Data Analytics Chat

Personalised Medicine & Why The Right Data Matters

• Ben Parker • Episode 54

Share episode

Send us a text

In this episode of the Data Analytics Chat podcast, host Ben Parker welcomes Steve Labkoff, Vice President for Development and Medical Analytics at Bristol Myers Squibb. 

Steve shares his incredible career journey, from starting as a traditional physician to pioneering roles in medical informatics and analytics across various pharmaceutical companies. 

The discussion explores the intricacies of personalised medicine, highlighting the importance of identifying the right data and the challenges it presents. Steve also highlights the impact of life experiences on his career, the value of mentorship, and his innovative contributions, including the development of video game simulations for medical training and the establishment of a hospital for AIDS care in Africa. 

00:00 Introduction to Data Analytics Chat
00:33 Steve Labov's Career Journey
02:09 Transition to Informatics
03:49 Major Projects and Achievements
09:54 Challenges and Setbacks
12:50 The Role of Mentorship
20:43 Personalised Medicine and Data
24:15 The Future of AI in Healthcare
32:14 Challenges in Data Interoperability
39:08 Conclusion and Final Thoughts

Thank you for listening!

steve labkoff:

I had been enamored with computers since my high school days. I was the guy who would stay late after school and play with a teletype machine that was wired into the city of Philadelphia. One of the biochem labs was doing scintillation counts with a beta counter. And it was taking them literally two weeks to get the data off of these scintillation counts from the instrument and then processed them into graphs and have an ana an analysis of any one assay. And I went in there with a early style laptop computer. I created a i, I wired my own RS 2 32 cables, attached it to the instrument. Itself, read the data coming off the instrument with a modem program and then pumped it through Lotus 1, 2, 3, and took what was a two week project down to about 30 to 50 seconds. It's not just more data. More data can actually hurt you if you don't, if it's not stewarded well,

ben parker:

Welcome to Data Analytics Chat, the podcast where we discuss the world of data, ai, and the careers shaping it. Today I'm excited to welcome Steve Labov, vice President for Development and Medical Analytics at Bristol Myers Square. In today's episode, we'll explore his exciting career journey and discuss the topic of personalized medicine and why finding the right data matters. Steve, welcome to the podcast.

steve labkoff:

Thank you. Thank you for having me.

ben parker:

Brilliant. I look forward to this. And I guess to start, what's been the biggest, best advice you received in your career?

steve labkoff:

I think the best advice I ever got was to make sure that I was not getting overly deep in any one specific area to find range in the work I was doing. In other words. At, when I started my career, I was working in what a, a group at Pfizer called Business Technology. And for the bulk of the time I was there, which was well over a decade, IWI was spending most of the time only in a technology facing role. And then a vp, a new VP came for the group and he said basically if I wanted to advance my career that I needed to. Basically do what other pharma doctors do and get into the core business. And getting into the core business actually opened up an entirely new set of domains where I could apply my informatics and data skills. But in the context of actually being a member of the core business, and that led me to a role in. In the medical affairs organization, which really opened up the door for a whole host of new opportunities that I simply wouldn't have had. I stayed with my head, my head down looking at the technology stack in, its in, its I'm blanking on the right word to say here. If I'm only looking at technology stack alone. So ultimately. Adding to my capabilities by adding range to what I was doing was a really good piece of advice by a guy named Jonathan White, who was a VP at the time.

ben parker:

Brilliant. Cool. I love that. Okay, cool. You've had an amazing career today. So do you wanna share with the listeners your career.

steve labkoff:

Sure. I started as a traditional physician. I went to medical school in Philadelphia. I did a residency in internal medicine university of Pittsburgh. Later I landed a cardiology fellowship at the University of New Jersey, A-U-M-D-M-J. But I was unsatisfied in that world'cause I had really gotten attached to the informatics community as early as medical school and was looking for a way to bring medicine and computer science together. And in 1992. I went to an informatics conference and a fellow from the Brigham Women's Hospital gave me his business card and said, you should look into this. You can actually make a living doing this this field of informatics. And a few months later, I gave him a call. I had only looked at one program. I interviewed for the program. It was in Boston at the Brigham Women's Hospital in Harvard Med School. And believe it or not, I turned down my first interview they called and said, we have, we'd like to have you interview here. And I said, not feeling it. And then two weeks later, the secretary of the program called me and said, almost no one ever turns down Harvard for an interview. You sure you wanna do that? And I took the interview and the rest was history. I ended up becoming a fellow. There was on faculty. Then for another 18 months after the fellowship ended, got recruited into Pfizer and spent 13 incredible years there doing all kinds of different projects because at that time, in the mid to late nineties, I was literally the only physician. Who had any type of informatics training in the entire company, and the company was huge back then, not as big as it is today. But if you can believe it, back in 1997, there were, I was the only MD with an informatics training or any kind of background. So that brought me into doing all kinds of interesting pro projects, some of which included building out infrastructure for a joint venture with CVS and Pfizer, which actually became, what's today is the MinuteClinics, where I got to write all of the data and all of the applications for the MinuteClinics for the pilot. I got to help build a hospital for, in Africa for AIDS care over the course of 2001 to 2005. And I even got to sit on a venture capital team at Pfizer for about three or four years evaluating new and exciting opportunities from a venture perspective. So I got a very wide swath of experience there. And then later, as I said before, I got into the medical affairs group towards the tail end of my career before I had changed roles. From there, I went into. A short stint at Deloitte Consulting. But almost as fast as I got in, I got plucked back out by pharma and went to AstraZeneca where I was asked to create and lead three departments. All revolving around some aspect of real world evidence, one in biomarker development, one in real world evidence itself, and another in. Clinical trial design and interpretation. So all of these groups I built from scratch and that was the kind of thing that gave, additional breadth to the work I was doing. I also got to lead a large group of 85 people at the Wilmington campus. When that ended, I ended up going over to, a medical informatics company called Intelligent Medical Objects, where I was helping to translate their interface terminology into a way to use it in the pharmaceutical industry. Later from there, I went to another small pharma here in Connecticut. When that ended, I became the chief data officer at the Multiple Myeloma Research Foundation where I really dug in and helped to create the largest medical registry ever that stood up for multiple myeloma's, a bloodborne cancer that is devastating. It's a rare cancer, it's only 30,000 new cases a year and. The organization was interested in building out a registry to do longitudinal data analytics on a large swath of patients who had been developed, who had been diagnosed with the disease. We built that, we stood it up. We recruited like a thousand patients into the registry before I departed. And they went on to to have that run. From there I went to a bioinformatics consultancy called Quanti. I was there for a few years and then on to where I am now, which is at Bristol Myers s.

ben parker:

Amazing. So I guess what got you into this industry to start off with? What was your, there's a lot of people, early in their careers, dunno what industry to get into. What was your driving force?

steve labkoff:

I had been enamored with computers since my high school days. I was the guy who would stay late after school and play with a teletype machine that was wired into the city of Philadelphia. And you. And then in med school or college, in college, I tinkered around with it a little bit, but in med school it really took off where I was part of what we called the note take. We had a note taking service and I was one of the officers of the group. And I would find new ways. We transcribed medical school notes, but I got hired by one of the professors there. To actually expand a medical computing learning platform that she had been developing in biochemistry. And I quintupled the size of that program. And then we published it with Simon and Schuster. And that was really the real big step that I took in terms of diving really deep. That was the Temple University Medical School. I also started a small business there, wiring laboratories up. We actually called the company Laboratory Automations where we'd go into a lab, look at their business processes and find a computer solution to speed things up. So the one of the ones that even to this day is still, I'm proud of. One of the biochem labs was doing scintillation counts with a beta counter. And it was taking them literally two weeks to get the data off of these scintillation counts from the instrument and then processed them into graphs and have an ana an analysis of any one assay. And I went in there with a early style laptop computer. I created a i, I wired my own RS 2 32 cables, attached it to the instrument. Itself, read the data coming off the instrument with a modem program and then pumped it through Lotus 1, 2, 3, and took what was a two week project down to about 30 to 50 seconds. As soon as the instrument was done reading the counts, it would send the data through the RS 2 32 port, and that data became converted almost instantaneously into finished graphs that were used. By the scientists, and that was, one of the better projects that we did while I was running that little small startup, it kept my indebtedness down dramatically. Because, when you're a medical student in the late eighties, it was not cheap. It's still not cheap. It's still very expensive. So that's what got me into it. And then I stuck with it on doing side projects all through my medical school and residency. And as I said after cardiology was over. I decided, I really wanted to dive into this deep and I felt that I could do more good over my career if I did that. It turns out I was right. The project I did in Africa as an example. Probably touched the lives of minimally to this day, maybe 10,000 clinicians and at least 40,000 nurses places a training facility for AIDS care and treatment and, I built all of the infrastructure for that place and wrangled a bunch of money from various partners who were with Pfizer, like for them to make major donations for the organization. And we stood up that facility and it's still in that, it's still in, it's still up and running to this day. It's called the Infectious Disease Institute at the Malago Hospital in Rera University in Kampala, Uganda. So that's those are some of the early catalytic events that really. Resonated in my career.

ben parker:

Yeah, it's amazing and obviously congratulations and I think it's'cause you're dealing with humans like, like it's humans life, isn't it? I think a lot of people speak to within the pharma world, it's that sense of achievement helping people, isn't it?

steve labkoff:

It's funny when I've been, I mentioned that I had been with Deloitte Consulting for a short time, and part of the reason that I went back into pharma so fast was because I missed the. The the calling there's a calling to this, right? You wanna do things that are gonna help society in some way, shape, or form. And I really I wasn't feeling that we were doing that to the same, in the same way at Deloitte as I was doing it at Pfizer. And that's what drove me back into the into the space.

ben parker:

No, it's amazing. So then obviously you touched on one, key moment that you mentioned. Has there been only other sort of defining mo moments throughout your career that's really impacted your, where you've gone?

steve labkoff:

Other major impacts, honestly, I think there may have been one which actually happened in 1981, before I got into the field, but it's been a driving force throughout my life. I had lost a favorite aunt to a really aggressive breast cancer in 1981, just as I graduated high school and started college. And, that drove me to one, want to find solutions in whatever way I could contribute to try to help with that. That really resonated when I got to the MMRF. The multiple Myeloma Research, foundation of Myeloma is certainly different than breast cancer, but the idea of using data to help cure cancer was irresistible to me to take the things I had learned in my, of course, my career and bring it to bear in a way that could truly move the needle on coming up with new therapies or new treatment. Strategies and using it all through informatics, all through data was, it couldn't really get any better than that from my perspective. So that early catalytic event to try to figure out a way to work in the oncology space really came to bear in a major way in that program 25 years later.

ben parker:

And obviously I think life moments do impact you massively. Yeah. Everyone's gonna go through a period where they have their moments in their career that are obviously not the nicest feelings, but I think if you are, if that sticks with you and through your. Job you're gonna, it's gonna impact you every day. It's like you've got a reason to get outta bed, haven't you?

steve labkoff:

Exactly, and I'm not gonna say that it's always been a. Peaches and cream kind of career. There's been ups, there's been downs, there's been layoffs along the way, but all through it, I, my, my North Star has been use the informatics in whatever area that I have been in, and it's been. The anchor point and I'll say one other thing. There's an organization I've been a part of since oh God. I went to my first meeting in 1983. It's called the American Medical Informatics Association, or otherwise known as a MIA or amia. And through my participation in that. Peer organization. I have met and worked with some of the best people on the planet in the field. That would be the likes of my first mentor, which is named Bob Greenes. And then later on with another mentor named Charlie Saffron, both of whom are still around, and both of whom I still work with to this very day. And, for people who are just getting into the field. I'll tell you the thing that you should look at is finding mentors early on. Finding people that you can trust and who can provide you with skill advice or life advice and drive your career in that direction, irrespective of where you're working because those relationships stick with you. Despite whatever job you happen to be in, they're the kind of things that are long term relationships. And those long term relationships are really what matter across a 30 or 40 year career.

ben parker:

Yeah. And I guess you touched on obviously an interesting point about getting a mentor. How did you do it, did you approach people? Did they approach you? Because obviously a lot of people. Obviously we'll find it beneficial, but I guess it's choosing the right one Who's gonna impact your career? How did you how did you end up getting mentors in your career?

steve labkoff:

That's an interesting question because I I'm I typically do seek out, people around me to let, I, I'm not shy about asking questions. I'm not shy about asking for help. One of the things that I found in my career is that if you're, if you don't know something, not only is there no, no sin in asking for help, but oftentimes you look smarter by asking for help because you don't need to, this is a team sport for the most part. It's not a an individual activity. And, I think. Some people get the impression that they need to shine on their own and across their careers, and to some extent that's not wrong. But I found what's been certainly better and more helpful to me has been by collecting trusted colleagues, in my life. I have many of them. And it's all been around. Keeping those folks close at hand and fostering those relationships across time. Because you, and like I also had a philosophy where if anybody ever asked me for help, like anybody, even to this day I very seldom say no. And I don't say no because I never know exactly where the help is gonna be needed for me at some point. And I might need to tap on that person for whatever. I have a pay it forward philosophy, which is effectively, the people who have helped me. I don't make any promises that I'll ever be able to help them, but I do promise to them that the energy they pour into me, I will pour into others who are coming up the, up the mountain. And I've done that religiously across my career. And I think everybody in the, in my world, feels and works in a similar fashion. So that's been a life phil, a life philosophy that has served me very well.

ben parker:

No I think most people do wanna help people. I think it's human nature, isn't it? Yeah. Obviously depends on the person's situation, as well. Have you ever faced obvious, you mentioned a couple of setbacks recent previously. So how did they impact your career?

steve labkoff:

Sometimes a setback, like a layoff can be quite painful. The first time it happened when I was at Pfizer, it was incredibly painful because at the time my friends, my social life my, basically everything was wrapped up in Pfizer. Being let go. Left me really. Wondering about everything at the time, but it also opened up new doors, new opportunities that I didn't anticipate. And it it took a little while to get over the shock of it. I didn't understand early in my career what it meant when McKinsey was in the room so to speak. And whenever McKinsey's been in the room thereafter, it's not generally ended well. They tend to come in when there's big reorganization taking place and they reorganize a group. And that's what happened back in 2010, but. It also opened doors that, again, like I said, didn't anticipate, and it allowed me to get into things I never would've anticipated. So one of the results of getting into AstraZeneca was when AstraZeneca ended, I got asked to be on a project by the chief Medical Officer there to work on creating, believe it or not, a video game. Simulator that would be used to teach doctors how to go from being clinicians into how to being scientists, clinician scientists. And it was a simulation based game that was based on storytelling and based on providing feedback. And I got to spend a year and a half actually doing something completely wild, build, I never in my wildest dreams would've thought I would get involved in literally building a video game, but there you have it. And it was, it paid well and it was interesting and, got to dive into thinking about adult learning and platforms and learning management systems, and it opened up a whole new world of challenges based upon. This, the cm, the CI CMOs idea that teaching could be done through gameplay. And I was a big video gamer in my day. And taking that into a world where it actually could be used in a professional manner was really interesting. If you think about it, simulation is one of the most tried and true ways of doing, of learning these days. Boeing has been using flight simulators forever. NASA's been using them in the space program forever. Why wouldn't we want to use simulation in a more mundane field like drug development? It's not quote unquote rocket science, but it is pretty sophisticated and the scenarios that you run into in drug development are exceptionally. Challenging in their own right and coming up with ways to simulate. That only seemed like a natural thing to do, but up until then, and I'm not sure it's even happening to this day, I don't think that the industry uses it anywhere near enough simulation in its work in a general state, in a general per perspective.

ben parker:

I think it's key, isn't it, to when one door shuts, another will open. But I guess when you're in that moment, it is the challenge of. Of the negativity of that situation.

steve labkoff:

Yeah, it can be hard. It can be very hard. But, I've bounced back a bunch of times from these types of setbacks and each time I've bounced back, it's been into something novel and interesting. Had AstraZeneca not ended, I probably never would've found the MMRF, and that was one of the most. Impactful and meaningful programs I'd worked on in my career right up there with the hospital in Africa.

ben parker:

Brilliant. So what what do you think has given you the genuine edge in your career?

steve labkoff:

The genuine edge, I think that my I think it. Trying to connect dots. I'm a puzzle solver at heart. I do the New York Times crossword puzzle almost every day. And it's funny'cause I didn't start doing the puzzle every day. My grandfather did the New York Times crossword puzzle literally every day. And when he passed away, I found in his apartment the last puzzle he did, which was the, literally the day he died and it was done, which was amazing. And it took on. I took that on as a sort of a life challenge to learn how to do the New York Times crossword. And today I can't claim I'm as good as he was.'Cause I can't solve Friday or Saturdays, but I can get Monday, Tuesday, Wednesday, and most Thursdays and Sunday puzzles are always fun. But I think that the solving puzzles has always been one of my. That's what turns me onto a project or a program. It's like figuring out why something happens in a certain way, and it's really what medicine's about altogether. Solving the puzzle of why a patient's situation is what it is. What are the perturbations in their care? What are the, what is their metabolic situation? What is their hormonal status, and how does that all. Reflect in what you're seeing in front of you, and then piecing the, piecing it all together and coming up with not just a diagnosis, but a treatment plan and being able to follow them over the course of time. So solving puzzles has been like my intrinsic natural position, and that's what's driven me for most of my career.

ben parker:

Yeah, look, everyone has a problem and if you can provide a solution, you are gonna stand out on you. So it's people want answers.

steve labkoff:

Yeah, exactly.

ben parker:

Okay, cool. So let's move on to the data topic. We're gonna look at obviously, your area in personalized medicine and why finding the right data matters. So when we talk about personalized ME medicine, how do we define the right, what the right right data looks like?

steve labkoff:

Personalized medicine is a very interesting, new approach to, to healthcare. Not so new, maybe 15, 20 years old at this point. But it's looking for the therapies that really take into account things that will affect the individual's very specific situation. Now, where that's led is into the domain of biomarkers and into the domain of rare disease and rare cancers. And largely so let's take myeloma for an example. The MMRF prior to my joining did a major piece of work called the Compass Study. And the Compass study looked at what was the makeup of myeloma? Was it one disease or was it more than one disease? And by looking at the data, specifically the genomic makeup of patients with myeloma, they had 1,053 patients who were in the Compass study and. They figured out over the course of the program that they weren't just, there wasn't one flavor of myeloma, there was more like 12 flavors, and they were all genetically medi mediated. So in other words, there were 12 different genomic subtypes of myeloma. And when you looked at the longevity of the patients and the aggressiveness of their cancer and all these other parameters, turns out that the genomic subtypes mattered. And if you had one genomic subtype or another, your course could mean a 10 year survival or a two year survival. And as drugs were being developed and they realized this, they could then target certain drugs in myeloma for certain genomic subtypes. The reason we went into building the Cure cloud, which was the med medical registry I was a part of, was to try to. Get that to the next level and observe what was happening in all comers with different genomic subtypes. Personalized medicine is really all about understanding. What is the therapy that will be very specifically tailored to you individually? Now, it's not one thing, it's not always one genomic thing. It personalized. Another personalized medicine approach is what's going on with cellular therapy and CAR T therapy these days, which is basically you take a sample from the patient. You teach their own T cells what the can their cancer looks like, and then reinject them with their own T cells. That's a very sophisticated form of personalized medicine. It's also very expensive. And that's the problem with personalized medicine is that in order to get very highly specific therapies that are good for one person and one person only that tends to cost, lots and lots of money and. Unlike solving a problem like hyper lipid, hyperlipidemia, or hypercholesterolemia with a statin like Lipitor, you know where you can treat the masses with one pill, all of a sudden you're treating an N of one with one therapy and that one therapy cost of fortune. So while there're, it's very enticing to want to go there, there's this societal challenge around personalized medicine that is a balancing act of, is treating the many or treating the one, what's the objective and can society afford it? And I don't think we should get into the philosophical issues on that. Discussion today, but it is part of the equation for when you do personalized medicine and when you're making a therapy therapeutic decision, either to build a therapy or to administer a therapy as to how that plays.

ben parker:

Yeah and would I be right in saying that, I think with the LLM boom is that now being able to analyze more data quicker and it's gonna it's having a like really positive effect on the industry.

steve labkoff:

Is it having a positive effect on the industry? So I'm putting a conference together at Harvard Medical School right now. It's gonna fire off in a few weeks. That's looking exactly at that question. In other words, we've been in this now for about three years with LLMs, with generative ai and, what I've been, the reason we put this conference together at the Beth Israel Deaconess Medical Center through the Division of Clinical Informatics was that, at least from my vantage point, there's been an insane amount of hype in the system about what LLMs are gonna do and how they're gonna change the need, change the realm of everything from how you order your hamburgers to how you get diagnosed to what therapies you're gonna get from a medical perspective. And what I found was that the hype, there was a big mismatch between the hype and the reality. So we put this conference together to look exactly at that, to say, what does good look like today? Because it was unclear to me that we were seeing the level of. Of goodness coming out of it that was commensurate with the amount of investment that was being made right now. And I stand by that observation at this point. And while we have found things to present at our conference and we are having a webinar series, and by the way, the next webinar on that is coming up on September the 10th, and it's a free webinar. You can sign up at dci network.org/events. And I think the event number is two 14 for the next one. Which is on, this coming week. But the question really is where are the shining spots? We're trying to highlight the shining spots in our conference because I think, from my vantage point, I'm seeing millions, if not billions of dollars being pumped in and thousands of dollars in value coming out right now. And it's the right now piece too, which is important to put out there because I think we're still very early on in this adventure. And I think, I like to use the metaphor if you look at this industry, the AI industry on the same similar timeframe as the Wright brothers to the space shuttle, which was roughly 60 or 70 years. We went from basically 1903 to 1969 for the man on the moon and call it the mid two thousands for the space shuttle. So call it, let's call it 90 years, I think on a similar scale. If you look at that right now, I don't think that we have even approached the barnstorming years of the 1920s in LLMs. I think there's an enormous opportunity, I think that things are gonna continue to mature. But, there's been a lot of things that have not matured alongside it. The pace that we need, which are the policies and the regulations that need to go along with it and the governance and all these things are, all these things are in a rapid development cycle, but they're not all running at the same pace. So government advi advisories on large language models and the use of AI in drug development. So that means FDA needs to weigh in or other agencies like the Office of National Coordinator, HIT, needs to chime in. And it turns out that, guess what? Government and regulators and governance models don't move at the same pace as. The next version of chat, GPT. And we are way in front of our, what's the expression? You're out in front of your skis in some domains and in other domains, we're making some amazing progress. In, from my perspective, the data needed for things like artificial vision, those types of applications are working and working really well. It's now at a point where. Artificial intelligence engines and machine learning engines can do a better job of radiology and pathology analysis of slides or x-rays or MRIs than humans can do. So that's a shining spot in the world there. But, some of the other things about can we bring a drug completely from a computer generated model all the way to a drug that's being used in patients? Certain things aren't gonna be advanced so fast. If you still need to do, and you still do need to do, randomized clinical trials, they take time and that piece isn't gonna be accelerated. You can't administer a drug and wait for it to have an effect in a human and advance that in terms of the drug's effect, that's a biological process. Things in the pipeline can be affected, so like patient recruitment and site selection and all these things can be advanced by. And accelerated by ai. But the rest of the stack, specifically around the AC administration of drugs and the collection of that data, that I don't think AI is gonna make much of a dent in that because it's a biologic process for the most part.

ben parker:

Okay, that's interesting, but what kinds of data are actually needed to make treatments tailored to individuals then?

steve labkoff:

That's the thing of it. It used to be you needed basically just clinical data. But today, you really need to have a much wider array of data types to look at things because, answers don't just live in one dataset, but they can, but they tend not to these days. You need to have what we call multimodal data. You need to have data that's coming in from not just the medical record, but the medical record plus the commercial claims plus, then you layer it in with genomic information or proteomic information. You might wanna layer in data from wearables or from internet of things that are interacting with the patient. It's the collection of these different data sets in this, early in the 2010s it was called Big Data, but now I think we're calling it multimodal data, where you try to look across multiple data sets to see if you're seeing a signal in one data set. What shows up in the other the concept of a gwas study was one of the early ways of doing that and. Today, it's not just looking at genomic wide studies. It's looking across genomics and the proteomics and the EMR and all the others that I mentioned. And that's where the machine learning techniques and the application of AI is really gonna help us because it's hard for humans to take in that many dimensions of information concurrently, but it's. It's much, much easier and much, much faster by using these new tools that we have at our disposal. That said, it's taking longer than you would've thought for some of this stuff to mature. And that's the disconnect that I was mentioning earlier.

ben parker:

Yeah, and I guess I think with especially health data, you need to constantly keep reviewing it'cause. a person spends time, not for myself. I used to have to inject myself'cause I had ankylos and spondylitis when I was living in the uk, but then I moved to Spain in the heat all the time. And then the pandemic happened, so then I didn't take the injections. And then ever since then, I've not taken any injections from my back. So different locations impact people, doesn't it?

steve labkoff:

Yeah, look, we have, you're in the uk, I'm in the us. These are two first world countries that have access to the latest and greatest of pretty much everything. So if you even have a rare disease. You probably can get treatment for it if you couldn't have, if you have insurance. And even if you don't, you probably still can get some help. But if you're living in, central Uzbekistan or, and I'm just making that up I don't know if you've got the access to the same degree of cares. You might in other places. That's, unfortunately reality of the world we live in. But so I don't know if personalized medicine will percolate itself into those locations or Sub-Saharan Africa for that example, for another example. But I think eventually it will percolate everywhere. It's just a matter of how do these. These therapies and these approaches, they have to be tested somewhere. They have to be done in a way early on so the rest of the world can follow suit. It's just a matter of what's the timeframe, who's gonna pay.

ben parker:

Okay. And so what, what are the big challenges then in bringing different types of data to personalized patient med care.

steve labkoff:

I think some of the big challenges are actually, there's always the challenge of interoperability and this is one that's been a bugaboo of my field since the day I started and it was predated the day I started. Interoperability is basically if you record a piece of data from one hospital or one one clinic, can you harmonize and use another piece of data from a similar. Location or different location and that those data can be rolled up together. Are they interoperable? Can they be used at the same time? And unfortunately interoperability remains a huge bugaboo. The US government has tried very hard in the last several years to influence something called the SCDI which is an interoperability standard for healthcare data. And that's a start. But, not being able to roll the data together and use it in a harmonized way is frankly one of the biggest challenges that continues to dog the field. And that is the case for not just personalized medicine, but across the entire informatics stack. Another big challenge is, is there enough? Do we have enough compute power to do the multimodal analysis that we're talking about? And it's not just about the power of the computers, it's do we have enough data and getting the right data in the right volumes and in the right and is, and making sure it's curated and mastered properly is another big challenge. Maybe it's not even the computing power. That's the big challenge. The biggest problem is making sure you have the right data and it's in the right form to able to perform the analytics that are needed to do here. Yeah.

ben parker:

Is that'cause? Is it'cause there's so much of the data is untouched in pharma.

steve labkoff:

No. I don't think data's untouched. Pharma. I think that I, if I gave you that impression, I don't think that's right. I think that the problem is that pharma has access to a lot of data and being, getting it into the form that's needed to do a proper, an analysis is a challenge. So if you're bringing in data from multiple vendors or multiple hospitals or multiple places, you've gotta have it so that A equals a, in other words, a value of a creatinine, which is a chemistry lab, has to be in the same units and has to be in the same form from everywhere you're using it. So you can say, what does creatinine do over time across the world given this therapy? And believe it or not, you'd think that's an easy problem to solve. It's not such an easy problem to solve. Different parts of the world use different standards. They use different units and frankly from hospital to hospital, depending on the instrumentation that they use, sometimes they vary. Hospital to hospital. That's a huge issue, huge challenge.

ben parker:

So then now that, is there like warning signs that an organization is working with the wrong data

steve labkoff:

I'm sorry, repeat the question.

ben parker:

Is there warning signs that an organization is working with the wrong data then?

steve labkoff:

So I wouldn't say it's the wrong data, I would say. Have is the organization. So one of the warning signs I say is that does an organization have a data stewardship policy? And by a data stewardship policy means is, what do they do at the corporate level or at, more than just a department, but at a higher level in the organization to Im to bring data in and to harmonize it in a way that can be used in a s in a smart fashion. In some places where I've been data stewardship is taken front row center seat in terms of make sure the data's in the right form before you use it in other places. I've been, that's not taken a front row center. And all of a sudden in those organizations their ability to do the analytics that are necessary become harder. So it's not that they don't have the data, it's that, that they have a lot of data that sometimes maybe even too much data, but that they haven't spent the time. Or energy for whatever reason, to ensure that they have the data standards and the data harmonized in a way that is analytics ready. And in terms of how to make AI really pay off, AI is only gonna be as good as the data that they use for the training sets. And if you back that out, it's how good is the data harmonized and how well is it organized so that the AI systems can make sense of it. So there's this sort of Maslow's triangle of. Of data need that has to be thought through. And some organizations that I've been around pay a lot of attention to that. Some of'em not so much. And that remains, and every organization eventually gets there because they realize that they've, that they need to fix that. That Maslow's Triangle.

ben parker:

Okay. And then so obviously not obviously, but we hear, I hear that like more data leads to better results, but then in healthcare, could that actually be making like the data for personalization harder?

steve labkoff:

It's not that more data leads to more result, better results or more results. I think the question is more of the right data that's been harmed? So back to what I just said, ensuring that you have the right data in the right form to do the analytics is in my, from my perspective, that's the key. It's not just more data. More data can actually hurt you if you don't, if it's not stewarded well,

ben parker:

Okay. Cool. Cool. And there, are there new or surprising sources of data that you think could reshape how we approach personalized healthcare?

steve labkoff:

I think, as the cost of doing genomic analysis comes down and doing proteomic analysis comes down, I think. And our ability to digest these. And these data sets are gigantic. They're gigabytes and terabytes of information on an individual. So coming up with novel and interesting ways to collect it, steward it, store it, and then analyze it, those are gonna be game changers. And I think that's evolving and it's been evolving pretty quickly in a positive way, but I think there's still a lot of room for that to improve over time.

ben parker:

Okay. And then how close are we really to making personalized medicine a reality for the average person?

steve labkoff:

I think we're there in many regards. Depending on what kinds of diseases you have, in some places personalized medicine is wor up and running and working very well. Cellular therapy is a good example. I still think cellular therapy is pretty early on in it's overall use. It's, you take your own tissue. Your own blood, you train it. With the tissue, that's the cancer and then you give it back. You can't get more personalized than that. Now it's still very early on in days of how that's working, but it is working and it is possible to get per that kind of therapy. My cousin who recently passed away from myeloma, in fact, he had a round of CAR T and I think it may have bought another six or eight months of life. I'm not sure. It's not out there for everything, right? So hypercholesterolemia is unlikely to have a personalized medicine approach. But rare cancers that are genomically mediated, they probably will and do in many cases today.

ben parker:

Brilliant. That's fascinating. Okay, Steve. Yeah, really appreciate you being on the podcast today for share some great insights. So thanks for joining us.

steve labkoff:

It's been my pleasure and I hope this is helpful to your listeners and if anybody needs to reach me for any reason, you can contact me at steven@lab.info.