start here in a couple of minutes uh while we're here I can just kind of give a background of what we're doing where we've been where we're at where we're going so this is an insightful dialogue between Ai influencers and Dr innovators uh we started uh the doctor innovator program about five years ago called dup digital Innovation upskilling program and we had certified hundreds of doctors in this dup program program over the years uh working more on the health care side uh with the doctor uh Focus uh then coming together with our sister company's uh akt Health uh we brought in the life science Pharma uh biotech side they've kind of intersected now uh in the middle as is going on uh everywhere uh that we see even in the in the physical world with Healthcare institutions working closely with uh life science and Pharma uh companies uh so this past year uh we began uh offering the uh Ai influencers and Doctor innovator dialogue we call it insightful dialogue these we wanted a little bit different format we wanted those uh using the the experts sharing their real life experiences so we're right at 8:00 a.m here on the East Coast uh so we will begin with our keynote speaker I'd like to introduce uh principal data scientists and Ai and analytics uh expert Dr Elias Abal Zed from Novo nordis 12 plus years uh experience in digital health and AI Innovation uh Dr Zed leads biomarker analytics uh to advance patient outcomes and Healthcare innovation he's a former associate director of data science at sopi overseeing digital biomarker uh development holds a PhD in biomed engineering from the University of Toronto and advanced degrees uh from Miguel and Concordia University he's a prolific off author p uh patent holder and frequent speaker at healthc care and AI conferences he's passionate uh passionate advocate for patient Centric AI Solutions enjoys soccer hiking and photography in his free time please Dr Elias wearables and and AI real world patient insights for better drug development thank you David for this uh nice introduction um good morning good afternoon all it's a pleasure to be here here today and uh to talk about an interesting topic uh wearables and AI I'm going to start by sharing my screen just a second do you see my screen yes we can great okay now it's full okay great so um I work in Pharma and uh you know one of the big challenges in Pharma and actually before I uh go ahead I want to make sure uh to express this disclaimer that the opinions expressed by myself here are not necessary reflecting the views of Nova Nordisk um yeah so one of the biggest challenges uh of Pharma is basically the high cost for a New Drug development which is estimated about 2.6 billion let alone the cost per patient in a phase three trials is around $50,000 yet we see 70% of these phase three trials fail and mostly due to inability to demonstrate the drug efficacy at this stage typically we go through stage phase one phase two but down really the longitudinal phase is phase three and this is the critical phase so one of the one one of the challenges that actually caused this problem is the fact that the assessments we do for patients in clinic which is has been the gold standard for long time are episodic and and subjective many of those basically um suffer from high interor variability but also the patient themselves when they're coming to visit a clinic uh cognitively their behavior is different than the real world life they're living in and therefore this is not actually reflecting patient Real Performance and that's what we see this big difference um in these assessments however in the last decade or so we started looking more at real world data and this shows to actually better track disease progression and and therapy effects and this is here a a good example a real example from a one-year patient monitoring in terms of it his function his or her functioning and as you can see those red uh bars reflect actually the conventional in clinic assessments and over one year these assessment have been taken every two to three months at the end of the year it's actually hard to understand whether the patient functioning improved um because we can see it's improving and then it's actually declining again however when we look at the continuous measurement of this patient functioning um and this was measured here by a wearable device for more for uh activity we can see although there's a variability from day to day the overall trend it's actually improving and that's what we would be missing if we're just relying on these episodic and Clinic assessment so this show us how real word data and evidence is important and actually wearable and digital devices can collect continuous and objective real world data and that's what we defined and is known as digital biomarkers officially one of the good definition I like on digital biomarkers is the one given by the European medical agency it's objective and quantifiable measures of physiology and behavior collected via digital devices so there's a wide range of wearable and digital devices some could measure physical signals like um motion pressure vibration some other can measure thermal signals like patient fever or other can measure electrophysiological signals U like electrocardiogram muscle movement and there's a different sort of uh type of variables how we can wear it and whatnot from smart glasses to smart watches so on so forth but what has been encouraging is actually the adherence of population and accepting more and more the use of wearable devices and here we can see from 2021 to 2025 the estimate for the US population we actually getting a increase in the percentage of population for their adherence in using wearable devices now one my argue I mean traditionally we've been using real world data in the form of patient reported outcomes which is basically has been a standard of care for real world evidence um and this is basically nothing but questioners filled by the patients themselves can be daily or weekly but they have huge limitations first of all It suffers from subjectivity um and a lot of missing data uh where for example patient you know may not be compliant to filling up every day or every week or even we've seen cases where patient uh forgets filling it and then he comes back to uh he or she comes back to fill it up and then they have this bias uh recall bias so the advantage of digital biomarket is they are objective continuously monitoring and passively collecting data so it's not disturbing patients so in this way we can see how digital biomarkers can actually augment extensively the patient reported outcome as another source of reward evidence here we see a uh two examples of comparing digital biomarkers and patient reported outcomes where in this this study cross-sectional analysis of data from the UK biobank uh the correlation between in clinic measured body fat or a dispos with varable measured physical activity versus self-reported physical activity we can see in the case of the varable measured physical activity there is a linear relation with the in clinic measured body fat whereas for the self-reported physical activity it actually tend to be overestimated by patient themselves where we see it's actually nonlinear and it kind of you know plateaus here where for the several measures of or for a given measure of body fat we see an overestimate in the physical activity uh self-reported physical activity another example is actually um looking at the activity tracking in patient who had undergone hip and knee surgery and this study was done basically after the surgery um all the patient had worn a um wearable wristwatch to track physical activity one cohort known as the feedback cohort U was receiving feedback from the wearable device about the number of steps they are performing daily the other group was blinded from such a feedback both group actually filled a weekly patient reported out outcome questioner about their um their estimate of their their own physical activity and data showed that actually as we go after the surgery uh the cohort or the group that had access to the feedback about their step Counts from the weable device actually achieved a higher level of activity in terms of Step counts um as we go further up to the six months period and this clearly show uh the importance of this feedback from the wearable device as a objective measure of physical activity all of which to say digital biomarkers offer important advantages um to improve drug developments first of all it allows quality of life and real world evidence measure of patients by allowing drug efficacy and safety to be assessed objectively in patients everyday life second it promotes decentralization where now much more assessment can happen in patients own home or everyday life passively reducing the need for excessive in clinic visit and therefore reducing patient burden this automatically lead to improving compliance in clinical trials as well as retention rates of patients last but not least cost reduction with wearable devices which are typically low cost compared to excessive cost of in clinic assess assessment we are able to collect much more data in shorter period of time so that reduce first of all the cost of data collection and allow Pharma companies to make faster and better decision for example in accessing a drug and decide whether to go from a phase two to a phase three trials and typically phase three trials are much more expensive trials now as you see and as you might be thinking I mean these varable devic are collecting data continuously this much be must be a huge amount of data and in fact it is it is a big type of data um and one can think how we can take advantage of all this uh huge and big amount of data and AI has been a instrumental approach to actually transform this big data into meaningful Health measures so here's basically how it typically Works uh with these wearable and digital devices we're collecting continuous passive data at home all this data it's typically transferred uh via Bluetooth and and wireless to the um sponsor or device vendor Cloud where analysis can be performed in the form of signal data processing but also artificial intelligence like machine learning or even the more advanced deep learning even nowadays we start looking into applying generative Ai and large language models and all this lead to derive important measures that could be like end points for example sleep measures physical activity Vital sign like blood pressure heart rate heart rate variability and even the the way a patient is walking or his walking is improved and all this measures can be validated in a clinical trial to become primary or secondary end points thus these are now digital endpoints that can be used later as drug efficacy and safety endpoints and of course another um another form of using this it can evolve to become a digital Therapeutics or software as medical device that can be also used outside of the drug development phase as a as a a companion app for patients now we saw how this framework work let's uh walk through a couple of examples walking speed estimation is actually one of the important measure specifically in neuromuscular disease like multiple sclerosis and even relevant in um you know patient who undergo surgeries for their knee or any lower limb surgery typically walking speed or gate um speed or gate even analysis estimation happens during a clinic visit where patient have to commute to a clinic and a physician would perform a plora of assess assessment um physical activity assessment and one of the most used one is the six-minute walk test where patient are asked to walk for six minute and then a physician would look at how many steps they were able to perform or how much duration they were able to cover in this six-minute walk test to actually estimate their walking speed however we can actually assess this at home or in the patient real life by using a wearable device for example a accelerometry based device like the one shown in a picture which happen to be worn here on the um on a belt on the waist and so this device is actually measuring the body 3 axis acceleration the body XYZ acceleration measuring this row signal which is um actually a large amount of data and then these row signal can be um processed through a traditional machine learning approach where we actually look at feature EXT raction and then applying a model like linear regression with some labeled data we can train the model to estimate walking speed another approach it's actually to use in more advanced deep learning where now we don't have to um extract features but actually feed in the whole row data and let the model train to optimize finding patterns in this row data that actually allow us to have an accurate walking speed estimation and again here you would need a set of label data so my team and I actually uh did a work on uh comparing these two models on slow walking elderly data from the device I showed earlier and we demonstrated with a deep learning approach we actually can achieve a better estimate of uh row walking speed of patients both in terms of reducing the estimate error and aligning uh into the um basically estimating the the the whole range of walking speed um and so with this deep learning model we showed that it can actually have a better estimate overall of the walking speed now one can say okay I mean deep learning model do perform better but they need also a large amount of data and this is actually a curse for clinical trials because collecting labeled data is a complex and costy approach yet there is also another approach in AI that can actually lower dependency on this labeled expensive data this is what is known as self-supervised learning how it works it's basically as follows we can train a general AI model by using a large set of generic unlabel data and here's an example for um estimating human activity which is a very important Endo in a lot of U disease areas when we are uh developing drug for example for mytina gravis where we would like to know how much time of the day a patient um is spending being laying down or sitting versus being active or standing and this approach actually takes a large set of generic unlabeled accelerometry data which is easy to find there's a lot of Open Source data sets really very unexpensive we don't need any enables we're just generally training a model called the encoder here um to be like you know estimate certain um self-supervised task and then we transfer learn this um primary model to our specific task but here we don't need to retrain the encoder we freeze it and we only train a few layers of uh deep learning and usually they only require a small set of uh labeled data and like that I'm actually removing the requirement for this large amount of label data where I still able to achieve actually even better than supervised learning when I have a small amount of labeled data so this way now I still have a good performing deep learning model that doesn't require a large amount of data and this is very beneficial in clinical trials for drug development in the last two years or so we all heard of generative AI especially with the um with the introduction of chat GPT generative AI is a type of AI that generates a new data based on patterns it learned from existing data two popular uh generative AI are the large language models in chat GPD and jimin and these are actually text based generative AI interfaces here we see a quick example I uh work through to actually show you how it works you can actually prompt a text and then it will actually um generate based on your text whatever you're looking for for example in this case I ask a um text based large language model that generates images to generate an image for someone holding a sign saying generative AI this is by the way a free open- source um website you can um play with it was okay although it did a mistake in U actually spelling generative but still it's something impressive now how is generative AI being being applied to um digital Health and Drug development um and recently there was some work published about what we call Health large language model where we're actually taking a large language model but now we can um input a multimodal data it can be wearable data electronic health data or any type of data about the patient and all this is entered as text to the large language model and then the large language model can give you estimate about several U health condition of the patient based on this different data sources as input here we see an example where actually um out of the Shelf existing large language model like gpt3 and four um were actually tested versus a completely fine-tuned Health specific uh small model of uh um of basically generative Ai and impress ly we could see that actually the existing model can still perform as good as a fine-tune model although we haven't fine-tuned those existing large language model so this is something impressive and encouraging that we could use of the Shelf model give it some input about the patient Health from multimodal um data and it still can do good prediction so we talked about digital biomarkers how AI is helping um giving important insights but how does the success perspective of digital biomarkers look like it's basically seen in evolving from an exploratory to a primary or secondary endpoint and again this is still an emerging experimental field however there has been few successes one of the earliest and the first success is the European medical agency approval of the first ever digital endpoint this started in 2019 where Emma approved the first secondary digital endpoint dried velocity 95 percentile in desant muscle dhy where actually after a lot of work it was demonstrated that this um wearable at home measure actually highly correlate with the incl clinic six-minute walk test test and further validation of this measure led the Emma in 2023 to approve this measure as a first primary digital endpoint in duchant muscular dropy there's a big advantages sure yeah I'm almost done there's a big advantages of this measure uh compared to the alternative in clinic six minute walk distance it allows to have continuous monitoring in home sittings and it also know it's highly correlated with a six-minute walk test but it's still more sensitive because it's continuous and can detect all the variability all of which to say um in this talk digital biomarkers can provide objective and continuous real world patient data for better assessment of drug treatment efficacy or safety AI can transform big patient Digital Data into meaningful Health Care measure and of course to scale further digital biomarkers in clinical trial and the health Healthcare sector in overall we need more collaboration between device vendors sponsors Regulatory and Healthcare Providers and Physicians this conclude my talk I'll be happy to answer your questions thank you wonderful thank you Elias that was uh extremely insightful we used to say I think it was 20 20 30 years ago I guess they've been saying it a while if you can't measure it you can't manage it so I I love this so now we're going to move to our insightful uh dialogue here with our doctor innovators and then at the end uh we will have participants uh ask their questions so feel free to put those in the chat uh for after the uh insightful dialogue here so I'll let you read the wonderful backgrounds here of our Dr innovators as they go ahead and ask their uh questions in the interested Time Dr Dominic sofa please your questions for Dr Elias thank you and my appreciations Dr Elias for wonderful presentation quite um insightful quite educa um yeah I have a couple of questions um arising from the topic um indeed it's evident that we medic devices are very very important and long way in in helping in as regards to drug development and and other in managing patient General however I am wonder where you practice are there any regulatory requirements for these uh weable medical devices I asked this question in the light of um the fact that uh you know the world is a global vill is often been said worldwide internet I mean information can move from throughout the world matter of minutes second so and I'm aware that when transfer of data across countries ACR continent uh sometimes there's requirements for instance for like medical transfer agreement across count so I'm thinking what this are there any reg Frameworks that exist surrounding the use of weable medical devices especially when it involves data being transferred for one part of part in for processing analysis and all that that's my first question um the second question also maybe Dr s can we uh have uh Dr Elias answer the first one and then come back for the followup thank you thank you yeah thank you um sorry I might have missed because the quality of the of the sound in it wasn't good for me but I think you're asking about whether there is some regulation for these digital biomarkers in terms of you know data transfer and um record protection whatnot right yes yeah uh yeah that's a very good question and and in fact there is um although it's still emerging um FDA and B also as well Emma has been working in the last few years and developing guidelines and the regulation for uh digital biomarker and overall they call it digital Health Technologies and just about one year ago actually in um December 2023 the FDA published their final guidelines for fit forp purpose digital Health Technologies and it does cover basically a lot of uh areas including the one you were mentioning um data privacy data protection and how um basically the data should be transferred in a compliant um platform endtoend compliant platform but it also covers basically the selection of these devices how it should be um you know suitable for the uh clinical population also it has to be well verified and validated um and also what are the risk and consideration which are important when we are considering uh patients okay uh thank thank you very I think we have four minutes um I don't know if you can go ahead second okay I'll just quickly just one um yes well the okay I have I just pick one of them regarding patient adherance patient engag medical weable devices um I I know it improves it improves but however at the same time you have to ensure the patient wears as at all times whenever he supposed or she or she supposed to so I know that creates another problem how do we ensure that patients always um wear this devices yeah that's that's another good question there's actually multiple factors to that um first of all the design of the wearable device um you know has to be patient Centric it's not bulky it's comfortable to wear and we've seen there's a lot of improvement over the years where if you compare for example a medical grade wearable device from five years ago to today's medical grade devices you can see actually how uh the aesthetic and how the Comfort has improved but also I think another important factor is to actually engage the patient um typically we in clinical trials we don't want to have biases so usually we don't show the feedback from the device to patients but this doesn't mean we cannot explain to the patient what we're doing and what's the importance of that and the third element and I think that's the most uh crucial is to actually use um Technologies like push notification on the patient phone or um the phone that the patient is uniform click and trial to send him a reminder uh of using these devices and we see combining these three factors we actually tend to get um High uh compliance rate in wearing these devices okay thank you I presume my time is up yes I think that is right about time thank you wonderful questions thank you insightful dialogue we appreciate that Dr sofa let's move to our second of third Dr innovators Dr Barbara Mado I'll let you read uh her background and experiences as Dr Mado asks her question hi David hi Dr elas thank thanks for the great contribution it was really great presentation in fact you have already answered two of my questions during the presentation which was great uh I appreciate that you have made it clear um the definition of biomarkers because in the beginning I thought biomarker would be pretty much related to genomics and that kind of thing and you made that really clear it made you made it easier for me to understand it was quite impressing that you said that the biggest challenge uh would be the cost of $2.6 billion dollar per drug which is a huge amount of money and you mentioned that 70% of the trials uh fail failure uh fail phase three right uh if I'm not mistaken uh how much do you think that these digital biomarkers can reduce these 70% uh clinical trial fa failure yeah uh and actually that's a very good question um I think there's two ways um digital biomarker can help um and actually it's more to preemptively understand whether the drug is working well already in phase two so typically phase two studies tend to be smaller and much less costy and and relying just on in clinic assessment may actually be deceiving because you're not actually collecting a lot of data in that short period of time this is where um actually wearable device because it's collecting continuous data allows you to see much more about the measure in in this patient in phase two and it can tell you that actually the drug is not working properly so you already can make the decision not to move to the most expensive phase three the other way these drug actually can help is because you're collecting a lot of more data in shorter period of time you actually don't need a large sample of patient in the trial and the period of the phase three trial can be reduced so already combining this less patient shorter duration of period of phase three trial actually reduce the cost overall perfect sounds great and uh it explains a lot in one thing that you have mentioned at the end was the collaboration between sponsors people that produce but honestly my question is from the beginning because what we see is that sometimes uh we call translation translational medicine is when you have a finding inside a university and this does not gain the market and does not gain the clinical practice and what strategies have you found a most effective in fering collaboration between the engineers data scientists clinicians because for me it seems that is one of the biggest challenge see is see what's happening clinically speaking to what has been developed how do you see this how this uh to Foster this collaboration yeah that that's a great question um I think over the years what we've seen um actually tend to be working the best um is uh there's two type of collaboration that needs to happen um and I'm talking from the farm perspective internally within the Pharma itself within the company uh we need to make sure like all the different uh stakeholders um are aligned from the beginning for example um clinicians uh regulatory within the Pharma company data scientist Engineers they all should be participating from the beginning um because when you talk together each from their own perspective can tell you what's important what's feasible what's not feasible but also beyond that external collaboration is also very important especially with regulatory at the end of the day for these um digital biomarkers to be used as primary secondary they need FDA or regulatory approval like Emma or whatnot so this is another important thing and I think what's also very important that we're realizing more and more digital biomarkers should not be a competition um Pharma can compete on The Compound on the drug but I think digital biomarket should be a pre-competitive effort because there need to be a consensus U for regulatory to approve a measure you don't want to end up with each Pharma developing their own measure specifically to them that's not gonna fly regulatory will not approve a in Silo measures sounds perfect so if you try to unify the information is going to be easier for the regulatory organs to approve it that's a great answer for me that's a great answer for me and as a final question if I have uh any time left um you're a leader in digital health and where do you see the future of AI and where B heading in the next five to 10 years because it seems to me that it's game field uh space and how do you see that for especially for drug development uh how much do you think that it can really hurry because you know that sometimes you can take five up to 10 years to develop something new how do you see that in dis yeah um so I might not be able to talk specifically of nov NIS because I'm not clear to do so but I think what I can say is generally going to be the trend um this field right now it actually start moving from being experimental to emerging especially with the approval of couple of uh primary endpoint like the one I mentioned uh but I think the the future combining Ai and weable together will have an immense potential to Revolution ize how we approach Health Management diagnosis and treatment and this I see it in the next 5 to 10 years for example one uh particular area that I think it will see um this impact is uh personalized medicine and predictive Health where now with wearables AI we can drive personalization of uh drug development uh giving the the right drug the right dose at the right time to the patient but also uh in terms of continuous monitoring of the chronic disease management and another way is to also integrating multimodal data for a holistic health view of a patient um and um for example also real-time clinical decision support where now with this weables we can continuously monitor uh in real time what's happening with the patient so I think these are the three pillars that I would see uh would be um you know a big impact that Ai and warbles will have on in the next 5 to 10 years thank you Dr Elias fantastic fantastic presentation as well thank you thank you yes is it is fantastic thank you very much uh Dr Barbara Mado thank you Elias thank you thank you Dr Elias Z now let's move to our third and final Dr innovator and again you can put your questions in the chat uh after our third innovator here I'll let you read uh Dr suan Tunda uh ol oluk Cuba's uh exemplary background as he asked his question please Dr hello everyone yes yeah uh good afternoon uh thank you very much Dr Elias for the insightful um presentation on variables and uh AI in drug development was quite uh insightful uh mine is going to be quick as some of my questions have already been addressed in the course of your presentation as well as the question and answer um session yeah just uh like you mentioned the uh variables the digital biomarkers actually have more objective um data collection as compared to the use of questionnaires that it has some form of bias um and the issue of uh adherence uh as I know some of these devices have to be worn uh on daily basis like virtually all true uh that has also been addressed so I'm just going to T my question to one very important aspect to uh the low and middle income country like Nigeria it has to do with infrastructural gaps uh electricity internet access access these are things that I know these devices will need constantly so I don't know in areas where have issues with uh electricity not being too reliable and uh some form of fluctuation in Internet and sometimes even uh significant downtime in Internet are there ways in which this can be addressed in order to make the varable technology viable uh in this uh region something like maybe um uh degreed collection of data to be uploaded whenever the internet is available something like that it's available yeah and in fact uh that's a important point and typically this is being taken into consideration specifically in the um in the new generation of wearable devices because yes um we came to conclusion a few years back that uh sometimes the environment of these where the variables are being used it's actually different from the um from the one that was sought at consumption time and actually the the the most um you know the latest wearable devices some of them actually um are made for such environments and this is where it's important to look at these aspects at the selection of the variables so a lot of these variables actually have um a memory on chip so they can collect data for a months for example without needing to upload and what will happen is after you can take that variable to the clinic and they connect it to a computer with a USB um cable and then download the data and also typically those wearable devices are made to be very efficient in terms of uh power consumption so some of it actually um don't need um recharging for a for one month so um uh again there's a lot of um of these features available in some devices and this will become important When selecting a device to keep into consideration um the environment it's going to be operated in oh excellent that's very good to know uh then lastly I don't know what roles should International uh collaborations play in order to harmonize the regulatory standard support the global scalability of these wearable devices yeah no that's a very good question and yes I mean um to be to tell you what's uh this the um how the environment it looks like now the two biggest regulatory players in the field has been FDA and Emma but actually over the last few years we started seeing more of sort of consortiums where um regulatory from the different countries are coming together to discuss because yes uh it's an emerging field and we need to have some standardization um because typically right now what happens is um if you want to develop an endpoint uh um you have to get approval from FDA from Emma and from the different countries you want to use it and they tend to have a little bit of differences although in general what they require is similar but there's still some differences and I think it's going to be important going forward uh to see this unification of Standards like we're seeing for example now in AI standards and uh we saw this act of uh in the European Union about AI Act and the same in us and I think this will become more Global and it will go towards uh unification so that you know we have one set of standards that fit all the countries and the needs oh yeah that's excellent thank you very much fabulous questions very important uh discussion thank you very much uh let's go to our uh participants if you have just feel free to um chime in or raise your hand or put into the uh chat and I was also some things came to mind we talked about uh different types of computing there's uh I was going to share something about the cloud fog mist and Edge here and also I shared some about the the starlink and apple direct to satellite uh Technologies now we have a question uh coming from uh Brazil Dr Leonardo aguar uh as an Innovative uh physician how do you see digital therapies and where able devices transforming patient monitoring and treatment and what opportunities and challenges do you foresee in effectively integrating the Technologies into your uh clinical practice can you share some examples of situation where the technology uh could bring significant value to Patient Care yeah um I mean I I think the advantage these uh technology is bringing is um is visible and huge so we're talking now going to continuous monitoring um realtime uh patient monitoring so you know doctor can know exactly what's happening with the patient um minute to minute of course this will come with challenges first of all integrating with uh hospitals and Clinic um we need the suitable infrastructure um we need to ensure also uh regulations you know this is Health Data so uh communication need to be secure data privacy should be ured so um a lot of these need to be addressed but I think already um we're seeing more and more um of these standardizations and realizations of such of systems and infrastructure especially with the revolution with the uh internet and uh infrastructure and also what you mentioned for example starlink that we're seeing it more and more deployed in even remote areas so this will allow uh integration of this on the on a global scale hey great uh great answer great question uh let's move we have uh one more here uh Robert ya from uh I believe it's Robert Yao in Singapore correct good afternoon I was slightly late so I'm not sure if there was discussion about clinical data perhaps exploring this method for data collection uh could be of interest and he has a link here online integration of quality focused data strategies in uh clinical trials um yeah sorry I'm I missed the the question so is it about the quality of the data from yeah see uh is there see about perhaps data collection let me see here there's a link let me take a look at the link uh online integration of quality focused data strategies into clinical workflows uh okay on how reg uh regular clinical workflows look like in participating site and how overburdened uh the medical person yes yeah yeah um I mean typically um you know with variables and and AI it's basically yes there is a lot of data um but also there is architecture behind how this data is being collected um so one of the benefit basically is that now you're collecting data passively which means patient is just living about their life and data is collected they even don't notice that and and also of course um there's not a lot for uh the clinician to do in that sense because this data is transferred and it's automatically generated so um the clinician don't have to spend time looking at the data what is the interest for the clinician is the end measures that the algorithms are generating based on this big data I hope I answered the question because I wasn't sure about what exactly is the the point here but I think uh yeah in terms of the infrastructure we have and the way the framework Works um it's not adding burden to clinician okay we have another question coming in here uh local here in the Boston area Dr Kimberly o Solin I have found that despite cardiac pacemakers being continuously monitored uh monitored the monitoring the alerts of the the alerts are delayed clients will be notified that the device is offline Weeks Later what's being done to optimize immediate SE of alerts uh to all parties involved with uh Alto allow for immediate response without paying for the service um yeah I'm not sure I don't have details about this service but I know from our clinical trials where we monitored for example um real time um Vital Signs including something related to cardiovascular um and that actually was during covid because it was hard to keep the patient in uh in hospitals um devices nowadays and their U the the the software and the system that comes with it actually allows you to keep track of uh the status of the device whether um the barrier is dep pleading or whether the device is online or offline um we're able to know that and alert the patient uh or send someone for intervention okay excellent thank you I think uh we have further clarification uh Robert uh had sent I believe the data he was referring to was uh he mentions from notes things like that I think Robert's referring to unstructured uh data here I see yeah yeah yeah I mean wearable data is considered unstructured data but what I want to say is the more and more we see with the AI now we're actually able to combine multimodality combining both structured and unstructured data and we're seeing really amazing results where now we're getting insights from different modalities excellent thank you any other questions feel free to uh you can either unmute or raise your hand I've got it I've left it open or feel free to write uh any questions uh in the chat we're opening up now to all uh participants who are here so anyone feel free again we will be sharing we have the udle AI bot here uh so we'll have the transcript we'll have the video we'll also upload to uh to YouTube uh and we will continue uh the discussion uh after if any questions do come to mind uh please send those over to us and we'll do our best to get them answered Dr EA did I see you unmuted there do you have a question okay any any other questions okay if there's no further questions again we'll be sharing the video the transcript uh feel free to send uh your questions uh F's going to uh share a little bit here on her screen okay it's up now the a I just wanted to kind of give a year in review here at the end of the year uh from our AI dialogue webinar series that began uh back in July and uh just wanted to recognize our Keynotes and topics here all the way through now uh to the uh the end of the year and we'll be continuing into 2025 key features you can see here insights from Top AI experts and Doctor innovators topics have included Ai and healthc Care data sharing Precision medicine surgery Women's Health chat vots Foundation models and wearable tech we've engaged 70 registrants and 40 participants uh per session uh reaching thousands uh via LinkedIn and email if you look at you know Impressions we have digital influencers and we're across a dozen different networks you know maybe in the thousands to to T tens of thousands of Impressions we again we work globally we've seen uh we have doctor innovators and others from all over the world we work in the Middle East uh we work in Africa Latin America so very Global uh diverse uh group uh we welcome other doctor innovators if you have friends colleagues that would like to join us uh in the future feel free the impact here is you know we're really fostering collaboration uh around the world it's a knowledge exchange those sharing their real life experiences uh Innovation at the intersection of AI and healthc care so thank you uh everyone for joining us we hope to see you back uh here in 2025 as we continue and fion is sharing the various sessions we've had throughout the year so really thank you everyone for being a part of this journey thank you Dr Elias for keynoting thank you our Dr innovators and participants have a wonderful rest of your morning afternoon or evening thank you bye bye take care bye bye thank you very much thank you