Preparing for Regulatory and Technological Change Across the Medical Device Landscape

Preparing for Regulatory and Technological Change Across the Medical Device Landscape

Show Video

Okay and good afternoon everyone um I think we will make a start um thank you for joining   today's webinar um which is Preparing  for Regulatory and Technological Change   Across the Medical Device Landscape. Please note that this webinar will be recorded   um if there are any questions please put them in the Q and A section which is on the top bar of your   team's window and we can run through these later during the Q and A section of the webinar. If you have any feedback throughout, please use the link in the chat that will be provided   um and we will get going. So I am Paula Sweeten, next slide please. Great, thank you. I'm Paula Sweeten um and I've been a Project Manager at Inno  Scot Health for just over two years um before joining InnoScot Health I previously worked   in clinical trials for the pharmaceutical  industry and in academic research management.  

Leading today's webinar we have Elaine Gemmell who's the Head of Regulatory Affairs at InnoScot Health and   our guest speaker Dr Paul Campbell, a Senior Clinical Advisor at the MHRA. Elaine joined    InnoScot Health in 2002 and helped to establish the organisation. She's a certified ISO 13485 and ISO   9001 lead auditor with experience of regulatory approval and CE marking for medical devices. She   has a BSCB Eng in engineering and previously worked for Motorola Semiconductors for 20 years.  

Following on from Elaine today we'll also have Dr Paul Campbell who is a highly respected clinician   and expert in the field of healthcare  information technology and regulations.   After starting his career as a pharmacist, um Dr Campbell transitioned to medicine and became a   consultant anaesthetist within the Clinical  Informatics team um at sorry became a in   20 school 20 over 25 years of experience across multiple clinical and digital domains. He's held   several clinical NHS and E- Health roles, including as a Clinical Director and Clinical Informatics um team in the NHS National Services Scotland. He's also a clinical advisor to the Scottish   Government Digital Health and Career Department  and is currently working as a Senior Clinical Advisor within the Software and AI Group at the MHRA. I'll now pass over to Elaine who will take us   through the regulatory changes that are coming to the medical device landscape and then we   will go over to Paul who will take us through artificial intelligence as a medical device. Can I have the first slide please Jane? So welcome to to everyone who's joined the webinar today um for a number of months I've   been doing presentations and training  sessions on how you regulate medical   devices and looking at what's happening with respect to the Europe and the UK. Today I'm  

going to do something just a bit different um as you're aware there have been lots of changes   in the the regulatory landscape due to changes in regulations standards been updated additional   requirements for notified number of bodies and just trying to keep abreast of the changing   technology and what has happened in the light of the UK exiting the European Union. What I hope to   do today is to summarise all of the changes and to provide some clarity and what we can expect to see   when these um changes are fully implemented. Can you go to the previous slide please Jane. So before   we start um what I will do is take a look at the regulatory landscape as it has been prior to the   UK exiting the European Union. So while the UK was part of Europe um everyone worked to the Medical   Device Directive and as part of the European Union the UK also worked to that directive.

Clearly this directive has been in place  for a number of years and there were some   deficiencies. Technology has moved on there's been new requirements, it hasn't really kept up with all   the new technology. So there's a recognition that we needed to to update and to make the medical   device regulations much more robust. So the Medical Device Directive, being a European directive had   to be transposed into law in the United Kingdom, and this was done under a statutory instrument   so the statutory instrument that amended the European directive set out three different   pieces of legislation within the UK that we work to for device regulation. There was the directive   for medical devices, one for active implantable devices and one for in vitro diagnostic devices.  

The CE Mark is the Conformity Assessment Mark that you apply to products to put onto   the market, to show that you've complied with all the requirements to make them safe and effective.   The new requirements that I talked  about being more robust, were set out   in the Medical Device Regulation which was intended to be implemented throughout Europe.   Being a regulation, this was directly  applicable in all member states and   there was no requirement to transpose it into into local laws. So the Medical Device   Regulation was set out initially published  in 2017, with a three-year transition window   so it was expected to be fully implemented throughout Europe and the United Kingdom in 2020.

Had that happened according to plan um the UK exiting the European Union had set out within   a withdrawal agreement that at December 2020 there would be a snapshot of particular laws   that were implemented in Europe and these would automatically become UK law. So had the Medical   Device Regulation been implemented in Europe in 2020, in May, at the end of that transition   window in December 2020, the new regulation would have transposed into UK law and we   would all be working to the the same legislation. So what actually happened? Next slide please Jane. Because of Covid, the implementation of the Medical Device Regulation in Europe was   delayed by 12 months. That meant that it was no longer going to be implemented in May 2020 and   was pushed out to May 2021. That meant that at the end of that transition window, the law that   was in Europe at that particular time was the directive so that was what transposed into UK law.   So now what has happened because of that delay, is that the UK and the European legislation   are different. Europe currently regulates to the Medical Device Regulation and the UK currently  

legislates to the Medical Device Directives. When the UK exited the European Union, they   um put in forward a proposal for a new Conformity Assessment Mark so that is the UKCA Mark and that   will be currently implemented for new devices regulated in the UK and will be applied to put   product onto the UK Market. In Europe the CE Mark still is applied for products put onto   the European market. We'll talk about some of the plans for accepting CE Mark and and the time scales   associated with that um a bit later, but currently the CE Mark can still be, CE Marked devices can   still be put onto the UK Market, however UKCA Marked devices cannot be put onto the European market.   There is a slight further complication with   um Northern Ireland, prior to Brexit trade between Northern Ireland and the Republic of Ireland was   quite straightforward. Both territories were in the EU, both shared the same trade rules  

but the European Union has very particular border checks that have to be done when goods arrive from non-EU countries, so when the UK exited the European Union had Northern Ireland   been considered a non-EU country there would have been additional checks put in place and this   would have implemented a hard border between the the countries in Ireland. To prevent this   happening Northern Ireland continues to  follow EU rules on medical devices and a   CE Mark is required to place devices on the Northern Ireland market. Next slide please. So let's look at the framework for Medical Device Regulation. For both the directive and   the regulation the process is the same. The first stage is to satisfy or to identify the   essential requirements that are set out in the directive that are applicable to your device.  

Just to simplify things for the purpose  of this slide I'll I'll refer to to   UK legislation so we're talking about the requirements of the Medical Device Directive. So once you've established what essential  requirements apply to your product you'll   then demonstrate how you're going to um provide evidence of conformance with these requirements.  This is normally achieved by using harmonized standards, although as a manufacturer you are   free to use other solutions to prove  compliance with the requirements.   Devices should be designed within a  quality management system framework. ISO 13485 is the harmonised standard that sets out regulatory requirements for medical devices.  

It was very recently updated to add annexes showing the relationship between the clauses   of the standard and the requirements of the European Medical Device Regulations,   and IVD Regulations. So it's also now  harmonised to the Medical Device Regulation. Implementing standards will result in a body of evidence that will demonstrate compliance with   the essential requirements. All of this evidence is recorded in a technical file which is pulled   together within the technical documentation, and the manufacturer signs a Declaration of Conformity   to state the requirements of the directives have been met and this will be also included in the   technical file. Both the technical file and the implementation of the quality management system  

will be examined by an approved body and the UKCA Mark can then be applied to the product   and it can be put onto the UK Market. There are a few exceptions to this conformity assessment route.  For very low risk devices that are in Class 1 that don't have a measuring function and are not sterile,   an approved body is not required to examine the QMS or the technical documentation. The process is   that the manufacturer signs a Declaration of Conformity, he effectively self-confirms that   you comply with the requirements and the CE Mark is applied and it's put onto the market. Next slide please. So we're now going to talk about the  implementation of future regulations.  

Clearly the directive was already established as being not necessarily fit for purpose in   today's new technological age and the UK had always intended to go to a much more robust   um regulation framework. So in 2021 the Medicines and Medical Devices Act was put in place to allow   the government the powers to amend Medical Device Regulations. In September 2021 the MHRA launched a   public consultation to gather views on future UK Medical Device Regulations. The consultation   set out particular scenarios with questions and these questions very closely aligned with what is   currently in place in the European Medical Device Regulation. Manufacturers, Trade Bodies Health   Boards, Universities and any other parties with an interest were invited to comment on the proposals.   In June 2022 the response to the consultation was published. If you're interested or if you  

want to see what the the thinking was, this is available on the MHRA website for you to download   and look at. The process was that the feedback received from the consultation was documented and   then the government provided some clarification and then set out their response, and in this   response there's quite a clear indication of what the future UK regulations will look like.  As a result of this response to the the  consultation there will be three statutory   instruments laid in Parliament in 2023 to implement future legislation. In April 2023   there was a statutory instrument put in place um to extend the time frame for acceptance of CE Marked devices into the UK, um and I'll give you some time skills associated with that just shortly.   In the summer of 2023 we expect a statutory instrument which will implement   much more robust post-market surveillance requirements for manufacturers. This will be done  

ahead of the full implementation of future legislation. So this is put on post-market   surveillance right up front and implementing at a much earlier stage than the new regulations. In Winter 2023 towards the end of the year we expect a further statutory instrument on   the Future Medical Device Regulations and this will be the first indication in full   of what the regulations would look like. The expectation is that the core aspects of the   future regime for medical devices will apply from the 1st of July 2025. Next slide please. So just very briefly what can we  expect to see in the new regulations.   As stated the questions that were asked very closely aligned to the EU MDR so it's   quite safe to assume that it won't deviate too far from what is implemented in Europe.  

Although we won't cover all of them today, some of the proposals for the new legislation   will include um changes to scope, what comes into scope for medical device regulations   and this will include devices without medical purpose. Some examples of this would be   equipment for liposuction, um anything with lasers, perhaps used for hair removal or tattoo removal,   dermal fillers, non-prescription  lenses, or cosmetic implants.  There'll be changes to the classification of some medical devices and this can be quite   significant for some manufacturers in  particular the implementation of a neural   for software may result in a lot of Class 1 devices which were previously self-certified   been reclassified to Class 2a and above which will mean that an engagement of an approved body or   a notified body will be required. Other changes to classifications will be concerned with IVF in   and assisting reproduction technologies, surgical mesh, total and partial joint replacements and   other technologies which potentially will go into the highest risk classification of Class 3.  

There will be new essential requirements introduced to the directive to make provisions   for emerging technologies such as nanomaterials and there'll be new requirements on manufacturers   to have financial covenant in place for compensation when adverse instance occur.   There'll be other changes that will provide clarity on the definition of Health Institutions,  and the exemption will no longer apply but there will be a requirement for registration of medical   devices that have been manufactured or modified in-house by Health Institutions. There   are particular requirements for a qualified person to be available to to manufacturers um on   a permanent basis on for medical devices to have a unique identifiers. There are lots more changes   that will come into force and if you're interested to to have advanced notification of what these   are I would encourage you to go and look at the response to the the consultation. Next slide please. So now we're going to look  at the the transitions. Next slide.

So the first statutory  instrument that I talked about   um that was laid is expected to be passed by the 30th of June 2023. What this is going to do   is to extend the period when CE Marked devices can be placed onto the market so there's there's   certain different types of scenarios that you you need to be aware of. If medical devices or   active implantable medical devices have been regulated in Europe to the old directives   assuming they have a current certificate they will be accepted onto the UK Market   with just a CE Mark until June 2028. If devices have been CE Marked through the MDR or the IVDR   or their custom-made devices that have been regulated to the Medical Device Directives,   they will be accepted with a CE Mark on the UK Market until June 2030. Next slide please.

So we talked about um post-market surveillance being implemented ahead of the future   um legislation the second statutory  instrument will put in place these   requirements for post-market surveillance. Under the Medical Device Directive it's   not particularly well defined and all it  says is that manufacturers should maintain   a systematic procedure to review experience gained from devices in the post-production phase   that doesn't give a very clear indication of  what should be included and also it leaves it   very open so manufacturers may do very different things. The legislation that will be laid in the   statutory instruments will aim to to strengthen the requirements for post-market surveillance   um and among other things it will include details of what should be included in a post-market   surveillance plan and post-market surveillance system. So it will be much more prescriptive   it will look at pro-active patient safety  monitoring, increased reporting of adverse   incidents, tighter timelines for reporting advanced events and an indication of the frequency of how   how frequently these reports should be updated. It's expected to apply from mid-2024 so that   will be ahead of the the new regulations that are put in place for the UK. Next slide please. So I've touched very briefly on the regulation of medical devices in in Northern Ireland   and as a result in Northern Ireland the CE  Mark will still apply. Now with the exit of the  

UK from the European Union the situation with notified bodies has changed and in the UK we   have approved bodies that are designated to do the Conformity Assessment. Currently these UK approved   bodies are only able to um approve devices to the Medical Device Directive and apply the UKCA Mark.   There are provisions whereby UK notified boarders can apply for designation to conduct   Conformity Assessment for Northern Ireland to the Medical Device Regulation. If that was   to happen there would be a completely new mark that would indicate that and it's the UKNI Mark. Currently there are no notified bodies designated to do that but the provisions are there should   they should they be appointed in the future. So to date only the CE Mark will apply in Northern  

Ireland. Next slide please. So you may have heard of the The Windsor Agreement that's been recently   negotiated it's a set of solutions to safeguard the supplies of medicine to to Northern Ireland and the UK and European Union have reached agreement to safeguard the supplies into Northern Ireland   unfortunately in this the medical devices do not receive specific attention so whatever was agreed   under this agreement does not apply to to medical devices. So the situation where you will have to   have a European notified body and a CE Mark for Northern Ireland still still in place,   so nothing has changed with that. There's a very clear indication however that should the Windsor   Frame Agreement go well there are provisions that additional um legislation can be brought into   scope and the hope is that medical devices will be added to the future scope. Next slide please.

I talked about approved bodies and these are bodies that are designated to do Conformity   Assessment within the United Kingdom. As of today there are four UK approved bodies   but they're designated to do slightly different things. So currently the only approved body that   can do all um devices, medical devices active implantables, and in vitro diagnostics is BSI.  SJS are designated for medical devices and in vitro diagnostic devices. Decra who's the   most recent approved body to come aboard can do medical devices and UL International can do IVDs. 

If you're going to select an approved body to to regulate devices in in the UK it's   important that you actually go and look at their designation because while we say that Decra have   a designation to do medical devices there are some different categories within medical   devices that they may or may not be able to do um when I last looked which could be out of date I   think it was the end of Q1 I last looked at it, while Decra could do medical devices they   were not designated to do active medical devices so they couldn't do software and they couldn't do   anything that was powered so it's important that you actually look at what they're designated for.   Six organizations have applied, in addition to this, six organizations have applied to   become UK approved bodies and there are more that have expressed an interest so the hope   is that as more and more um approved bodies are designated um the strain that notified   bodies are under, or approved bodies are under um will be eased. Next slide please. So in addition to what's happening in the United Kingdom with respect to notified   bodies and approved bodies being able to certify products um there is a recognition in Europe that   there is a a very real um constraint  on notified bodies being able to   take devices that were previously approved to MDD and to update them with respect to the MDR.   So the transitional periods for compliance with the Medical Device Regulation has been extended   the reason for this is to avoid medical  device shortages and to ensure continuity   of supply without compromising patient  safety or performance of the device.   So manufacturers have been given more time um to comply with the Conformity Assessment of devices   being regulated to Medical Device Regulation as long as they're covered by a certificate issued in   um accordance with the Medical Device Directive and that that they have engaged with a notified   body and have a contract a written contract in place with them for future Conformity Assessment.  Under these conditions, if they  satisfy both of these conditions   Class 1 devices can continue to have the CE Mark and be placed on the European market in   accordance with the Medical Device Directive until 2028. Similarly with Class 2a devices, 

Class 2b and Class 3 devices have  to be in compliance by December 2027. So that is all I was going to say with respect to the the future legislation clearly there's a a lot   of things happening and in line with changes to the regulation, one of the very high profile   issues that's been talked about is the regulation of software in general but in particular how to   regulate artificial intelligence. So I'm now going to pass over to Paul who's going to talk   specifically about artificial intelligence  as a medical device. So over to you Paul. Ok I'm about to hopefully share my screen Okay thanks so much Elaine for an excellent presentation, a great coverage of of the ins   and outs of the the regs there but as you said we're going to touch on artificial   intelligence as a medical device now uh  and yeah I'm going to tell you about about   the the ins and outs of that, the problems, the challenges of that and hopefully   Wrong button sorry folks. So Elaine you mentioned to start there this actually software as a medical   device you know and I'm sort of going to start at software as a medical device before getting into   AI as a medical device. if from the legislation point of view it's all classed as software to  

be honest and it's probably just worth reflecting on the breadth of where software now touches the   healthcare landscape it's pretty enormous it's sort of involved in everything now you know   triaging diagnosis, we've got digital medicines, companion diagnostics, clinical calculators, IVD  software so so software is intricately involved in and all this now in doing great clinical   things and also potentially if it's not done  well creating clinical problems and and error.   So software is really quite a large part of the healthcare estate now and and hence the the   review globally about the regulations round about software and I'm going to go on to the next bit   and again that's just this indication that  artificial intelligence is currently part   of software, a sort of subset of software. I would note though I think we all know, you just have   to look at the newspapers and you'll see that artificial intelligences is becoming ever bigger   depending on how you define artificial  intelligence as well that the size of that   that circle there can be bigger or smaller because even the definition of artificial intelligence is   complicated but worth noting that it is all part of software and we can consider it that way just   to start with and so in terms of starting uh  the name MHRA that the Software Hub is a new   page that we've published and we've got a bunch of sort of guidance documents on it this is one   of the most recent ones this is a sort of start point and and we try really hard to let people   focus on what's the intended purpose of your software, because often it's very nebulous   and it's very hard and if you've got a nebulous purpose it's quite difficult to prove that it   works. So we've published this guidance about how to craft and intend the purpose, Elaine mentioned the   technical file before, this is a core part of your technical file for your product so some guidance in   there. I suppose in summary briefly it focuses on what does your device do, who does it do it for, who   are the people that use it, and and also the user environment or the use environment. So these these   four points at the end there are worth considering uh for your intended purpose but there's more   detailed guidance on on the website there. In terms of all software and particularly AI but all software  

we have to go into the next thing which is that to try and classify the device Elaine mentioned   Class 1 devices or Class 2a devices then there's 2b and 3 so as the classification goes up   that that's in in respect to the risk going  up and there are some notable features about   risk when it comes to software you know at the click of a button you can turn on software to   millions of people that they now have access to the click of a button you could run an analysis on   thousands, tens of thousands of people's you know records and clinical records that could lead to uh   output of information that determines treatment plans for them so the size and the scale   of this is is really quite of note and I think  increasingly we noticed that there have been a   few, if you look back through the press there have been a few notable fairly large-scale IT problems   in healthcare but also outwith healthcare  so when it happens it can it can be big   and most of this non-linearity of of a impact from a change you know again a simple minor change a   bug in a line of codes can can have a very very large effect literally because of a minor change   and a bit of code so so that's quite of note. Two points at the end are worth, human computer   interactions I think that there's a growing  evidence base that we need to be better at   that in fact that that's taken account of in the in the consultation and in our sort of software   roadmap that we're developing to provide guidance. We need to understand not just how the tech work   but how the tech works with the humans in the environment they use it in, the clinicians   and patients that are using it so that's really important and then finally the, Elaine mentioned   about post-market surveillance within that everyone just uses the phrase adverse events   roughly speaking but that it's it's notable that post-market surveillance includes assessing for   indirect harm, indirect harm you know you use information that's fairly pivotable in in a   patient's treatment or care and harm occurs then that would be classed as indirect harm. So software  

is very much in the space of indirect harm but but that is covered by the legislation and and   by what we're trying to provide guidance  around and we recently released a software   specific device uh specific vigilance guidance, again that was just published last week again   on on the Software Hub page, all the references will be in this slide deck for anyone who wants it.  Of course you know all these things bring you risk but they also bring unique benefits so it's just   you know why we're here it's why so many people are using it because there is actually you know   fairly significant potential what can be done to help the whole collective of patients you know   clinicians, health services, they interact between them all so lots of potential benefits there   uh but yeah we have to consider all  those those risks that go with it.   So I've covered sort of software in general and this is a talk about artificial intelligence as a   medical device but hopefully that's positioned it as a subset within it but but how is it different you   know in essence you have an input you develop a product, that product has inputs of information it   goes into the computer, it does a thing and it then it does that to a certain level of performance or   desired or declared outputs you have those outputs and the outputs go to the patient   how's that any different from anything else. Well these are possibly some of the bits that that   are specifically notably different for artificial intelligence and by that is that this is because   you are fundamentally sort of teaching or training or running a model where you're putting in large   data sets to train and create models in the first place even if that if that's a fixed model you're   still running this on large data sets, so it brings a huge challenge around the bias we can go into   this you know the bias of the literature, of the data that's gone into it, the population you've chosen,   the features of that population, is that population the same as the population you're using it in, does   it disadvantage certain character characteristics or certain groups, so bias is a real challenge.   The next one is interpretability uh so we have sort of explainability and transparency they're   subtly different but it's two sort of different things that the way I sort of think about it you   know as a clinician is the transparency element, could I look into the software as such and could   I understand what it's doing so when it's giving me an answer I can I can understand why it gave   me that answer, what were the features that mattered that made a difference and then the   explainability if the patient for example were to ask so you're telling me the the computer says   I have to start this treatment of this medicine now doctor and why, why is the computer   saying that if you know can I explain that to the patient if I can't explain why the the artificial   intelligence has concluded this then that's  a problem and creates a problem of trust.  

Another really unique feature for artificial  intelligence is sort of adaptive adaptivity   or changes in the program, there's a number of features such as sort of drift in the data and   the data analysis that can occur and even just updating the software, updating your data model   how do we do that, these are fairly challenging things to do you're talking about changing live   medical devices so changing devices takes a fair amount of work and and what are the plans that we   can put in place for doing that and and what risks are associated with making those changes. So these   are really they're not the only ones but they're probably some of the biggest and most important   specific issues about why artificial intelligence is sort of over and above software and it's a   bit difficult and as Elaine mentioned earlier you know we're currently on operating under   the MDD Legislation in which at the time it was written you know there was no mention of   artificial intelligence so so we're trying to make all this work and provide guidance along the way.   Hope that helps with that, uh I I mentioned I've said a few things before I just wants just to   see we've got this change program uh roadmap this is this is on the website and it covers   a series of work practices on qualification, classification, pre-market, post-market and and   cyber as well becomes important and there's a remaining three which are specifically on   artificial intelligence so you'll see we've got AI rigor that's all about the the evidence that the   proof that your product works and it does work in an unbiased way and adaptability and adaptivity   so we and we are already delivering on this and intend to continue delivering so if any outputs   we can as and when we can alongside legislation uh by way of guidance so that we can provide as   much information to sort of developers, entities, health services, patients, markets, you know so so   that people can understand where we're going with us and hopefully so far we've actually   had some pretty good positive feedback on on the guidance we've put out of late.   One of those elements of guidance uh that we did is one of the work practices related to I think   a good machine learning practice so if you are in the in the landscape of developing artificial   intelligence or machine learning programs this is a piece of work that the MHRA did with the US FDA   and Health Canada and it produces a start point of here are some principles to focus on if you're   developing machine learning practices and some really really good ones in there you'll see in   there that you know the bottom the performance of the human AI team uh they have to provide users   with clear and essential information, the training data sets should be different from an independent   from the test sets they used in the training and testing is different otherwise you're in a way   of potentially marking the same homework so so this is this is available on the web and and is   a very useful set of principles if you're setting about you know developing a product in this space.   Then again I've mentioned that we have  produced this and what we're calling our Hub page   we are trying to specifically pull because there's so much knowledge, information, documentation in   this space we're trying for the software side of things to pull all together into our Hub   page which we we launched there in April, I was going to say about six weeks ago so hopefully   that's useful and people can find much there and we hope to populate that further as we go along.  

Elaine stole my thunder the the future consultation is a great place to go to to get to get you know   to get the the results of what was concluded from that consultation you can see the government   response you can also really interestingly see what what the the sort of marking from   the respective respondents be the industry or healthcare services so so that's a bit about   software and and AI and the specific challenges, hopefully that that's been useful. If if I may   I'll just touch very briefly on a couple of other things to consider. This landscape of medicine and   healthcare that we're living in is unfortunately is becoming increasingly complicated and and that's   you know dangerous enough concern and it's because we're doing increasingly you know   more advanced things and we're improving health but it's of note, Professor Cyril Chantler   you know and any physician who's involved in in many reviews endless of changes to smoking   so legislation so uh anyone in that position, I just thought it was a useful sort of quote and and worth taking note of   something a bit more detailed than that in this complexity, on the right here you you see this is   an analysis of oncology treatment patients, cancer nurses, research staff, from 2012 to 2015 so even   in that short space of time and that's now eight years ago, this was in Australia, you can see the   growing depth of complexity within elements of sort of complex clinical care and so   we have to recognize we're working on a really complex system particularly for AI, particularly   for software you know you may think you change one thing over here and and that will lead to a   predictable outcome, unfortunately that's not always the case in a complex system and it's   not necessarily always a case in artificial  intelligence as well so we have to be very   aware of that and we also have to be aware that when we roll out any of these products   they go into a system that's not, it's complex but it's not static, the system and the people   round about it adapt and change as well and so it's within that environment that you really have   to make sure your products are shown to work. A couple of other things you know there's now some  

some evidence, this is a paper from last month, I'll not spend too much time on it the long and short of   this one is that this was a review of machine learning devices in the States and it's shown   that you know there are adverse events that are harm that occurs it's often with uh there's a   large scale problem with the data inputs to the machine learning that there's a lot of harm there,   but then also use problems, there's a there's a real challenge for a relatively small number   of issues related to use problems but actually they're more likely to be related to harm so I   just thought it was a useful paper again all the references are there I thought that'd be one more   taking note of for anyone interested. Another point of consideration is that workflow automation so   if you look at this this diagram here we've got the pale blue in the middle of treatment and care   delivery you know that's the sort of the clinical end of healthcare but healthcare is all of these   different parts to running a healthcare system and actually we're sort of taking some of the   most complex technology and we're deploying it in probably the most complicated part of   that healthcare system, straight into clinical care so this is a really useful paper that   indicated that that's not normally what most other industries have done is if they become   an increasingly digitized or particularly using artificial intelligence they tend to approach   you know less complex areas to start with to learn and throughout that technology so administrative   operational you know surveillance so so it  just makes the case that there's lots of other   advantages that can be found across the whole of the healthcare estate   that could use AI and and it would all benefit the whole of the health  care estate but of notes we tend to have jumped in   straight at the most complicated bit in healthcare generally speaking so that's an interesting paper   and the last thing is the evidence, so we have the intended purpose what does your product do but   how can you prove your product does a thing, what evidence have you got for that and the evidence   questions really quite difficult because it depends who you're asking, are you proving it to a regulator, are you  proving it just to get an idea published but  you're not actually proving the product, are you   proving specifically that it's safe, you have to think about the whole life cycle of the product   uh and and of course with all new technologies we notice you introduce a thing, and you will end   up with unintended consequences, things that were not foreseen, so you have to be aware of that and   capture some evidence around that. Again humans, and getting on to the end here we have to   get is it economically viable, to have a valuable product you have to do the sums on this and work   out if putting all this in place, does it make  sense as economically, but is it good enough will   people use it, so I think that evidence thing I'll sort of summarise with my own little algorithm,   I suppose it's somewhat artificial, these are in my head three questions I get asked quite a lot   you know does it work, is it safe, is it useful, and I think anyone embarking on or involved in  products should you know constantly reiterate those those questions - does it work, is it safe,   is it useful, and can you prove that uh and yeah if you can do that then you will create valuable   products so there is a lot of value can be found there but I'm going to pause now I'll come off   screen, I believe there's a question session and I hope that's roughly to time folks. Thank you. Okay thank you very much um Paul for a really interesting um summary there that was very clear,   and we have had a few questions come through in the Q and A that I will put over to both you   and Elaine um but we have only got five minutes or so left I think, we'll do five minutes of Q   and A and then just wrap up the presentation. If anyone does have any other questions please   feel free to send them through to us, and the whole webinar will be made available online   um after this session. So the first question that came in was do Class 1 devices on the UK Market by   virtue of their CE Mark require to apply to the UKCA Mark prior to the 1st of July 2025? um I'll pick up that one um from Ed hi it's nice to see you here   um so CE Marked devices were originally intended to be accepted onto the UK Market until June 2023,   that's now been extended as you indicate to to July 2025. The extensions that we talked about  

um really follow on from what's happening in Europe with the extension to the requirement   to transition to MDR so the only Class 1  devices that this further extension would   apply to would be ones that would require the engagement of a notified body, so with the new   regulation that would be any devices that had been up-classified, that originally had been Class 1,   but now you have a higher classification and require a notified body to assess them or reusable   surgical instruments which is another Class 1 um function that now requires notified body so   to answer your question for Class 1 devices that don't require any notified body engagement they   would have to be your UKCA Marked to  the UK regulations um prior to July 2025. Thank you, thank you very much Elaine um we've got one more question in the Q and A  section right now which we can do but if anyone has any others please feel free to pop them in.  The next question is will there be a  more streamlined way to determine if it   is a medical device and what class it comes under? So again I can I can pick that up and by more streamlined I'm not sure if you mean   what's currently available, or the way the new regulations are set out, there are guidance   documents that are published and currently there's a lot of guidance around the Medical   Device Directive, the new guidance for the Medical Device Regulation and for future UK   regulations are very slowly coming online so to answer your question um there will be um   guidelines on on classification that will set out how you determine what classification your device is   and the way to determine if it's a medical  device is to go to the the definition and they'll   probably be a new clarified definition within the new regulations and if you can answer yes to   any of the sets of criteria that are set out in the definition then it will be a medical device.  There will also be guidelines on borderline  devices which again will will come on stream   as and when they're available and there  will also be um the ability to go and ask   for advice from MHRA if you're really  not sure about borderline medical devices.

Okay that's great thank you Elaine, and we don't have any other questions in the Q and A right now   but as I mentioned earlier please feel free to get in touch um if you have any questions down   the line. In the interest of time I think we'll just carry on just now and so next slide please.   So thank you for joining today's webinar, we have recently started um a webinar series here   at InnoScot Health and we've got two more um coming up in the next few months, the first one is on the   21st of June 2023 um and it's on protecting your Intellectual Property, we'll then take a break for   the month of July and uh previous slide please, then we've got one more in the 30th of August   um which is the Clinical Entrepreneur's  Webinar. So they will all be available   um to register online um at the address  that's given there. Next slide please.   So just to sort of wrap up um on some of our latest news that we've had here in InnoScot Health,   um we actually work with a range of partners and provide regulatory support and consultancy   to them and we're pleased to say that in  April of this year we congratulated our   partner MR CoilTech on being awarded um ISO 13485 accreditation which is the   official certification they gained after being recognised by the British Standards Institution   or BSI, for operating a compliant quality management system and meeting strict guidelines.   MR CoilTech was set up in 2015  um to commercialise radio frequency   coil devices for a new generation  of ultra high field MRI scanners,   they're a Glasgow-based company that work in the Imaging Centre for Excellence the Queen Elizabeth   University Hospital, where there is a CE Marked Seven Tesla Scanner that the company use to   develop coils specially tailored towards the needs of clinicians and researchers around the world. 

We're very privileged to have assisted them with gaining their ISO 13485 accreditation and we are   continuing to support them in the development and commercialisation of their coils. Next slide please. So if you have any um queries you can  visit our website here where you can come   and visit us we have a lot of information  online but you'll also find the details of   everyone who works at InnoScot Health, ways of contacting us and booking a consultation   with one of our Innovation team if  you'd like to do that. Next slide please.   So thank you very much for attending um here at all of our details if you wanted   to contact us via email, if you have any  questions after today's webinar. Equally or  

on Twitter and on LinkedIn, and thank you very much for attending. And thank you especially to   Paul and Elaine for all of their insights and  the information they've shared today, thank you.

2023-05-31 11:40

Show Video

Other news