Jeana Konstantakopolous, MATTER (00:13):

Hello and welcome to the Data Science Innovation: Driving Alzheimer's Disease Insights. I am Jeana Konstantakopolous, the Senior Director of Partner Engagement here at MATTER and MATTER is a healthcare technology incubator and innovation hub. Built on the belief that collaboration between entrepreneurs and industry leaders is the best way to develop healthcare solutions. Our mission is to accelerate the pace of change of healthcare, and we do three things in service of this mission. First, we incubate startups. Since we launched eight years ago, we've worked with more than 800 companies that range from very early growth stage startups to larger companies, and we offer them a suite of services to help at every stage of their development. Our member companies have raised more than $5 billion to fuel their growth. Second, we work with larger organizations such as health systems, life science companies, payers, foundations to strengthen their innovation capacity, and we help them find value in emerging technology solutions by unlocking full potential of both their internal innovators and then creating more human-centered healthcare experiences through system level collaboration.

(01:24):

And third, we're a nexus for people who are passionate about healthcare innovation. We bring people together to be inspired to learn and to connect with each other. We produce a lot of programs like this one today, including large scale events for the broader community and small forums exclusively for our members and partners. Today's event is being conducted alongside our Brain Health Innovation Challenge, which you'll hear a little bit more about at the end of this program that we're producing with support from the Lundbeck US Charitable Fund. It is an independently managed nonprofit 501C3 that is committed to the responsibility and appropriate support of programs about restoring brain health. The Lundbeck US Charitable Fund is wholly owned by Lundbeck, a global pharmaceutical company specializing in brain disease. But for more than 70 years, Lundbeck has been at the forefront of neuroscience research tirelessly dedicated to restoring brain health so that every person can be their best.

(02:27):

So for today, for years, data analytics has been used in healthcare to fuel faster, accurate diagnose to inform decision-making, personalized treatment, improve patient care and outcomes, lower costs and more. But with the recent advances that we're seeing with big data and generative artificial intelligence, more organizations are exploring how these new ways can take modern data science tools to address persistent healthcare challenges. So talking about challenges, one of the key challenges right now in advancing care for our growing population of older people is living with Alzheimer's disease and the related dementias. And we're seeing that there are a wide range of disparate sources of raw data, including electronic health records, personal health records, patient portals, health related smartphones, wearables, and lots of unstructured data out there. And the question is how can we gain meaningful insights? Well, hopefully our panel today will help us to dig into that. Today we're joined by Mary Furlong, a leader in the longevity market. Elizabeth Powers the Vice President and general manager of US Regulatory Science and Study Innovation, and IQVIA and Ryan Urbanowicz, research scientist, computational biomedicine at Cedar, Cedar-Sinai Medical Center, and the co-lead of Tech ID and Training Corps at Penn AI Tech in the a2 Collective. And our conversation today is going to just dig into this topic. So with that, hello everyone.

Elizabeth Powers, IQVIA (04:11):

Hey, how's it going?

Jeana Konstantakopolous, MATTER (04:13):

Hi, Ryan. Hi, Elizabeth.

Elizabeth Powers, IQVIA (04:16):

I'm good.

Jeana Konstantakopolous, MATTER (04:18):

Great to have you guys join us today. Before we really kind of dig into the topic at hand, I'm going to go around the table and have you tell me a little bit more about yourselves and how you've come to look both at the growing older population in the US brain health as it relates to Alzheimer's disease and dementia. And then this question of data. All right. So I'm going to start with you, Mary, if you can tell me a little bit about you and your role.

Mary Furlong (04:46):

Well, I've been at this a long time. I'm a serial entrepreneur, have started three companies and I think have raised about 250 million in corporate sponsorships and venture financing for startups that are building companies in the longevity market. I produced the Longevity Venture Summit, which I've done for about 20 years, and the Washington Innovation Summit. And I have a podcast called Longevity Deal Talk in terms of brain science. I'm an advisor to the Canadian Brain Health CBI Group. I've been part of Poit Science since the beginning, and I'm recently judging the UK competition for business plans related to research and dementia.

Jeana Konstantakopolous, MATTER (05:37):

Thank you. I'm sure your a wide array of experience here will be of great use today in our conversation. Elizabeth, can you tell me a little bit about yourself?

Elizabeth Powers, IQVIA (05:47):

Yes. So as you said, I'm vice president and general manager of a group within IQVIA called Regulatory Science and Study Innovation. Our primary mission is to find ways of unlocking access to clinically rich data, including new data sources, some traditional data sources like electronic medical records, but also new data sources like wearables and so forth, big data sources, small data sources, and really figuring out how to use those data sources in a way that has scientific credibility and rigor. And so we are in the early stages of building out some new research networks that focus on CNS, including Alzheimer's, and obviously there are a lot of exciting treatments coming to market for Alzheimer's and dementia. And with that comes a bolus of research sponsored by pharma companies, and we are heavily involved in various efforts around that. So really happy to be here with you today.

Jeana Konstantakopolous, MATTER (07:03):

Thank you. We're excited to have you and certainly have someone who's riding the wave, so to speak, of what's happening out there in the ecosystem. Last, but certainly not least, Ryan, we have you and I think that you are our resident data guru, so if you would please share a little bit about yourself, that would be great.

Ryan Urbanowicz, Cedars-Sinai Medical Center (07:22):

Sure. I'm Ryan Urbanowicz. I'm currently an assistant professor at Cedar Sinai and Medical Center as well as an adjunct at UPenn. I run the Herbs Lab. We do research in the development of machine learning and artificial intelligence methods as well as our application to a variety of biomedical target data points or objectives. And our lab specializes in development of automated machine learning tools as well as interpretable rule-based machine learning. So I'm very much coming at this from the computer science side, data analytics side, and I've gotten involved in Alzheimer's research in particular over the last couple of years through the Penn AI Tech and a2 Collective that's funding research grants for technologies and ai, especially applied to Alzheimer's and dementia and other aging issues. And I approach data kind of agnostically because I'm involved in a lot of domains and also because many of the challenges and questions and data science are kind of universal and generalizable. But there's a lot of things that I think about in terms of Alzheimer's that are unique in terms of data quality, what features to collect, things like that. So anyway. Yep.

Jeana Konstantakopolous, MATTER (08:39):

Great. I think we have the perfect set of perspectives to get our conversation underway here. So let's kind of pivot here, which is, I'm going to start with you, Elizabeth, which is Alzheimer's disease and related dementias are on the rise. As we know, our older population is growing and because Alzheimer's disease and dementia tend to have a later onset for most people, it kind of goes in hand with the prevalence of these cognitive disorders increasing. It means that we're looking at more people encountering these diseases. And what are you hearing about some of the challenges in the marketplace as it relates to this?

Elizabeth Powers, IQVIA (09:19):

I think I'll kind of organize this along the patient journey, if you will. I think first of all, there is patient and caregiver fear and uncertainty about what is happening with what is happening to me, what is happening to my parents, my aunt, my grandparent, my brother, my sister, and deep fear about knowing an answer because of the implications of that from a caregiver perspective. Then once someone is within seeing a clinician about this, we're not. So I was just sitting here thinking, I've been working in Alzheimer's for almost 20 years now, and I think we're still seeing very inconsistent practices. It's not like frankly, oncology where they're relatively clear lines of care. There's not even clear diagnoses. And that is still the case. Now, my own personal hope is that over the next five to 10 years with new therapies coming, it will overcome both physician, patient and caregiver resistance to a diagnosis because there are treatments oftentimes when you're in therapies where there aren't really meaningful treatments, it can be very difficult to get to a diagnosis. And then there's just record keeping. Actually what gets put into the EMRs is very different from physician to physician.

(11:13):

And then the last thing I'll say is then there's the burden on the caregiver, whether that is someone in a nursing facility, step down, step up care facility, assisted living facility, or just in a home with family, there has to be a limit on the burden that's put on the caregiver, and that's something that really has to be taken into strong consideration both in terms of treatment and care and in terms of data collection for research. So I'll just hit pause there.

Jeana Konstantakopolous, MATTER (11:54):

I think that's a rich ground for us, I think to play in. But Mary, I'm going to pivot to you next. I know that longevity is something that you think a lot about and kind of this growing population of concern around brain health, not just by the way, do I have OID disease or dementia, but preventatively, what can I be thinking about to do that? And kind of having a community of peers focused as well. What's kind of your take on what's happening in the marketplace?

Mary Furlong (12:25):

I thought I might size market. So the longevity market's an 8.3 trillion market, and then there's a lot of riches in the niches. So if you look at the boomers, they are at the top end 77, so in three years they're going to be 80, and then you've got 20 years of older adults coming behind them. So it's a huge market. Now, the opportunity for innovation in the home, in the care setting, in the adult day setting, in senior housing communities and in places with dementia wings, that's really important to look at. But we're just at the very beginning. I mean, more people watch Wheel of Fortune then that's their cognitive fitness. And so if you take an issue like driving, which I'm very concerned about right now because you look at the number of people who are not going to be able to drive in the next 10 years, and we're not prepared for that in terms of accessing resources in the home. So some of the analysts think Uber Health is one of the most important new brands or brands out there that could play a role, but lighting can play a role, pharma can play a role, and there's a huge staff shortage. So there's really got to be brand new models for how we find train and retain caregivers.

Jeana Konstantakopolous, MATTER (14:07):

It certainly, no, there's not. I think that maybe that's part of this, right, which is that there's so much, not just from a standpoint of talking about the number of people that are potentially impacted by this, but the myriad of kind of concerns that they're starting to have to consider not just healthcare and data, but these larger access issues that you pointed to that all have a role to play in things like diagnosis or care. So really interesting. I think I want to take some of these things and maybe Ryan, I'll come to you next, which is this notion of data we heard from Elizabeth about data and relating to this kind of group being in lots of different places. As someone who spends their days looking at these troves of data, what do you think the current state of certainly Alzheimer's data, but maybe broad more broadly, this older adult and population level data that could start to have a play into things, what does that look like to you?

Ryan Urbanowicz, Cedars-Sinai Medical Center (15:17):

I think like a lot of biomedical domains, the data is distributed, it's siloed, it's messy. We're still figuring out in many cases, what are the right variables? What is the right information to collect on patients? What do we need to be collecting in order to target care or to monitor for care? So there's a lot of unanswered questions in terms of just knowing what to gather correctly, let alone how to do it well, and thinking ahead is really important. I think that's already been brought up. We need to think about what the future needs are going to be and make our data collection systems adaptable so that as our understanding of these issues changes, so can be the way that we collect our data and the way we utilize our data to translate it back into patient care or help or whatever it is that we want to focus on.

Jeana Konstantakopolous, MATTER (16:14):

And maybe as a way to take another kind of step back here, when you're looking day-to-day at these datasets, are you only thinking about things like the older adult and Alzheimer's disease, or are you looking at other kind of patterns and approaches that you're taking elsewhere?

Ryan Urbanowicz, Cedars-Sinai Medical Center (16:34):

Yeah, no, I wish I could say I'm entirely focused on Alzheimer's disease, but no, I think very broadly about a lot of biomedical outcomes and work with a number of data types. One small example, I was involved in a clinical trial for, I forget, I'm getting my biomedical outcomes mixed up. But anyway, in this project they had clinical trial data from multiple sites and just harmonizing the data from these pretty well structured programs was an absolute nightmare. It took two, three years just to bring this data together before we can even really analyze it, and just this as a reflection of what the current state tends to be in the medical field in terms of being on the same page from the get go and how we're going to collect information or how do we make use of the data that's already out there, how do we bring it together and leverage it in a way that is reliable and trustworthy.

Elizabeth Powers, IQVIA (17:36):

Ryan, if I can just dovetail on that. I mean, you just said that you were involved in a clinical trial and there was an enormous amount of effort to harmonize the data across sites. That's under the best of circumstances where sites are entering things into a pre-designed ECRF case report form to feed into an electronic data capture system. When you're working with real world data where even if the EMR is Epic, every system is configured differently, much less than adding in data that is occurring from outside the actual side of care, but getting a sense of is a person's activity level, what's happening with certain biometric, biometric data, heart rate, sweat anxiety, these are all things that are relevant to patients, people, people with dementia, and that data gets very hard to integrate and is incredibly messy.

Ryan Urbanowicz, Cedars-Sinai Medical Center (18:55):

Absolutely. And one other quick sort of side note, in addition to when we're thinking about collecting new data, another big point that might be worth discussing more is thinking about patient privacy concerns. We want to be collecting this data so that we can make best use of it, but also how do we do that without patients feeling like they're being monitored or giving away their personal freedoms?

Elizabeth Powers, IQVIA (19:24):

There's actually a question in the chat about passive data collection. So in IQVIA's business, we're really just beginning to see large scale studies, real world studies come through where sponsors are hoping to have passive data collection through a wearable. I mean, that's among what it really has to be in a certain way, but actually having a validated tool for that. In my experience, it's still very much early days and in our view, it is critical for research and dementia and Alzheimer's to have this for a whole range of reasons tied to things I've already said. But I think we're still some years at least two to three, if not five to 10, from really having the sophistication of tools and sensors to be able to do this easily and without burdening the patient or the caregiver.

Jeana Konstantakopolous, MATTER (20:43):

It's interesting because I think we do this in healthcare lot, Elizabeth, which is one of the first questions I asked you. You started with patient experience and caregiver experience, which is that individual level of healthcare, which is so intimate and so personable, but at the same point to make that a really meaningful kind of evidence-based experience for them. We depend on population level data, we depend on the rollup of all of that to enable that personal experience. So maybe Ryan, a question for you. As we talk about these disparate realms of data development, production living, little living spots, how do you think about population level data and unlocking some of those insights where either for this older population or you thinking that we need to look and what's your experience as someone who is looking at things like AI to uncover some of those things?

Ryan Urbanowicz, Cedars-Sinai Medical Center (21:48):

So there's already an incredible wealth of tools out there for machine learning and AI. There's some great advancements. Obviously that's still an evolving field as well. There's a billion unanswered questions, but in terms of thinking about analyzing this kind of data, the first thing that I worry about that sometimes can get overlooked I think is the data quality and collection. And that is absolutely essential. You might've heard the phrase garbage in, garbage out, right? It is tempting to have this magical thinking about machine learning and AI. It's like, I'll just pass it to the tools and then I'll get something good. And the real meat of all of this is always going to be what variables do I collect? What is the quality of the data I'm collecting so that I can leverage these awesome new tools to really make the boat most out of the data, but at the same time not fall into pitfalls like bias, right?

(22:48):

Bias is a huge one. Making sure predictive algorithms are fair, that they allow fairness in their decision-making capabilities as well as the information we glean from these tools and models. One, in terms of methodologies and making use of the data. I mentioned earlier, one of the areas of research I'm in is automated machine learning, and I'm actually just writing on a paper on it right now and surveyed a large number of auto mail tools, which is making it a lot easier for people to use machine learning and hopefully do a better job than the problem with machine learning analysis pipelines is that there's a billion ways to make one, right? Everyone has an opinion, everyone has belief, and there's a lot of right ways to do it, but there's also a lot of wrong ways to do it. So paying attention to the evolving machine learning AI in terms of how data analysis conducted I think is important. We all taking a role and being critical of how we do that, I think is going to be really valuable in the future.

Jeana Konstantakopolous, MATTER (23:59):

Well, maybe kind of taking a page from that, this notion of garbage in garbage out, there are a lot of efforts currently underway such as the National Institute on Aging is creating some data repositories. There are a number of other organizations that are looking to kind of create these data pools of information that has been well collected so that people who are looking to train their solutions, train their ai, have data that is relevant and meaningful. I'd love to hear maybe Elizabeth and Mary, if you know about some of these kinds of efforts underway, what your thoughts are as far as these data collection and repository efforts that are happening.

Mary Furlong (24:47):

Well, I know it from the NIA funding, so there's $160 million I think every year that they are funding. And I think about 60% of that is going into innovations related to brain health. And so we can look to the work of some of those entrepreneurs and see what they see. But maybe I should turn it to Elizabeth to say more.

Elizabeth Powers, IQVIA (25:16):

So there are some data sets that are large enough, but with the evolution of treatment, I think what we're seeing is that those dataset, what needs to be in those data sets is evolving. And so they don't always fit the bill depending on what you're wanting to research. And I do want to address a couple of points. Ryan made a point earlier, and there are a couple of things in going on in the q and a chat here about, I'll call it broadly social determinants of health. Part of what we're seeing with the coming surge of the existing and tidal wave of Alzheimer's research that we're seeing is the need for more diverse data. And what that means is that you can't just go. So typically large pools of data have been driven through academic research centers. Those research centers skew wealthier, wider, I'm sorry to make mass generalizations, but this is what we see again and again regardless of therapeutic area.

(26:49):

And so we are seeing a push on the part of pharmaceutical companies and specialty organizations to really try and get attached to community community points of care. And what that means is that you're tapping into physicians who are not accustomed to research. They want to be part of research, but they don't have the practices, they don't have the staff to support it. And even then getting, there's also in the chat a little stream of, yes, so much valuable healthcare information exists outside of a point of care and getting access to that also means that you're skewing richer, wider. And so it's really, I think there's one challenge in getting to the community physicians. There's another challenge in getting at the data that is happening outside of a care setting and just part of activities of daily living that is really important. And again, I think we're probably five to 10 years from really having good validated ways of collecting that data. I hope it's faster, but I think that the reality of that term validated means that it's a medium term thing, not a near term thing.

Jeana Konstantakopolous, MATTER (28:34):

I think this is an important thread to talk about, which is the notion of social determinants of health. I think, Mary, you had that really interesting comment earlier about transportation, and I know in some of our earlier conversations that we had leading up to today, you talked to me a little bit about banking and financial records, and you've talked to me a little bit also about the role of that secretary of state and the driver's license and what this means for older adults is to kind of, I'll say outside the healthcare point of data that still have incredible relevancy. And I'd love for you to just maybe talk about a few of those things. Yeah,

Mary Furlong (29:12):

I have a really great example. So I am renewing my license, and so I'm going to be 75. A lot of my friends are doing the same, and so they all have to take and prepare for the driver's test. And so I took the AARP driver's class and I had found, my husband found for me a really good program with AI built into it. So it doesn't just teach you the rules of the road, it helps you understand rules of the road. So in this class I was in, which was live, I said, oh, this program will really help you understand the signs and everything. Well, people said, I don't have a computer. So when you realize that digital access to your point about data sets is not there for everyone. So the notion of just what Covid taught us is we first have to help people get digitally literate, and then that's another way to gather information.

(30:14):

Otherwise, you could just have someone check the box and say unfit to drive. And so I think that's a big place that you see it. Banking is another, so older people are looking for things to do, and so they want to go and talk to the banker, the savings and loan person. They don't want the automated necessarily want the automated ATM because it's part of their socialization. And so it's the book clubs, it's the pharmacies, it's these local places where older adults appear that that's where you begin to see that they might just be losing it. So when their boyfriend tells them to withdraw some of money, someone they met, then the banker now sees that maybe something's not okay with mom or dad.

Jeana Konstantakopolous, MATTER (31:08):

So it's looking at maybe some of these unconventional sources of data, and not just data, but potentially thinking about this a little bit more humanistically contact points that we have in our community. But for a lot of those contacts, there is kind of a record. There's something that exists out there.

Mary Furlong (31:31):

I think the surgeon general's report is really important about loneliness and even the fewer hours that families are connecting, fewer places that people can go and gather. And as they get into their eighties, they have often fewer friends because they lose their friends, they lose some of their friends. So we have to think locally and we have to think very broadly about how do we reach these caregivers and how do we train the caregivers, many of whom do not understand the nuances of the research reports on things like dementia, but they're in the front lines day to day.

Jeana Konstantakopolous, MATTER (32:09):

I think it's a really interesting point. Maybe Ryan, I'll ask you, we've talked about some of these kind of unstructured data things that we see, for instance, like notes in a health record or maybe a banking flag or things like that. How do you, as someone who's looking at data sets start to consider some of this unstructured data that's out there to help weave a more rich and thorough story about the that we're trying to analyze?

Ryan Urbanowicz, Cedars-Sinai Medical Center (32:42):

Sure. So first off, I guess I should acknowledge that I'm not really an expert on analyzing unstructured data. I certainly collaborate with those that do. And this is an exciting time to be in that area of research. There have been over the last 5, 6, 10 years, pretty incredible advances in natural language programming use of large language models, deep learning methods in general to work directly with unstructured data to analyze that type of data. And I think it's important to keep an eye on these emerging methodologies. It's hard to keep track of because there's so many people working in this domain, it's hard to decide what is the most valuable research to focus on. And part of that is that from a machine learning data analysis perspective, there aren't really established rigorous benchmarking approaches. Every problem is different. It is challenging to say, well, this method works better than this method for these reasons.

(33:44):

It actually turns out to be a whole can of worms. And so there is currently a little bit of faith in saying, I'm going to use this method and run with it. But in general, trying to understand the advantages and disadvantages to any given approach, typically with these new methodologies that rely on deep learning. The biggest trade off for me is that you're most of the time giving up interpretability, which in medicine is often a huge selling point for methodology. We want to be able to trust our models and understand the predictions that are being made by 'em. And that's always been a big disadvantage of deep learning despite the hype and the attention that deep learning has received. Not to say anything deep learning is also incredible. It does some pretty amazing stuff, but it's just one tool in the toolkit, right? We've got to remember to use the right tool for the right job. And often that's not going to be deep learning for some of those reasons that I just mentioned.

Jeana Konstantakopolous, MATTER (34:47):

With that, maybe Elizabeth, you're sort of in the throes of where the rubber meets the road between these things, which is maybe traditional and conventional data thoughts, and then the understanding that they're not capturing necessarily the full picture. How do you bridge that chasm and what are some of those things that you think about day to day?

Elizabeth Powers, IQVIA (35:08):

Well, one is let's start with counts. Patient counts like how many to answer a certain question with the required degree of rigor for the purpose. How many patients do you need in your analysis and how are you going to get there? On some level you would say, oh, it should be easy. There are lots of Alzheimer's patients, but actually finding the patients. And then someone mentioned federated data in the chat. I would add this idea of federated data to Ryan's list of wishful thinking. It's just not that easy because everybody's structured differently. Similar terms, even among very, the leading experts or maybe even especially amongst the leading experts are used somewhat differently and means somewhat different things depending on who entered the data. How you bring all that together in the end means that actually getting sufficient patient numbers can be tough.

(36:49):

I think that, I dunno, I feel like we've covered a fair amount of this ground, but we do think a lot about are there big data records that we can go to? So we have another study, another in another therapeutic area where we actually are using driving records as an indicator of safety events. I've heard of research around gun ownership and Alzheimer's. As someone said, we now doesn't take much to get a gun. A lot of people are going to have Alzheimer's and they're going to have guns. And that is when someone mentioned this research to me, I was like, oh my God, I'd never thought about that. That's really, really scary. So we think about things like that. Someone has mentioned prisons, we've been exploring work with prisons. It is very hard for a whole range of reasons, but we do try and push ourselves to think about how big can you get with some degree of reliability depending on the research purpose. Maybe I'll pause there.

Jeana Konstantakopolous, MATTER (38:20):

Well, it's interesting. I want to maybe pivot the conversation just a little bit as we are getting towards the tail of our conversation here. The first thing I guess I want to ask a little bit about is you mentioned how many people do I need to have a successful study? What is that number? And I would ask because of the varied nature of Alzheimer's disease and dementia, which is that it involves a person who is not cognitively at normal function and sometimes is depending on other people to help guide them to appointments or create continuity in their day-to-day life and structure and medical records. Is that part of the reason that you all think that we have difficulty getting access to this is because of the nature of the disease itself and the necessity for that? I'll say caregiver role in its acquisition and regular study.

Elizabeth Powers, IQVIA (39:21):

So I would say that a lot of the challenges in accessing data for Alzheimer's are in many, many therapeutic areas. It's just, it's the reason even real world research is supposed to be more efficient and cheaper than clinical trials. And it isn't always. But I think for all the other societal reasons that we've been talking about, Alzheimer's presents and a special challenge, Alzheimer's psychiatry, dare we even say the word pain, where you begin. So we begin to ask ourselves, is it that actually there is that the science is nascent because there's a lack of data or is there a lack of data because the science is really nascent? Because when you, I mean, look, I know there's so much research going on in Alzheimer's, I get that. But when you look at the way research has evolved relative to how research has evolved in oncology, inflammation, rheumatology, those sorts of things, the science isn't as mature as those disease areas. We simply don't know as much about the evolution of the disease, the methods of research, the methods of clinical practice as we do about these other areas. And I do think that the lack of data, it becomes an iterative thing.

Jeana Konstantakopolous, MATTER (41:10):

Well, I think that's actually really interesting and where we want to end our conversation, which is talking about insights, because the hope is that through using data more constructively, that we are able to glean some of these insights that help us create better care pathways that empower caregivers to have more meaningful interactions. I'll just say all the good things that we hope for here in healthcare. So let's talk a little bit about that interpretation part of it. We're talking about all these things, these massive data sets, these social determinants of health, lots of points of M for data to enter into the picture. How do we make that meaningful? And going back to this patient experience, what does that maybe look like for this disease state? And I know this is asking us to be a little bit of that half full kind of perspective, but I think that's a good place to sometimes go. Maybe I'll start with you, Mary. Thinking about a lot of people, where do you think people who are going through some of this would want to see this data that they're participating in their data? How would they make it meaningful or what is maybe some of the hope around that?

Mary Furlong (42:32):

I don't know if this is the exact answer to that question, but what I see entrepreneurs doing is focusing on the care managers. And the problem with being a caregiver, it's an average of 49 year old woman who's already got a full-time job in a family. She's very club sandwiched. And so to add anything else into that role is hard. And a lot of the caregiving is done by the family, but recently I've been seeing a kind of new category. And companies like the Key, I think do a really good job of finding great caregivers and educating them about dementia. So I think you have to look at the corporate role of who is playing a role in the caregiving economy. In fact, we're having a conference on that in December. And I also think AI can play a role. So I think AI can play a role in the training. So I think that we have to look at the people who are touching the patient, not the patient themselves, to build data sets around the caregiver economy because that's the one that's going to be on the front lines pointing to this person has a problem or this person doesn't have a problem. There still will be a need to have research with the end user. And groups like Cabi and the group in the uk, they're beginning to create panels where you can aggregate and find some of those data samples.

Jeana Konstantakopolous, MATTER (44:17):

Thanks, Mary. I think that's food for thought. Ryan, when you think about the synthesis, so what of it all, what are things that you're thinking about?

Ryan Urbanowicz, Cedars-Sinai Medical Center (44:29):

Sorry.

Jeana Konstantakopolous, MATTER (44:33):

Yeah, I mean, we have, all data is wonderful, but if we can't distill it to have a meaningful insight that can change either a care pathway or inform how people should be approaching their disease management, just data numbers. So how do we do that? How do we look for those insights?

Ryan Urbanowicz, Cedars-Sinai Medical Center (44:55):

Gotcha. I mean, so first off I mentioned transparent or interpretable machine learning. That's one element that I always sort of fall back to as being important. And like I said, there's a lot of hype around AI technologies that are very much black boxes. They're opaque. And so I'd like to see more research and researchers developing and using methods that are directly interpretable. There's a small subset of us out there, but we're largely overwhelmed. So that's one thing to pay attention to. And focusing on understanding what are the variables that are important in our data sets and what else we should be collecting. So when we're analyzing data, I feel like one question we should always have in the back of our heads is if I was to do this study again in the future, what would I want to collect instead? What would be better variables or a better way to collect the data?

(45:53):

Another thing that's important on the topic of data size and fairness, most people when they think of data collection, they think more is better. And that is in most cases, true. We need more data to have more power to more have more confidence. But with larger data sets, typically they're going to end up being messier. They're going to be more heterogeneous, which is both good and bad. It's good because we're representing a greater diversity of people, hopefully if we're doing a good job gathering a broader dataset. But the downside is that a lot of methodologies are not really set up in analyzing data to consider this heterogeneity. And what I mean heterogeneity outs, I don't just mean different backgrounds, but I mean heterogeneous associations where if we're trying to predict an outcome, the factors that contribute to the occurrence of that outcome can be very different for different groups of people.

(46:52):

So this group of people over here, they get the disease due to these genes. And over here it's this combination environment and some maybe other gene or and beyond. And a lot of methodologies that we have, they're just trying to put together the one best holistic model that's going to make a decision for everyone. And that's a problem in itself from a methodological perspective. This is an area that we're interested in, we study, and again, I'd like to see more people just take this into consideration and think about what methodologies could we develop or could we improve to tackle those kinds of problems. Might be a little bit in the weeds, but

Jeana Konstantakopolous, MATTER (47:32):

Oh, you know what? This is all about creating these data insights. So I don't think it's in the weeds at all. And hopefully we have people sitting on the line that are thinking very much in the same way that you are about this. And maybe Elizabeth, I'll give you kind of the last call on this one, which is what are those insights? What's the things that we're hoping to glean? What are some of the things that you'd like to glean as you're thinking about this?

Elizabeth Powers, IQVIA (48:00):

So when I look at what one can look at the pharma pipeline for treatment, and it seems to me that there is a shift to earlier and earlier treatment. This is harder. This is hard to do because people don't, diagnosis doesn't always happen early. And someone, clearly, I've been monitoring the chat all along here, but someone mentioned something about could we do something like what was done during Covid, where many people, unfortunately I think not enough, but many people shared their data in some way. And I think that being able to get access to data in a way that allows us to understand what early diagnosis really looks like and begin to develop some more definitive early diagnosis, there's a lot to be overcome in that. We've talked about a lot of that here just in the last hour. But I would really like to begin to see more rigor around early diagnosis, more concrete understanding of what early disease looks like and what both patients and provider and payer systemically, what does that need to look like? Because that is not where our system is early diagnosis is not where our system is geared to.

Jeana Konstantakopolous, MATTER (49:42):

I think

Elizabeth Powers, IQVIA (49:43):

It's going to be essential for good treatment.

Jeana Konstantakopolous, MATTER (49:46):

Mary and Ryan, nodding aggressively. Go ahead.

Mary Furlong (49:50):

I just want to say I really agree with you about the identification of a patient population that can participate in that early diagnosis, and I think that's where you're going to get the motivation of the end user to participate.

Jeana Konstantakopolous, MATTER (50:09):

Absolutely.

Ryan Urbanowicz, Cedars-Sinai Medical Center (50:12):

Yeah. I think this is a good example for an opportunity for maybe citizen science. How do you get people engaged? How do you give them incentives to be engaged in either providing data or helping us to understand early detection and at the same time link that to a direct benefit to them?

Jeana Konstantakopolous, MATTER (50:35):

Well,

Mary Furlong (50:35):

Wild idea, which is why not make it fun? So AIS Innovation has partnered with AARP to stimulate new people to play games, and they launched new games like Monopoly and Trivial Pursuit with many generations. But one of the things older people talk about is using gaming to keep their brains active, like the crossword puzzle. So if you kind of begin with what they're doing, 150 million people watch Wheel of Fortune talk about an audience and a population. And I was at a memorial for a woman who was 102, and the secret was Wheel of Fortune, keeping that brain active in her book club. So I kind of think we need a big initiative to get people to get motivated and to make brain health as important as physical health and cardiac health.

Jeana Konstantakopolous, MATTER (51:41):

Well, I love this. I think we've heard everything from citizen scientists to extended multi-generational gameplay from Mary and Elizabeth's concerns about keeping the caregiver involved and staying patient-centric. I think ultimately what we've heard is there's lots of opportunity here. So I'm actually going to say at this point, thank you to all of my guests today for your really poignant insights for your experience in your relative fields. I think there's a lot to be done here. So with that, thank you for joining us today. We're thrilled to have had you, and we hope that you'll visit us for more information. Have a great day.