#249 Unveiling the Future of Personalized Medicine with Real-World Data and AI: A Discussion with Nadia Lipunova

0:00:02 - Mehmet
Hello and welcome back to a new episode of the CTO show with Mehmet. Today I'm very pleased to have with me from London the UK Nadia. Thank you very much for being on the show. The way I love to do it is I keep it to my guests to introduce themselves, so the floor is yours. 

0:00:17 - Nadia
Thank you, that's very kind. Hi everyone, my name is Nadia. I'm an epidemiologist by training, so my background is spent across sectors of government, academia and now industry. I worked across multiple diseases in real-world data and real-world evidence. Now I work at HOLMUSK and work in the field of psychiatry. 

0:00:39 - Mehmet
Great. Thank you again for being here with me today, nadia. I'm feeling that people want to know about the work you do. It's a tech show and we discuss different aspects, but I get excited every time we talk about something that touches health. So this is why I get excited about today, and the first thing I would love to ask you is to give us an overview about the two things you just mentioned, which is real-world data and real-world evidence. So if you can tell us a little bit about these two terms an overview and maybe later on you can explain how they are impacting the healthcare industry, Absolutely. 

0:01:28 - Nadia
I love talking about real-world data and real-world evidence. So this is my prime time. Real-world data overall and evidence are data sources and any analysis streaming from that that are generated outside of clinical trials. So these are all data generated, let's say, when you go to a doctor and you're not part of the clinical trial, but certain items of information get recorded and sometimes they're used for, let's say, population level screening or population level statistics. You know, just a simple question of answering how many people do have cardiovascular diseases. It's impossible without these, say without real-world data. So this is a key item that has been used for a very long time but it has never entered really individual level statistics or trying to foresee what, generate hypotheses or confirm hypotheses to such degree as it is now. 

And this is a field that has grown substantially in the last 10 to 15 years, primarily due to better recording of health data. Before, let's say, 20 years ago, we had very broad statistics about health states and health risks and symptoms on populations. Now, with electronic health records or EHRs, we have far more. Now we can actually try and carry out you know, cholesterol levels and how does that impact cardiovascular risk or what kind of symptoms? Let's say, a patient with schizophrenia present to a hospital, and how does that impact their response to treatment? So it gives a lot more granularity, which is really important for us. And as we enter well we have entered, I guess the area of personalized medicine and the goal of personalized medicine, these aspects are incredibly important because we know that every patient is different and there's actually a saying that I think was still from Hippocrates, that it's much more important to know a person that has a disease than to know the disease and then go back to it. So obviously it's important to know both, but individuals differ a lot and there are certain things that we need to know on a population level. Now why, overall, this field is important is clinical trials are exceptionally important in establishing efficacy, and this is still the gold standard and which I think will be for a while. 

But we don't know what happens to patients outside of clinical trials. Clinical trials are designed to select patients to eliminate any kind of confounding or any kind of reason they think right, this is probably a fact that is not due to the drug, but something else. So you don't want to have that ambiguity. But by default they select very few patients. So a lot of patients, or most of the patients, never go into clinical trial, whatever the therapeutic area is, and by default they don't get represented. So in schizophrenia, for example, that could be up to 80% of patients are not represent, not eligible for clinical trials, which means we don't know how to treat those patients, and then response rates are not the same that we see in clinical trials, hence why it's really important. So it is adding a lot of real-world data and evidence to the understanding of those patients' experiences, to the current knowledge. So it's not aiming to replace clinical trials by any means, but it's aiming to add more information about what happens outside of that setting. And how do we sometimes better inform clinical trials? 

0:04:51 - Mehmet
That's great explanation, nadia, and thank you for explaining that to us Now, out of curiosity, because I spoke to a couple of executives who work with mainly companies that they focus on the clinical trials and we know where the data comes from. Now, in your situation, when we talk about real-world data, real-world evidence now, are there some, let's say, hospitals, clinics that enroll with these programs so you can collect these data? How does this work? 

0:05:27 - Nadia
That's a very good question and that's one of the more important questions is where do the data come from? Because that already shows what is the validity of these data items and which degree we can use them. The most common setting for these data to come from healthcare provider is usually a hospital or secondary care, more specialized care, because they have a lot of data in-house and that is mainly captured by electronic health records. So a while ago mentioned 20 years ago a lot of things were on paper so there's not much of a trail that you can really pick up and turn into a patient trajectory. Now, with tech and electronic capture, we can actually follow up a patient in their presentation, diagnosis, follow-up, prognosis etc. So that makes it possible and it's usually most accurate, most well populated in a hospital setting. The data can also come from primary care setting or let's say, gps. That capture a bit more broader population and it usually represents a setting where diseases are not serious enough to go into the hospital or they can be managed in a primary care setting. 

Recording of those information really varies within and between populations. Some countries or some healthcare providers have it very well recorded and some really don't. So it's a bit more uneven than, let's say, secondary care, but usually from hospitals. Now when we say hospitals, we always assume and this shoe in a lot of therapeutic areas that all of these data are generated by clinical personnel. Whoever assesses you puts in, let's say, suggested diagnosis, then it gets coded, but it's coming from a clinical staff assessment In psychiatry specifically and some therapeutic areas, there is a lot of room for patient reported outcomes that were not that well captured a while ago. They're still not captured good enough, but they are becoming a very important aspect because how the patient feels and that can stem from pain, how much pain does the patient feel, how do they perceive their quality of life, how can they engage in life? Patient reported is also something that gets sometimes captured in the system. So clinically informed patient reported outcomes and sometimes there are surveys as well that combine the two, but most frequently a hospital. 

0:07:51 - Mehmet
Okay, that's great. Now you mentioned psychiatry. From the normal medicine point of view, I can understand and you explained it at the beginning about the trials and the way we do it. So maybe you have a disease and then they start to do some experiments and then they extract data and so on, but in psychiatry, what is the gap that actually you're trying to fill here? So what is missing, I would say, which you are trying to fix it out. 

0:08:26 - Nadia
That's a great question. I think you already picked up on one of the most important gaps and the difficulties in psychiatry is that we don't have biomarkers. We just don't have an equivalent, let's say, to measuring cholesterol or measuring a biomarker. And I'm not saying that it is very easy in other clinical areas or more physical diseases. But psychiatry lacks a lot of those just because we don't know. So it becomes a much more difficult entity to tackle. So one is just the lack of physical biomarkers that would be objective for us to judge the state. 

There are attempts but they're not consistently used and because of that subjectivity it bleeds into the data that we work with. So a lot of the things we need to always consider the subjectivity at input, at source, and that it may or may not be completely true to reality. So, say, patient reported outcomes are very frequently used. An example for limitation that it is subjective, it is something that is not a biomarker and that's across all therapeutic areas. But this is our in psychiatry it's one of the most important information sources. So we have to bear the uncertainty in data recording and we have to close a gap of just not having biomarkers. So we need to do quite deep into data to have the closest comparator. 

0:09:47 - Mehmet
Great, you know, like this is very I'm trying to think about it very useful, I would say, because by doing this, you can have, like let's call it more labels, and then you know, this will help later on to get this. And there is I'm trying to mention this after you so for people to understand, like, because data nowadays is everything, and this is why you need to have good classifications of the data, and I think this is very important, especially in psychiatry, because it's, like, as you said, it's very broad. You know it's not like you do, as you said, the block test or something like this. Now I'm interested to know, you know, to you know when we get this data. So I'm sure, like there are, like in the background, some engineering solutions that they play a pivotal role in managing this data. So can you give me some examples on how you know this data that comes? You know what happens next, like, what are you know the things that can happen to start using this information? 

0:10:51 - Nadia
I really love this question because this is a very commonly omitted aspect of real world data overall and use of it, because there's a lot of steps and there's a lot of plumbing you can call in getting these data into any kind of analysis. That is extensive work on governance and that's very qualitative aspect, but also bear some quantitative metrics that we need to employ on data quality, on how do we use this data. A lot of data that we, into user de identified, and that's a science and a work stream on its own, because we need to have good solutions that are scalable, that are effective to de identify data, and the acceptance of error in this area is incredibly low. We cannot risk that. So you need to go so high in terms of your confidence that we can scaleably and well de identify data that is taken. 

A lot of tech solutions on its own, then, usually working with multiple healthcare providers, we need to standardize the data and pull into one big data set, which entails quite a few things A we need to work on the structure. We need to also work on the values, standardize those, make sure that we analyze each data set as part of a broader population. That takes a lot of finicky work, but it's really important because our analyses are, because it's healthcare. We are very aware and we would like to avoid as much as possible any bias that we introduce in the analysis and if there's an effect based on one healthcare center, then all of our analysis will be valid. So we need to take great care into data preparation, and data engineering is probably one of the most important teams in real-world data and real-world evidence. 

Overall, it gets sometimes an unsung hero. I think and there's a rule of thumb that 90% of the work is data prep and 10% of the work is the analysis. It's relatively easy to work to run a CNN or a regression analysis on a clean data set. It's very hard to get it to that point. So there are multiple steps the identification, which is a tech solution on its own, then it's data management, data engineering, and then finally we go into analysis. So we have to have tools appropriate for each stage. 

0:13:06 - Mehmet
Yeah, and I know from not only in healthcare the easiest part is the presentation of the data at the end, but the hard works happen. So I always tell people, when you see these fancy graphs and these fancy interpretations of any kind of data, I say this is the top of the iceberg. There's a lot of work that happens in the background and data engineers they work hard to clean this data. And, by the way, it's funny or nice to know the methodologies are the same but the verticals are different. So, for example, we're talking about healthcare, but it applies the same thing for any vertical. Now, because here you're dealing with different type of data. So here we're dealing with, maybe, someone who wrote a note about this patient, what the symptoms they saw. So I'm sure like and I'm going a little bit geeky here, I would say I'm sure like you would be using a lot of natural language, processing NLP, to leverage this data, am I right? 

0:14:06 - Nadia
Oh, very much. So A lot, a lot of NLP in healthcare and, I have to say, in psychiatry especially so, because, going back to a few points that we mentioned, the data that we work with are highly subjective. I mean, don't have that many lab tests that are really useful for psychiatry, so we do rely on clinical notes that can be semi-structured, unstructured completely to a very high degree, and that applies to symptoms, to the diagnosis itself, because sometimes it's not that well recorded in the code of the maith and we need to cross-reference that. So anything and that relates to current experience, whatever patient is experiencing at this very moment, and also in terms of their history, because that's very important when we consider their real estate response to treatment. We want to know what happened in the past because that influences what happens after that. 

So we rely on NLP quite a bit and the NLP approaches of course differ to high variety. Sometimes we use a very quick, you know you can do a reg X on some of the data items that you need and that will be satisfactory. And sometimes you need to go into a full model development. That is quite laborious, computationally resource heavy and requiring, but that is the only way to extract the information, curate the information that we would need. So NLP plays a very significant role, and I have to say that I think this is one of the approaches that will be as prevalent as it is today, if not more. 

0:15:32 - Mehmet
Yeah, and the reason I asked this? Just I like to explain sometimes the reason I asked the question because so people tend to think usually and you know I'm targeting because I surprisingly, although the city will show I have people who are not from the background but they think you know computers, you know all that they do is like analyze the zeros and ones and you know the numbers. And the reason I asked about NLP because you know natural language processing is what allows a computer to take something that we wrote or we said. Maybe sometime you can transcribe this into something that a computer can analyze later on. And this is will bring me to ask you, nadia, about the buzzword AI, like what role AI plays in you know real world data and real world evidence, like how you leverage it for getting better results. 

0:16:29 - Nadia
That's a very good question as well. I have to say, the adoptance currently is patchy. We coming from and I'm not a doctor for full disclosure but from the medical field, of community and working in medical sciences we are skeptical to a very high degree, but also optimistic about the potential because there are so many things that we see could benefit from data capture. You know, patient being, let's say, or clinician examining a patient, recording the notes that can be automated to a high degree and extracting, let's say, key items at the source. There is high potential Helping out with some of the more administrational tasks that play a significant role in healthcare, because data recording versus what happens in practice to could be slightly two different things. So that can be automated to a very high degree and AI specifically can play a massive role because it would add not only the automation but also the interpretation to certain degree. Then we're talking about clinical decision support, which I think gets most of the hype and most of the promise from, especially, population, and I think a lot of people who watch healthcare and AI is hoping that AI will help to decide what to do with the patient's trajectory, what kind of drug to give, what is an optimal drug to give what is an optimal dose, and that's a very difficult field. So we are very skeptical again because the error patients for error is very, very low. We can't risk it because at the end of the day it is a patient and something goes wrong. That is a penalty that nobody wants to take, nor they should. So, skeptical but optimistic. And then, finally, we go into more in depth treatment strategies, which I think are still a little bit in the future In terms of to which degree is adopted. 

Now there are use cases. I can't say that it's a full AI independent mechanism. Let's say that is operating somewhere, or at least I'm not aware of that. There are good studies and good examples where AI in supporting clinicians and that was specific example in radiology, which is quite good for imaging is being assessed, that it improves outcomes in terms of assessing the diagnosis. So that was helpful, but it still has a person next to it. So currently I think the realistic position of AI and healthcare is supporting the staff in terms of replacing certain tasks. I think we are very risk averse and you can see that AI and tech overall adoptance in healthcare is much slower than in fintech or than in any other areas. We are a little bit slower because of the impact. 

0:19:07 - Mehmet
Yeah, 100%. But I believe you know, like with any other field, also in the market, as you mentioned, you gave some examples. Now in medicine we like to say let's take a second opinion. So maybe AI could be like a second opinion or maybe first opinion, but the second opinion comes from a human. I see it honestly this way because of course, we will not keep the AI fully to take decisions, especially around our health about our lives maybe. 

So I'm optimistic on this because on previous episodes we discussed this and we saw, you know, how beneficial it could be, especially for, for example, areas in the world that even they don't have proper healthcare coverage, for example. So this is what gets me excited about that. But you mentioned something around how usually you know the cycles of adoption in healthcare is different than other industries. So in your experience, like other than you know, being skeptical like what are other differences that you can tell us about, especially in tech for for healthcare? And you know why, for example, adopting the technology that comes from a new startup is also like a challenge. 

0:20:30 - Nadia
Yeah, there are a few. I think the most common ones and the most important implications that availability of data is relatively unequal. If we consider the broad tech and I know I'm combining a lot of different industries versus healthcare is quite different. We don't have as much data and we are very data greedy to make any kind of analysis to be inferential on applicable on a population level. So data availability is one aspect and it will continue to be so. Realistically, there are certain developments that say EHRs being available has massively increased the efficiency of healthcare analysis and adopting more advanced techniques, from ML now to AI. So there are movements, but it will be always difficult because, again, these are patient data and that is under a very strict governance framework, which will continue to be so, if not become more strict, and that also differs between countries and even within countries. So data availability is one. The other aspect is that we deal with a lot of well, probably a lot of industries can say that a lot of messy data. 

we deal with much missing data and we don't always know if it's missing or it's just not there and therefore not recorded. So can you say that a patient does not have any, you know arrhythmia if it wasn't recorded, and that's important. So we're dealing with a lot of uncertainty and using appropriate statistical analysis is quite difficult. So then you take an AI approach or take an ML approach, or take an ML approach and you need to adapt it to your specific use case. So there's a lot of rework which takes time. 

And let's say, if you go into NLP and imaging analysis, where NLP at first was developed not at first was developed, but best use cases let's say, if you give an image or and side by side text description, in medicine there's a lot of weird ways of describing that image, because usually if you have, let's say, a chest x-ray, you would say no pleural effusion or none of these symptoms are there, which is difficult because you have an image, you have a text and they are separate, whereas in a traditional task you would have, you know, a picture of a cat and that would be a description of a cat. So it's confirmatory. So it's a very way of thinking in medicine. We see in the dose and in the data and we need to adapt those. So missing data is one, availability of data is two, and then a very convoluted way of thinking and phrasing things in medicine. Yeah, but do you? 

0:23:03 - Mehmet
think like these challenges will be overcame somehow in the future. 

0:23:08 - Nadia
Oh, definitely, especially the way of thinking and phrasing of the things, and that is developments. I think over the last five years. Let's say NLP itself has done so much to overcome that. I'm actually very optimistic about that. Availability of data yes and no, we will have more data just by default, but I think it will always be a struggling move. Always be behind, let's say Pintek is my go to example. Missing data I think partially. There are some things that will be recorded better, but it always comes at a cost at an unfortunate clinician because we don't want to overburden clinical staff to record for the sake of recording. But again, we want to deploy AI, ml models. We want as much data as possible, but that's a tension point. So I think to some degree, yes, but methods for NLP and alike are certainly overcoming that problem. 

0:24:04 - Mehmet
Yeah, for the last point you mentioned about, like, causing an overburden for the clinicians, and this is because they need to do a lot of data entry, basically, and I believe AI can help here, in my opinion, right, so they can rely on this. I don't know, maybe with what we are seeing now from the ability of AI algorithms to understand the language and even they can analyze the way we are saying something, also combining that with visuals, I believe, yeah, it will take time, definitely, and I know, by the way, like with a lot of labs around the globe, that they are trying these things nowadays and they are trying to build, actually, what's now known as large language models to get this data out of Now. I want to come back to an area which is also important, like we touched base on it, which is mental health, right? So what do you see challenges that can face real world data when it comes to mental health? 

I know we talked about psychiatry, but mental health is something which is I discussed on the show a lot honestly, because a lot of people, especially in the past few years with the pandemic, with what's happening, closures and so on. So how, what would be the challenges to apply real world data on mental health? 

0:25:40 - Nadia
Fantastic question. We are considering this every single day and that's the thing that we live and breathe, aside from the known challenges such as subjectivity of metrics, that we don't know what to measure exactly, and it's not just us but a lot of clinical community are also debating certain measures Like what does reflect we being well Cause? It's a difficult question when you think about are you happy? It's almost. You know there are very few people who can very directly answer this question. It's almost always a philosophical question Well, am I fully happy? So this is always a spectrum and it's these are difficult questions to answer for us, let alone patients who have more important symptoms. So the subjectivity of that. 

One of the things I think is important to acknowledge is that, exactly to say, a lot of people struggle with mental health problems, and that is a massive spectrum. So you have broader population that has very different symptoms and a lot of comorbidity along the way. So even if we consider no hospitalization, so a patient is just having depressive thoughts and a lot of anxiety, these are already two different things. This is already comorbidity and that is very difficult to tease apart. What is the different? What do you treat first? Or how do you treat first? What is the best approach to do that? And that goes into all the more severe diseases as well. It's very difficult to tease apart. What is the cause, what is the main complaint, what is the main issue that is affecting patients life, and what is the best treatment strategy for that? So there's a lot of chaos and clinicians are doing the best that they can given the current developments, and so comorbidity is one. 

The fact that a lot of people struggle on different spectrum, I think, is two because there's a lot of information. I think one of the interesting things that probably is a bit more prevalent in healthcare is that we have data that are regulated. So, say, data that we are working with are under very strict governance. There's a lot of data, especially pertaining to mental health, that is not regulated and it's everything that you Google about yourself or it goes into a lot of recordings. So that's an interesting balance between developed in a very strict framework and something that is developed in a more less regulated aspect. 

So we deal with two different sources of analytics and how they impact all of those. So let's say, even NLP approaches that we spoke about. If you check publications in the last three years. A lot of those are on electronic health workers and there's a substantial amount that analyze popular social media entries as well. So I think a challenge is to use a part. What is what is reliable, what is valid, where does the data come from and how broadly we can apply this information to? Is it just limited to the study, or can we say something about a broader population? So if it worked for someone, does it work for a lot of people? 

0:28:51 - Mehmet
Yeah, so it's again the same story of the data real ability that you mentioned a few moments back. Now, nadia, you have to deal with different stakeholders. So we talked about some clinicians, we talked about, you know, the data scientists and engineers. I think, because of the governance parts, we need to talk to policymakers and any other stakeholders. And how do you manage, like for me, like someone in your position should be kind of a product manager, you know, because product managers they have to talk to engineers, they have to talk to sales, to marketing, to executives. So how does it go for you? 

0:29:35 - Nadia
Also very good question. It is sometimes challenging, but I think it's also one of the more interesting parts of the work that we do overall, and I'm epidemiologist, so I was always not a doctor, not a medical doctor, but I work very closely with clinical staff and I'm also by no means a statistician, but I work with analytics. So you're always kind of in between and you try to draw the best examples but you're never master of either, so a jack of all trades. That is the same case in what we deal now. We have to work with healthcare providers and clinicians and also anyone else in that healthcare provider setting, so that can be from governance, from chief information officer, so a lot of different roles, industries, of course, a very different stakeholder. 

And the way to deal with this, I think, is to have a lot of empathy, and that's quite. I think the only way to be honest to deal with this is to have a lot of empathy, is to imagine and try to understand where the people are coming from. So if you work with, let's say, a clinician, I need to really imagine what does it mean if I would ask him like, can you just record these five new measures? And that was just like yeah, sure, that will take additional 10 hours of my day and that will sacrifice patient time. So trying to really understand and hear what people are saying, I think is the most useful thing. And same goes to the industry. They have a lot of pressures, variety of pressures, and the more we understand it, the better we can cater to all of them and also balance probably different stakeholders, because we never work in isolation. It's not only the industry that we're, so we have to make sure that you're trying to understand all of them and balance your approach as well. So a lot of empathy. 

0:31:17 - Mehmet
Yeah, I can see you need to be very patient and you need to be. You know, yeah, like it takes, it takes a lot of yeah being patient. I would say, like, you know, like, because you need to deal with different people, you need to have the ability because, you know, I told you like it's very similar to being a product manager and you know, I know some of my friends are and you know they need to deal also with people who come from different backgrounds, or the empathy that you mentioned. 

It's really, but you know I love these kinds of challenging roles. I would say, because you know I believe you, they make this of the work. I can see that you are very excited about this, what you're doing, nadia. So if I want to ask you, how do you see this field the real world data evidence in healthcare a couple of years from now? 

0:32:16 - Nadia
I think a lot of the things that we are experimenting with now will either cement in their use. So, let's say, certain NLP approaches, certain tech solutions, certain methods that are highly reliant on available tech and data, I think we'll be further explored and some of those will be used very commonly. I think there will certainly be new approaches that we're going to start experimenting with. So some things are going to be satisfied with, some things are going to be new. Will it change completely? Will there be a breakthrough? I don't think so. 

I think one of the most appeal and difficult parts of working in healthcare analytics is that it is running a marathon. There's very few examples where all of a sudden something happens and then everything changes in clinical care, and I think this is where you mentioned patients and the most patient it takes is to sit with it and know that. It will take years, which sounds quite hard, but it needs to be Body of evidence is difficult to acquire and it takes a long time and multiple efforts. So it's one study will never hopefully it will never be something that just changes healthcare. You need to have at least 30 things corroborating your findings, and this is just causality that we deal with a lot. So whether it will be changed completely and drastically, probably no. 

Will we continue adopting things that are already developed in terms of AI and ML? Definitely, and I think you will see much better uses because a lot of people are doing that. So we will have more examples to learn from and also to develop. So I am very optimistic. I have to say, I want to say quite optimistic. I am very optimistic because there's a lot of things that are untouched and I'm keen to see how they play out. 

0:33:59 - Mehmet
That's fantastic. I lack optimism. I lack optimistic people. Usually I'm optimistic by nature, so I get excited also as well. And to your point, because in the show we focus not only on the technical aspects or, let's say, the detail, behind the scene aspects of what we discussed. We give hints to founders or to be founders or entrepreneurs, and when it comes to healthcare, I always my guest actually mentioned that I would say I second this you need to have, in addition to patients, you need to have the perseverance I would say, because you face a lot of obstacles. You have the policy makers who would say, no, you cannot collect this data and all this. So, yeah, it looks to me I'm not by any mean, I'm just a fanatic of healthcare tech. I can say anything that can enhance our lives. So this, I get excited myself. Any exciting project with what you are doing today with the whole mosque, nadia, anything coming? 

0:35:10 - Nadia
Quite a few. There are certain things that we have published already and I would invite anyone interested in mobile data and psychiatry certainly take a look. They outline some incredibly elegant approaches and to analyze into Zeta and to make an inference about a population. A lot of the conversation that we have is about severity at presentations, so this is very interesting in how it affects future risk and I would certainly invite visiting our website and having a look, but there are quite a few coming, so definitely keep an eye out. 

0:35:44 - Mehmet
Sure I should have asked this before. But who are your customers? I would say who you deal most with. Other I mean not for the clinicians, I mean who use your product. 

0:35:58 - Nadia
By far the biggest clientele is pharmaceutical companies. So we work very closely with a variety of different well and different pipelines and we work across different stages of their development as well. It's sometimes, let's say, phase two, more planning of the work and seeing how it benchmarks against what they have thought of their clinical development phases to be, and then we can also go into phase three and later points in describing populations, seeing how it compares to the results that they had. So, but certainly pharmaceutical companies are our main clientele. 

0:36:32 - Mehmet
OK, got it. Well, we can find more about your work and you know HomeMask. 

0:36:41 - Nadia
Certainly certainly visit HomeMaskcom where you can see a lot of publications and presentations of the work and we do have high attendance and conferences, so we try to really disseminate the work that we do. I think there are quite fantastic few rates. I am biased, of course, and for personal, you can always, of course, check my LinkedIn and I try to share keep most exciting pieces of HomeMask there as well. 

0:37:05 - Mehmet
Great. As we come to the end, Nadia, I have to ask this Is there anything that you thought that I should have asked you or anything I missed and you feel free also to mention? 

0:37:21 - Nadia
That's really good question. I think we covered it all. To be honest, I think we covered Key things are really important. 

0:37:29 - Mehmet
Oh, that's great. Some people they think that this is a tricky question. It's not by any means so. When I prepare, you know, for the for the recording of the episode, I try to make sure that I covered everything we wanted to discuss about. But you know, like we are humans, we can do mistakes. I could have the jump on anything. Some people no, no, this is not a tricky question. It's not like a kind of a getting grating for the show. So it's just for me to extract as much as possible useful information from my guests, because I learn also a lot. The audience, they learn a lot. 

Well, Nadia, thank you very much for your time and you know especially people who works in your industry the healthcare. I know how much busy they are. So thank you for taking the time for being on the show today and you know I appreciate also all the information you shared with us. I would make sure that you know the links you mentioned are also in the show notes and this is how we end the show usually. Thank you very much for everyone who tuned in today. Keep the feedbacks coming. I'm enjoying reading them and also, if you have some, something you don't like or something you wish that I could do better. I would love to read these things. You know I love compliments, but I like also to see anything that you're not liking also as well, if there's any. And thank you for tuning in. We will meet again very soon. Thank you, okay. 

Transcribed by https://podium.page