ashin might the ultimate presentation of this this workshop is AI and virtual reality and integration for neurological health care I mean to say that it can't be applied to the Majin Erik case but my maze is at the moment focusing on neurological health care some of the case examples I will be using these with stroke and I take a particular perspective as the chief artificial intelligence officer of my maze from a cybernetics perspective which is somewhat of an unusual word we'll come back to it but it's quite different from the bottom-up concept of AI it's more of a top-down perspective looking for feedback loops control loops which are particularly relevant in the healthcare perspective this is going to come out in the presentation I trust okay so first of all let's define the problem that healthcare has and why AI in virtual reality has an opportunity to be applied here so the first point is is that there is a limited number of healthcare professionals per capita in the UK it's going down at the
moment and that's not necessarily unusual even though the need for healthcare is increasing as longevity increases neurological healthcare damage to the central nervous system actually increases so there is a crisis developing in the industry and the costs of some of the technology that we are using to diagnose and provide therapy in these neurological health care is not becoming that much cheaper faster and so the cost to provide care is a significant issue so we really need to address these issues not least in the industrialized new developed world but also you extend that access to the newly industrializing the less developed world I'm actually talking at a United Nations sustainability Development Goals conference on the 25th in Geneva my mace is based out of Los and in Switzerland so there's a bunch of really important issues here defining the problem and the solution and it often comes out that AI is a solution before defining the problem so I definitely defined the problem first the solution for many Fork is to the use of AI integrated
with virtual reality particularly in the therapy set side of things and the I tend to look when I look at a solutions benefits for four factors productivity we should be using AI to augment people or perhaps replace people so that productivity increases the quality of the diagnosis let us say should improve at very least of human activity otherwise the the business case the use case is is theory ated the opportunity for AI is to personalize to make healthcare less of a robotic opportunity which sounds weird but the idea is is that with AI there should be time for particular patient individuality to be recognized which in the traditional model unfortunately is often lost because the the time pressure for healthcare is so op oppressive there should be the opportunity to discover new Ornelas se patterns which maybe can be lost in the confusion of time in a human case so for example an individual may have stroke but also may have depression may have a dementia and recognizing the interplay for particular individuals is a
real issue for many human diagnosis systems so the goal is to you know provide a solution with AI and virtual reality to have a massive implication in pact across these use cases and it's sufficient enough to refer to it this paradigm shift as you know to shadow to parallel the industry 4.0 and 5.0 concepts to say it's a Health Care 5.0 a major paradigm shift and having said that there's great opportunity here there are also significant risks and I will touch on those because the solution must be made in the context of the risks and not just the opportunities so we're going to very quickly look at some healthcare opportunity the size and magnitude of it as indicated by some independent consulting management consulting groups I tend to use this PwC graphic a lot in my presentations it's a cumulative growth curve in terms of the dollar opportunity actually a pound GBP opportunity the number you probably won't be able to see it unless you're very close to some of the screens but it's 15.7
let me get make sure I just trillion pounds by 2030 which is what some 11 years ago and just do it's very easy to get lost in that number fifteen point seven trillion plans that AI will impact cumulative over there of the over the years that's twice the amount of gold that has ever been mined in human history will occur because of AI and this is not just PwC there are many commentators and we're tracking on that growth health care is recognized actually as the or second major use case contributor but that's a I I've also mentioned about VR there's a there's a bunch of VR companies in the health care market space my emotion is indicated by one of them thankfully by citrus pits but in terms of this major portion by sector mostly VR is in the entertainment industry but the second-biggest application is in health care so there's a real resonance of opportunity here I mentioned I was going to talk about risks and recognizing my time as well this is a
I think world World Economic Forum graphic and it shows it's very difficult to see with the shading here the benefits versus the negative consequences so the greatest benefits in terms of emerging technology is on the right of this diagram so you have biotechnology artificial intelligence and robotics and many of you will recognize the hype which is too often talked about in AI in terms of existential risk and bias and what-have-you so it's considered a very risky technology but on the the positive benefits which are applied actually that's on the right and the negative consequences that restaurant is on the on the on the vertical scaling it's also the same so the biggest opportunity but the biggest risks and that applies actually the things that are also relevant in the healthcare side whether that be new raw technology or proliferation and of the ubiquitous IOT sensors which collect behavioral analysis that collect a bio signal data so we have to recognize the opportunity when we're building solutions but recognize the risks would make sure we mitigate
those from the design principles in the first case okay I'm going to talk a little bit now about why I think it's worth seeing the paradigm shift benefit and why I refer to it and others have started to with to the shifters health care 5.0 so let's just recognize some of the points I've mentioned earlier there is an ongoing global health care crisis costs increasing access to human health practitioners is actually being stable it takes 15 years to train a clinician a doctor so even if there was the money to invest it would take a significant lag to meet to make a major impact with traditional human technology or human beings a whole bunch of stuff here rampant corruption in in many countries of the world we need systems with transparency the lack of patient engagement patients are being treated in a less personalised way for the sake of productivity that's a major issue the the fact that we haven't fully fully gone into the digital transformation of the healthcare industries helping to build large
data leaks of patient data that we can mine effectively for the better utilization of those services and with there are security and privacy issues there so all of these contending opportunities and risks are causing kind of a stalemate situation in terms of building better healthcare so if you look back in the history of the paradigm shifts in the provision of health care you'll see that it's gone through three four five the last one is the artificial intelligence personalization a previously art of digitalization automation industrialization and basic production so you can I believe these slides will be made available and don't intend to go through each each and every bullet point here the point I'm making is that there are paradigm shifts not just in industry but in terms of specific applications such as in healthcare and that personalization is driven by three factors the quality of healthcare provided the responsiveness the timely responsiveness of health care being provided people not having to wait three to six months for hip replacements etc etc and the fairness
so the access to this service whether that be because of the economic circumstances of individuals or whether they live in well-developed or less developed countries so there's a whole bunch of issues they're driving 5.0 and one of the points here and I'm starting to refer to some of the concepts which I believe are cybernetic inertia in nature is this loop and continuous continuous learning opportunity so we're in a situation where we can use a day I for diagnostic prediction and classification and we are in a position where we can use AI to make appropriate interventions early interventions personalized to the individual but we wish to build this in a holistic model where the data is being used by all the components in the delivery of the individual processes such that we can affect that productivity benefit and learn from that and where I'm going to come back to a reinterpretation of this slide later on in my presentation which relates to a patent-pending process that I have written for mine maze in this respect so
let's look since I've been talking about mine me is at some of the ideas that we have a products which are relevant and Stroke which a relic which have been fda-approved and I'm going to go into a little bit of those next so let's talk about mine motion which is a product which we are bringing to market this year and what indeed have already started the selling process through Germany and in the UK and the USA in other countries of the world and this is a new raw Riya ballot rehabilitation program whereby we are specifically focusing on stroke patients who you have had an incident there I've gone through the so called acute phase which is basically treating if that if the stroke was due to a hemorrhage which is bleeding in the brain which is a potentially fatal condition and that has to be interacted with in the first minutes no we're not talking hours of that condition the majority of stroke cases incidents are however ischemic stroke where blood is not so much
leaking from a rupture but has a blockage and oxygenation is being denied to the neural tissue and after a few minutes again that neural tissue will will die and individuals start to have characteristic conditions in terms of loss of mobility in terms of slurring of speech etc etc and again action has to be taken as soon as possible and there's a period of time when an individual's recognize brought into treatment of a few days to to prevent further damage that's a few days in terms of treatment in terms of therapy then there's the sub acute phase when the brain in a mysterious process that we do not fully understand is engaged in neural plasticity whereby there is a process an organic process where the brain attempt to transfer functionality from one part of the brain to another part of the brain and individuals may start to see small improvements but unfortunately that new neural plasticity phase is quite short this phase that we call the sub acute and what we really need to do is
to try to interact and treat the patient in the most effective way possible as early as possible in that sub acute phase because if we do saw we will have a faster recovery now as far as there are my best collision clinician colleagues advised me we are in a situation where even with the best therapy it's difficult to improve the end state that means to say we can predict with a degree of accuracy knowing the algorithm is called prep - and it's basically uses things like shoulder arm and finger extension so if you calculate the metric of the damage from those together with information from the National Institute of Health together with some magnetic stimulation to see if circuit area so to speak in the neuro cortical column has been damaged we know that this patient is going to get to this end state what we don't know is that if the end state is going to be achieved in two years or in three months now because of the cost of therapy we want
to do it as soon as possible and the patient wants to recover as soon as possible so what we have to try to do is to do the most effective therapy as soon as possible and to do so with the most increased dosage of therapy that the individual can sustain now with limited clinicians available with limited let us say expertise because not everybody is an expert clinician or therapist that often is not occurring individuals get physical therapy even in the subacute phase for an hour for one or two days in the week and then after having gone from the subacute phase into the chronic phase which is like week two or three going onward four to six months or more then their outpatients typically either coming into a clinic once or twice a week for one hour or two hours or doing X at home sometimes with therapists in attendance sometimes not and that is not the best way to increase the tolerance of dosage of therapy it's not pharmacological typically it's therapy a specific
focused targeted there is achill therapy for the condition of the individual to excess if they have problems with movement in the elbow then have elbow focused therapy if it's the shoulder not being able to raise beyond a certain level then find therapeutic exercises which focus on that mortar deficiency I trust I make myself clear and the way that we are we believe we can do that is using AI to diagnose and to do problem do the prognosis but also to advocate the best therapy for the individual and to do that to advise in the subacute phase the therapists to do the right things to help build up the quality of that therapist to do the most effective recommendations but in the chronic phase when the users are at home to have at home AI enabled gamified very virtual reality to be able to use motion detection systems for the kinematics to determine the the improvement of an on improvement and to pivot the therapy to keep the individual being occupied with the correct therapy that's
what we're trying to do at - and indeed we are doing so we have the mind motion Pro platform which is in typically applied in that subacute phase and the mind motion go in the chronic phase at home and we're using AI we're using virtual reality in gamification games to be able which are targeted to specific motor deficits the companion is an interface to the system that is on a tablet and that builds that the family of the mind motion platform this is a video unfortunately it's not working but it doesn't really it's a nice video in the sense that it tells you what I've just told you but we don't need to go and spend time on that because that we've only got 10 minutes this particular diagram I'm afraid it's not coming through with perfect clarity I hope you can see it on the screens I hope you'll be able to access it but we've sketched out the journey of the therapist for specific individuals and what unfortunately I think it'd be very
difficult for you to read is we envision the training content but in terms of being continuously evaluated not just in terms of the kinematic improvement but the emotional journey of the individual if our intent is to engage the patient in such a way that they are excited and motivated and happy with the therapy and they're not depressed who are not in pain who are not fatigued by the therapy then we and if that's being done at home without a human in in in attendance we need our AI system to have the effective computing capable to say I know I'll say it in English I think my patient is fatigued because he's dropping his head I can detect that by kinematic motion or because she has galvanic skin resistance which suggests fatigue or whether there's a facial recognition maybe eye tracking which suggests distraction so all of these effective computing multiple modality impact I call it artificial empathic intelligence that we can engage with that and then with a virtual assistant who's playing the therapist in
attendance to suggest comments like are you tired mr. Jones shall we take a five minutes break because if we can engage with the patient at that empathic level the opportunity of keeping them engaged and constant in terms of their commitment to use the system may be the difference between six months of therapy three months of therapy and that three months of therapy's effect is it's very valuable to the patient to the recover as quickly as possible the normative behavior but it's also effective in reducing the cost of providing their therapy this goes a little bit further we'll skip past that so here is a gamification scenario so there's an immersive environment we call it the video arcade and we have a gate a specific game this is called the whack-a-mole challenge you'll be familiar with it some of you may be where we overwhelm apply a player with actions that they need to take in a specific time frame the actual action is to hit them all on the head and then another mole will
pop up and then you hit them all in the head so you're making movements like this now those movements may be very difficult they may be impaired because of the elbow and the shoulder movement the patient may be moving their toss or so that they can move that so we want to be able to sense that movement and say oh that's not normative can you sit back in your chair and extend your arm for that to advise the movement the kinematic movement which is relevant to them to the impairment improvement so we define the strategy we design a game to do the right therapy here we actually monitor the movement of the upper limb in the case example I gave and we are monitoring the movement of the elbows shoulder the wrist and we build a workspace and that workspace if the therapy is going to be its achieving the goals of improving the motor deficit is going to increase in time so we were not only monitoring the kinematic data but we're also monitoring
the game performance and making inference as to what we now call a a a level progression evaluation or recommendation engine where we are saying ok we're now ready to increase the level of the game to speed up the number of Mol's to begin we're or we need to change the game because we've achieved our elbow improvement but we now really need to focus on more shoulder related gamification so here's a can we check this so we have developed kinematic detection motion detection we originally started out using Microsoft Kinect which many of you be familiar with now Microsoft went through a period of time where they retired that product for about a year and a half and within my maze we bought up 2000 of these systems as a stock so that we would have the ability to fulfill that and they did in Microsoft didn't provide a replacement until about 2 or 3 months ago and it's still on backorder and for the Microsoft Kinect as you were DK development kit so what we did
to make sure that we and we're not how can I say snooker to use an expression is to build our own motion connection visual system and if we just tick on this video again please I'll just point out the how can I say that accuracy of the human to the movement and it's the it's the actuators in the robotic sense it's the elbe it's the it's the joints that were really seeing and looking for and deriving algorithms to determine the work space okay we do other things in mind maze we take not just provide gamification we can provide that through a traditional virtual reality headset on the left you can see a diagram an image where we're also doing the G on the right what we're doing is actually using EMG sensors to detect the motion of facial muscles because by doing that we're able to do things like detect that this gentleman over here is only eighth floor of our office is actually snarling in road rage whereas this lady is pleasurably experiencing the
driving experience it's now actually that kind of emotional affective computing is what we're after and we can do that with visual image in two phases so we don't have to have a mask we can actually have a camera to detect that affective computing okay yeah I really am coming up for time this is an example of the mind mask with integrated with EEG so we're able to detect many of the affective cognitive states of the individual I'm just going to go quickly through into the cybernetic concept which I promised we would do cybernetics is a concept that was developed really in the sort of mid part of the last century and prom was prior to artificial intelligence and basically the concept unfortunately over the diagram some of the words are off on the left here was basically the scientific study of control and communication in man and machine and it was a top-down perspective ai is a bottom-up perspective where we seek to develop cognitive functionality equivalent in an artificial way but from an emergent
top a bottom-up exercise and it's a transdisciplinary concept because it increase in cludes control theory information theory optimization general systems research and that's really useful in the healthcare environment on the image on the left there you see a pacemaker actually I'm not sure whether this was the pacemaker of a specific individual but let me just mention poor Benedict the poor prior to the current one was a cyborg in the x2 the context that he had a pacemaker that was a cybernetic device that would took the input about what signals were going on in the heartbeat and and basically augmented that control process but from such small beginnings we're moving into BMIs where with EEG and other sensors we are detecting intent there's an almost day by day there are new papers and articles about the extent which BMI is really helping us and in I think it was at the not the Olympic Games it was at the World Cup in Brazil about six years ago there's the walk again project whereby a brain machine
interface was used for an exoskeleton to help paraplegics walk and what was really interesting by the way from that walk again project is something that wasn't expected have occurred which was people with the BMI and with the physical exercise of walking were actually recovering spinal connectivity the the the power of the mind in being asked to generate a signal to drive an exoskeleton was causing not a neural morphogenesis that a recreation of neural connectivity is really interesting Coralie of that research so what's going on in a typical feedback cybernetic process is there is an input desired so let's move your arm up now there's a controller there which is a digital model in a certain sense in the modern sense which suggests that this model can be done with this kind of action but then you apply that signal to the actual control plant the arm and you will or you will not get the actual result that you expect so dependent on that you feedback the process it's this loop this this continuous loop
which is the cybernetic principle and we measure and detect of a variety of input data we estimate do a diagnosis of the health state we go through therapy and prognosis and guidance and then we go back and we're constantly collecting and managing that process so it's it is a cybernetic loop so what we've done in mind maze is create a holistic model of how this all works so we're basically looking at patient data developing a database we have multiple therapies that we can that we can use to attain good outcomes ideally so therapy's daunted give a good outcomes that you only give fair or poor results we want to be able to collect this data so we can do data mining and then once we've done data mining we can look for new neural network models of therapies in a digital model and keep that in a feedback process I recognize that I'm very short on time and so this cybernetic healthcare is taking in genomic data and medical knowledge data from sensors and personal
models and through cybernetics is helping to keep that as a constant floor and feedback process so just ending up with a few courts here the CEO of Google basically says that AI he expects us to reshape healthcare you have to work through some of the regulations for privacy and the like but he sees that it's one of the biggest areas that AI will impact in the next 10 or 20 years and Tony Parisi of Unity Technologies on the VR side again sees that the impact on education and therapy is a kind of education we are educating stroke victims to recover their functionality through the use of vr so with that thank you very much I do apologize for being two minutes over time bear with me for that if you have any questions I'll be happy to take them out now or privately thank you very much