do we actually monitor the patients and to ensure that we pick up the their deterioration and as a result of that intervene at an earlier stage than then when it's too late the data out there is very clear including nice and I could see the previous chair of nice here the National Institute of Health and clinical Excellence have recognized that substantial problems are there in identifying a deteriorating patient and you may say why that happens to be the case we actually have a huge data that we measure uh in a hospital setting I don't know if the patients or members of the public but certainly most of the clinicians here are fully aware on the volume of data we collect and this is 2016 we still have them on these flimsy pieces of paper half the time it's not even beside the bed measuring all sorts of things physiological data biochemical data including stool charts urinary output so on and so forth the problem with this type of technology is a subtle change in in any way is rarely recognized until such time as there is an acute drop in the blood pressure or a real uh increase in the pulse that will May alert the nurse who might be available to detect that and once that detection happens obviously she or he has to inform someone who knows how to deal and how to manage that case the second issue in the pathway of care of someone who's deteriorated is to intervene start firstly by getting the right person to see the right patient in the right time you might think that is very easy but in actual fact in about half the patients that escalation of care if you go back retrospectively and look at that information is delayed and this was the publication of a member of our team PhD student with us Max Johnson who looked at a syst atic review I Mayan I just say this is not a unique problem in the United Kingdom it is as you can see the US Australia and Japan about a 20 to 50% delays in identifying these patients and the Box on the right hand side show shows you the correlation between the delay in picking that up and its impact on morbidity and mortality now let me tell you how we communicate when someone is deteriorating what's the first line of communication to someone to come at least deal with that emergency in a ward environment if you go up to and this picture was taken from uh one of our Wards yesterday because we wanted a picture to show you what the technology we're currently use the bleep system is about 50 years old I'm sure there some of you here will swear by these but I'm just about to show you something better uh and uh that is what the house officer or the senior officer will wear and at the same time on the right hand side one of the commonest Handover notes that is given between two doctors leaving a shift and handing over the notes to the person taking on the shift and what's interesting something else we really as we studied this on average a junior doctor on a shift of eight hours will receive about 100 different calls or jobs in samar's hospital the acute medical service just the acute medical service forget about surgery and all the other disciplines they receive about 1,000 jobs and tasks a day you go and ask anyone what are these tasks no one has a record of them you get a bleep this patient needs a uh some Bloods done this patient needs a catheter done this patient needs that and that is the type of messaging that the the individual receives and they haven't finished that job they write this sheet on the right hand side to hand over the patient so it's a fairly cumbersome thing the nurse goes up to the phone after she detects a deteriorating patient calls the operator the operator will page the doctor and this poor doctor is receiving a significant number of messages on his pager and he or she has to make a decision where to go and without any information in terms of prior priority I izing the patient that they need to see next so the main core and the hypothesis of our work has been how do we have a better detection systems and how do we have better system to intervene the first one is early warning signs and how we can pick those uh signs of deterioration and the only way we could do this is what we are currently using in all aspects of our life whether it's shopping Banking and that is the digital technologies that are available to us and also a better use of data I famously said in 2008 you can only improve things if you measure them and that's the reason I remember working with many colleagues in in this room I remember giving a talk in which we talked about really having quality accounts and measuring quality because if you can measure it you can actually improve it now other sectors are very smart they've done this before just look at the number of serious and fatal accidents the mortality from road traffic accidents over a period of 30 years now how come they could achieve this and all the Investments we have in the Life Sciences all the technological advances that I showed you earlier could not be translated in improving patient safety in a hospital environment and how did they do it in reducing road traffic accident a fairly ranging Technologies over 30 years from seat belts to break pads and finally the last technology I don't have a very new car but I've seen and I've driven one which is the what they call the assist uh uh assist driving technology when actual fact it alarms you as a driver it tickles you I think when it drives in the seat that you're about to hit the curb or you're about to hit the car beside you I love to have one of these which might actually push me to recognize someone who's deteriorating so most of our effort has been how do you actually translate technology in enhancing patient safety information technology is the way to do it but the current systems we have are cumbersome they've taken years to install and I don't want to go to the fiascos of the cost of the information technology systems that were installed installed within the NHS I'm going to share with you a couple of slides the end user view ultimately I know this because I'm a doctor doctors only use technology if it's going to make their life easy look at some of the uh some of the enduser uh uh views that we've captured it took one hour to teach a doctor how to prescribe paracetamol through the it system I don't think I'm stupid but it could be more complicated and so on and so forth in actual fact you don't have to ask the doctors and the nurses who using these systems ask the regulator CQC I'm not just some of the reports from their inspections data quality issues have also been highlighted of total of 460 records that could have been completed 245 were valid 280 were invalid and 155 were missing so despite the investment if you don't have an intuitive system that the doctor or the nurse feels that is actually it's helping them to deliver a better care to their patients they're not going to engage with these tools and the end of the day they become an administrative tools in which the Finance director could calculate how many patients came in the door how long did they stay there rather than really looking at the huge potential of information technology which has transformed everything else in our life but not necessarily in healthcare so we've been looking at how could we engage clinicians at the front line identifying the gaps and using more simplified Technologies in developing these Technologies and and this whole idea started with some collaboration uh as you know I come from Imperial uh which is probably the best endowed institution when it comes to the power of engineering science software writing and beside Imperial there's another wonderful organization called The Royal College of Arts I don't know how many of you have visited that but design was something else that I recognized as I matured will have a quite an impact on delivery of healthcare so this is the Helix Center which is a joint venture between Imperial College and the Royal College of Arts so this is the pod which was this was funded by hefy it's actually in the middle of the hospital and the reason we put it in this wonderful peeron between the medical school and the old building uh is because that was the best way of really getting that staff coming into there and really coming up with their clinical problems and at the same time patients who came in there really sharing some of the challenges in terms of the wars they were in and we have a number of designers there you can see Maya who is in there and uh number of others there who essentially designers Engineers software writers who really try to identify these different gaps and develop the different apps as you can see here which is uh which having an impact on a number of different Pathways of care but the one that has been very unique which has really been championed by one of our lecturer Dominic King who's there somewhere there he is uh is a lecturer in surgery back to some of the challenges that I was telling you earlier how could we capture this information better how could we develop an early warning uh systems in which a doctor could prioritize some of the task that they've been given and then intervene and start the treatment so hark is essentially bringing all of these different inputs blood test physiological monitoring and combine those together and really deliver it then to the individual clinician looking after a patient in a ward environment or even in a primary care environment so I'm just going to go through some of the description of it this is a task being ordered it's very simple task you can order it within 30 seconds and I think what's unique about this the person who's ordering that task will actually identify the person of the right seniority in ordering that task the task is created and then the task is received by the individual who needs to ask on to who needs to respond and act on that command the prioritization are significantly easier obviously with certain algorithms you can see here the actual doctor will have a priority of different jobs so at least you could go and see the patient who's sickest in the short shortest period of time rather than having a list in front of them and trying to make a judgment call without real the necessary information so if this is some of the work which was the output of evaluating this at Imperial College NHS trust mainly on the samar's site you can see from a functional perspective there were 11 supplementary communication functions with the device information transfer 22 out of the 24 CL critical factors were better communicated using the app and digital app and finally the response time as you can see had significantly improved by by 30 37% less time taken to respond to a task request so we've developed this you know it's amazing because you go around you think NHS is a basket case but the NHS is not you can actually create Innovations in the NHS NHS should be leading Innovations we've reached a stage after the develop that Dominic and a number of other colleagues in the department recognized that really to scale this up and trial it even within the large Imperial College Trust on the three sites we need some help and we were delighted to meet Mustafa and his colleagues and deep mind you know some of the I I was there in the office yesterday some of the smartest people I've come across and the idea of developing a partnership which has already worked extremely well at roal free as you may know and really taking this up and scaling this up and I know the NHS trust leadership are very excited about this and they've been absolutely brilliant in facilitating this Innovation coming through so I think I want to share with you an example a very simple what I call a Frugal Innovation that could have a huge impact in improving communication identifying sick patients preventing deterioration and escalating care and thanks to the group which is the uh this is the launch of our patient safety and translation Research Center nearly three years ago in the House of Lords and that Dame Sally Davis there who is the director of uh nihr and I'd like to thank finally obviously Imperial College for facilitating this the trust who really with open arms have embraced this Innovation andr but also Imperial Innovation so on that note I'm going to now hand over to the smart guy who's going to tell you how we just take this over and really have an impact on patients and their safety thank you thank you very much so thank you um ARA for a wonderful talk and for framing I think the patient safety challenge that we have taken on so my name is masafa I'm one of the co-founders of Google Deep Mind before I talk to you about the patient safety challenges I want to start with a little bit of a background on Deep Mind who we are and what's really motivating us I'd like to put it to you today that one of the defining characteristics of the Modern Age I believe is that we are overwhelmed by the complexity of the of the systems that we've created all around us we have the smartest experts people who trained for decades and decades struggling to respond to uh some of our most complex problems whether they be climate change or building sustainable Food Systems of fair production or whether it's taming our complex financial markets in some sense what we're struggling to cope with is to make sense of these vast streams of data these high dimensional temporal abstract streams of data that are coming in from all of the senses that we've created around us that somehow describe the nature and structure of the world today 5 and a half years ago I found a deep mind with two of my colleagues and in some sense it was a response to exactly this problem we recognized that we were overwhelmed by complexity and overwhelmed by information and we posited that over the coming decades the real benefits and Innovations and productivity gains that will come in our civilization will be those that we derive from better understanding the vast streams of data that we're creating around us if we could somehow learn the nature and structure of this data well enough to be able to predict its course over time then that would better help us to intervene in this complex world around us and at its heart Deep Mind is very much a British research organization so we have something like 250 staff 180 of the very best AI researchers postdocs phds and professors in the world over the last 12 months we've been lucky enough to score two nature front covers first uh this time last year with the nature front cover for our Atari dqn work where we trained an algorithm to learn to play multiple Atari games old school uh Atari games from the 80s purely from raw pixel inputs just in the last couple of weeks we've announced that we we've become the first ever system Alpha go to beat a human professional um at the game of Go in the next couple of weeks we're actually going to be playing the world champion in Korea for a million dollars in a 5-day Live Match and what we've tried to create at Deep Mind is a kind of hybrid organization we've tried to bring the very best that a focus on long-term research has to bear and learn all of the best bits from Academia and bring those Into the Heart of our organization while simultaneously combining those things with the pace and the scale and the agility and dynamism that we get from working as a company at scale and all the while thinking what can we do that would be meaningful what really is our purpose what are we really focused on and that's where Deep Mind has social impact at the very heart of our core mission in some sense I think this makes us uniquely placed to work within the NHS we have the resources and Technical expertise of one of the best technology companies in the world and yet we also have the agility of a startup structured as we are in these small cells that have a great deal of autonomy and can work really really fast but all the while driven by the kind of meaning and purpose of a social impact organization so let me start with what I think is a very brief diagnosis on why healthtech has underd delivered over the last couple of decades in some sense it's sort of struggle to do what nurses and doctors want what they know that they need one of those reasons I think is because it's consistently delivered I guess what you could call a kind of one siiz fits-all or top down system rather than something which has been co-designed in collaboration with clinicians directly typically the characteristic of these deployments is that they've been very large scale step change introductions where there's really only one version you buy a product off the shelf you deploy it to a thousand person organization and it and it updates and improves very slowly and this isn't the the characteristic of the sorts of apps that you'll be using on your phone every day and that you're also familiar with the other thing that I think has has really slowed thing down things down is that typically the companies have stuck with proprietary standards which tend to create closed systems and that means that they've defined their own database schemas they've defined and hand curated data in a particular way which is the reason why a lot of your technology systems don't talk to one another and I think that's a really big mistake that prevents both academic research CL clinical epidemiology s and also startups or doctor hackers who have a great idea from getting access to that data and being able to experiment and deploy their own prototype app we've already covered some of the common complaints but I'm sure you'll be all familiar with the questions of why on Earth are we using pages in 2016 where else in the world do we have pages in production in such an important environment there aren't enough desktop computers on a ward I've heard this over and over again in the last nine months and it sort of amazes me when you do find one the batteries probably died or the smart card reader doesn't work or there's a key missing in the keyboard and there some really fundamental infrastructure that doesn't seem to be quite right quite remarkably I've actually experienced faxes in operation I I genuinely didn't know that these things were still in use until and and until until I started collaborating with my friends at the Royal free and and at Imperial I've also heard the complaint that we wait for hours sometimes even days for x-rays to be transferred between hospitals clearly this is something that should be digitized now look this is obviously an incredibly intimidating arena for technologists in some sense the history of the last two or three decades represents a kind of daunting Prospect this hasn't been a particularly successful place for Technologies to work but I want to argue that if the NHS needs new technology as ARA has so rightly put then the onus is really on us um technology needs to find a new way to work with the NHS let me try and present you a few of those ideas today this is actually incredibly simple ask the question what does the user need and when you've asked that question ask it again and again and again and let that be the Guiding Light shaping everything that you do as a technologist doctors and nurses already know what the problems are they work on the front line day and day out they're very smart people they're very creative and often they even know what the solutions are what I've heard is that doctors and nurses have been demanding one mobile first why can't I have data in my hand at the right time I need the right piece of information right in my pocket so that I can make the appropriate diagnosis or take the appropriate intervention this needs to be sharable across teams why can't I communicate this information in my own views on on on the current state of the condition of my patient to my colleagues and this should be super easy to learn to use no one's ever had to go on a training course to learn to use an app that you've downloaded from the App Store and the truth is that these Innovative workarounds are already in use you know it's sort of a taboo topic but the vast majority of doctor doctors and nurses today are using WhatsApp to communicate patient identifiable data to share images and they do their best to keep it uh anonymized but clearly there are risks to the current system they don't comply with the most upto-date NHS standards and that's something that we've got to tackle but frankly we really can't blame clinicians they're motivated to really make a difference and they've committed their careers to helping patients to have the best possible care why wouldn't they take advantage of what they obviously see to be a tool that makes their lives easier and I think the challenge for us is how can we deploy some of the best user designers some of the best Engineers some of the best machine learning researchers to work on some of the problems that really matter so let's tackle the first patient safety challenge that ARA put to us how do we identify which patients are at risk so what's actually going wrong here why aren't we why aren't we detecting them on time well part of the problem is that we're not recognizing their physiological deterioration and that creates a breakdown in reporting and then we don't get access to the data in real time largely because we're dependent on these sorts of hand uh amended charts they're really difficult to share they're very difficult to interrogate you only get information from them when you go to seek it that's an important Dynamic you only get information from them when you go to see it it's exactly the opposite of the way that your smartphone works your smartphone tells you when there's an important piece of information that you might need to pay attention to so let's take the example of acute kidney injury that we've been collaborating with the RO fre on contributes to about 40,000 deaths annually and quite remarkably it's been estimated by NHS England that around 25 % of these cases could be preventable and this cost something on the order of a billion or so pounds in direct NHS spending it's a national priority and and in the last couple years the NHS have issued a patient safety alert to try to frame um how uh clinicians respond to this but essentially this has been a PDF with The Cutting Edge best practice emailed to everybody and stuck up on posters and I think that that can be improved so I'm going to tell you about a collaboration that we kicked off with Dr Chris Lang um and members of his team at the Royal free over the last six or seven months or so Chris is a consultant nephologist and kidney expert and he's also the associate um medical director responsible for patient safety at the free so we set about co-designing directly hand inhand with clinicians at the free the first step is how can we understand the user who is the user what do they do how do they work on a day-to-day basis we did scores of focus groups and onetoone interviews we try to empathize with their day-to-day experience by shadowing them and trying to learn from what happens minute by- minute hour by hour during their day and it was great fun for us we use that to try and identify what their unmet needs are today and we use those needs to then prioritize our own requirements as you see here in some of the images immediately as soon as we've got a sense of what their needs are we start to build things we we prototype we experiment we sketch things we do wireframe and then immediately we want to test them the answer is rinse and repeat this process actually works we go over it every 48 Hours as soon as we've heard that there's a new insight that somebody's had we go and put it to wireframe we sketch something out and then we take it back to the nurses and doctors and we say how does this look does this make sense how would you like to change the buttons the color the choice architecture the menu options does this flow make sense we learn we iterate we test and we constantly improve the other component to our codesign has been to try and actually map out the complexity of this pathway so it's been estimated there's something like 900 to a thousand different patient experienced Pathways in the hospital system so what does it look like for us to actually try and map this complexity out well from the patients perspective it's it's pretty painful lots of different things are happening to this patient throughout the process but if we zoom in a little bit we can see that this red part pathway here is where the vast majority of the activity takes place and in some sense this is when the patient has already developed some kind of life-threatening complication of Aki so what we decided is how can we go and look backwards what happens before that part of the process how can we do better risk assessment by looking at the underlying comorbidities how can we do a better job of real-time monitoring of the data in order to try and proactively intervene in this patient's care in order to drive their experience down a parallel pathway that is more stable so obviously the best way to do this is to build something and test it here is the um app that we co-designed with our colleagues and friends at the RF free I'm going to run you through what it might look like so we log in immediately you see uh the patient list and so this could be sorted by Ward or by specialty or by consultant and so if we go ahead and just tap on um the first patient we see that this patient actually has a pretty serious acute kidney injury level three we're able to browse through the results of the retrospective Trend analysis and we can see that there's been about a tripling against Baseline of this person's creatin they're in a pretty bad way and we also want to take a look at the um potassium we see that it's very highly raised at at seven so this could indicate hyperemia we may want to zoom in and take a look at the CRP which is also extremely raised and compare the trend analysis against our creatin zooming out again we see that lactate is at a 5.1 and this might give us an indication um that this patient's cause of AI has been um a sepsis if we want to see in a little bit more detail we can see the retrospective blood test results exactly who ordered it when what was the time and what was the value in more detail we can also zoom in and see a little bit more about the patient details you can see here that this patient has come in with uh type 2 diabetes and hypertension and also if we look down a little bit further we see that they originally presented with abdominal pain and were subsequently diagnosed with acute appendicitis we also have the contact details at hand so that we can contact both the NEX ofkin and the general practitioner so let me show you a little video now of I think clinical teams like to be able to Define problems and team up with people who can help solve them it's been very interesting to have this collaboration between Hospital in a company like deep mind we're very interested in the idea of streaming at speed clinical data to mobile platforms because most of our teams spend a lot of their day on the move because of doing their clinical jobs in my work I'm quite involved in implementation science which is a way of testing clinical changes in practice working with deep mind health has followed that model completely that way of testing iteratively testing so planning something doing it studying whether we thought it was good or not is extremely valuable what's nice about helping to design something is that we have a say in how it will be useful what I've seen about the design of the app is that the screen is clean and it makes sense and there isn't too much information on it it's not all in your face technology should support your work should support what's going on with that patient so if you can communicate that easily to people it's going to save time and it's going to help you do your job and it's going to help you get to your patients quicker and so getting to your patients quicker is of course the part of detection which is only half the story to respond to ARA's second patient safety challenge the real question is how do we actually manage the intervention on once we've identified which patients we need to go and see at the bedside and this is where I think the delivery of effective care has been delayed according to ARA and Dominic and their research and up to 50% of deteriorating patients and their response was really quite incredible they went off and created hark to support nurses and doctors to deliver care much more effectively clinicians responded 37% faster with hark than with pages and so tonight I'm really proud to be able to announce that Hawk's going to be joining forces with deep mind health and becoming an integral part of the service that we go on to provide let me take you through an example of what it might look like to manage tasks with Hawk so let's take the the case of requesting a medication order the patient clearly needs a prescription the nurse goes on to bleep a doctor the doctor hopefully calls back in good time the doctor then will go to the ward and have a face-to-face conversation with the nurse decides that yes it's right that we need a prescription and then go on and inform the nurse that the request has actually been completed and then finally the patient uh is able to receive the medication of course this assumes that you get through on the first bleep and that the poor nurse isn't just bleeping constantly because the doctor that that he or she is trying to get hold of is actually on half term this week or is busy in theater all day this is what should happen really doctors should be able to prioritize and manage their tasks on their phone they should be able to accept a task from a nurse and tick to say they're handling it they should be able to have a conversation with a nurse or a fellow doctor in messaging in their app about that specific task and then they should be able to send a confirmation message when it's been done completely remotely saving a vast amount of time so this evening I'm really pleased to be able to announce that we're going to be launching streams by deepmind health and our objective is for the first time ever I think to integrate both the detection and task management into a single platform the plan is for it to be a beautiful and intuitive and most importantly really easy tolearn experience that anybody can download in the in the clinical setting and start using streams will deliver the right data to the right clinician at exactly the right time and really the objective here is for us to try and shift some of the 97% or so of activity in the hospital today which is reactive further towards activity which is proactive and ultimately preventative and this of course is where our Cutting Edge analytics and machine learning comes in how do you prioritize the series of alerts that go to a patient how do you ident go to a doctor or a nurse how do you identify which person on the clinical team should be receiving the right task and how do you ensure that they've been followed up in good time and of course the real question here is ultimately how will we be able to provide patients with access to this data how we be able to empower patients to play an active role in their own treatment and in their own experience and really play an accountable a role in holding accountable the treatment that they receive whilst they're in hospital so I really want to leave you with the idea that this time I think it's really important that we have to do things in a fundamentally different way and for us this is about partnering and co-designing with clinicians clinician Le technology is actually a new way of thinking about how we co-design interventions in the system I truly believe that together we can revolutionize patient safety and this evening I'd like to invite anybody and everybody who's interested to ping me an email personally and come and hang out in our office and come and talk to us about the apps that you have in your mind the things that don't work for you and your visions of how this might be able to work come and join our partnership thank you\n"