Hello everyone. A very warm welcome to our webinar today to level up your lab data, new use cases for improving time to treatment. My name is Daniel Chancellor. I’m going to be your moderator today. I’m representing Norstella, one of the webinar hosts. Norstella is the organization that’s brought together several prominent pharmaceutical intelligence providers all under the same roof with that shared goal of accelerating access to new therapies. We’re sitting on a wealth of data, whether that’s the clinical trial and pipeline intelligence from Citeline, consensus forecasting and financial modeling in Evaluate or indeed patient access and payer analytics within MMIT. We’ve now augmented all of this with a range of real world data sources from claims to medical records, patient demographic information, to lab data, and it’s lab data that we’re going to be focusing on today. All told, we’re sitting on around 70 billion data points all integrated and housed within our NorstellaLinQ platform.
And this platform is really revolutionizing how pharma companies are approaching decision making throughout the drug lifecycle. That’s enough about Nostella for the time being. If I can just keep your attention for just a little bit longer, have a few housekeeping items to go over before passing on to my expert speakers today. Firstly, thank you to Fierce for hosting and promoting this webinar. You’ll see on Fierce’s platform the abilities to submit questions to our speakers. Please do so. I think we’re scheduled for about an hour today and this includes audience Q&A, so we would love to hear from you. If we end up running beyond the full hour and your question hasn’t been answered, I’m sure our speakers will be in touch would welcome getting in touch with you afterwards. The session is being recorded. The recording will hit your inbox within 24 hours of the event closing.
So if you do have to drop in and out, that’s okay. You’ll be able to access this and indeed share this recording with any colleagues that haven’t been able to join us live today. Just lastly, if you would like to use closed captions or subtitles, I think these are available within the video player in the bottom right. Okay. So today we’re focusing on lab data. Labs sit within this general real world data RWD universe alongside perhaps more widely known and more widely used data sets such as prescription claims and electronic health records. Patrick, I’m hoping you can better explain what do we mean by lab data and how labs differentiate from these other sources. But first, I’ve given the Nostella intro and pitch already. Perhaps you could also introduce yourself and where Quest Diagnostics sits in this space.
Sure. Thank you. I’m happy to be here this morning and talk with everyone. So my name is Patrick Winniewicz. I’m part of the healthcare analytics solutions team within Quest Diagnostics. That’s the group within Quest that leverages the Quest data assets to build data driven and analytics solutions that are based primarily off of the plethora of lab data that we have. So just a bit about me. So I’ve worked in healthcare technology for about 25 years, adding that up now. I’ve been at Quest a variety of different roles for 13 of those years and the last five years really been focused on the data and analytics side of the Quest business. Then your question, and maybe just talk a little bit about what we mean in the lab data here and why it’s so interesting is that I have such a rich set of data.
It’s very close to the patient. It gives great insight into what is happening at the patient level and includes things like of course de- identified data here, right? But the demographic information, the ordered test, the resulted tests, the value of those tests, diagnosis codes, provider information, facility information. All of that included in this set of data that we have. And from a Quest perspective, we’re talking about 15 years of longitudinal data. And I know as we talk here, we’re going to talk about some of the different use cases, but just to set that stage, 15 years of longitudinal data, 65 billion data records. At Quest we see about one in every three adults every year. We have approximately half a billion, just north of 500 million new lab test results every year. So it’s an evolving and growing set of data. And then the last thing I’ll say about lab data and maybe just a bit about Quest is people tend to think about Quest as being a quote traditional lab, right?
So basic chemistry testing, et cetera. So our portfolio testing, they’re north of 3,500 different lab tests and that includes over 900 genetic tests, 1,000 esoteric, different esoteric tests that are available. So it really is a broad set of testing that’s available and allows you to get, as I said already, pretty close to the patient and what’s happening with their care.
Great. Thanks. That’s a really fascinating start. Maddie, sorry to have kept you waiting for so long. Please, can you perhaps introduce yourself, tell the audience a bit about your own role at Norstella and perhaps why we’ve all come together on this webinar?
Yeah, absolutely. Very excited to be collaborating with Patrick on the webinar. I’m Dr. Madeline Naylor. I lead our clinical team at Norstella. I’m the vice president of health informatics. So my team’s role is to take our client’s asks and best translate to our data assets like our lab data asset that Patrick was speaking of, of all the different data points that are in that data asset. I use lab data daily in my clinical profile and our clinical builds. As Patrick was noting, the lab data just provides such deep insights into the clinical and biochemistry profile of a patient. It’s the key driver in diagnostic and therapeutic decisions. So when we’re thinking of lab data, we’re thinking of using it as we use it for earlier detection. We use it for disease progression, patient monitoring, clinical trial, patient identification, and for also to find sites.
And then as Patrick was also noting, it’s for the genetic data. That’s a key component of the biomarker and genetic data of using it for biomarker signaling and targeting. So lab data in Norstella’s data lake data universe is a key part in how we build our profiles, identify and find the targeted patients in a de- identified manner, as Patrick noted, and really help bring those treatments to patients in the market.
Okay. Thank you. So Maddie, staying with you, is it fair to say we’re seeing more interest in using lab data and what would you say is, assuming that’s a leading question, but assuming the answer is yes, because we’re on this webinar today, what do you think is driving that?
Yeah, we absolutely are seeing more interest in lab data. I think several components. One, because of the insights it does provide for the patient profile. Like mentioned, it’s a key driver in those diagnostic decisions in the therapeutic decisions. When we’re looking at lab datas when I’m building clinical profiles, I typically start with the lab and build from there. I can get the insights I need from the lab that’s occurring that’s happening at the patient level. Also, genetic and biomarker targeting is a key, especially in the oncology space on how we can target these biomarkers and really get more effective therapies for patients. So looking at those aspects again in the earlier detection, the biomarker targeting the patient outcomes, we’re able to use lab data across the board for all of those use cases, which is why it’s being such a significant driver in this space right now.
So perhaps a reflection of the types of drugs that our sponsors are developing, perhaps more heavily lean into the required lab tests to accompany treatment. Patrick, do you have anything to add on, I suppose, the evolution of lab data? You’re saying the data itself has a 15 year longitudinal history, you’ve got a long history as well with Quest What have you seen evolve over the time?
Yeah, I mean I’d say that over time we have seen this kind of change or expansion perhaps when I think about five or six years ago and still of course claims and RX data still tend to be the most common commonly used types of data. And so when I think about when I began in the HAS role at Quest, a lot of our discussions with sponsors and customers was really around awareness and what’s available in the lab data, collaborating and talking about how you might be able to add the lab to as an additional data source and explore. And we’ve seen that change where in some cases lab is the driver because of that information that can be garnered and organizations have expanded their understanding and knowledge about what is available in the lab data, that it’s been nice to see this shift of maybe moving from a label think about lab to really seeing a tremendous amount of value in the lab data.
So let’s make it tangible. Oh, go on Maddie. Sorry, I was interrupting.
No, no. I was just going to say, I think one piece to add is the timeliness of the lab data as well, which has been a key for us. When you think of claims data and the especially valuable data set as well, but lab data is resulted much quicker. We can get those insights at a faster level. We can help with the earlier detection and we can really impact these treatment decisions for patients and providers. So from a lab data, it’s also the breadth and depth of the insights, but it’s also the quickness of the insights we’re able to get as well.
Absolutely. That brings us to the title of the webinar, which is accelerating time to treatment. Perhaps Maddie staying with you, can we make this tangible? Can we talk about specific examples, diseases, perhaps not naming drugs, but treatment settings where pharma companies have been able to employ lab data successfully and that has really made a meaningful difference in that closeness to the patient speed to insights.
Yeah, absolutely. So we’ve had several clients speak of the lab data and the assets that we’ve been able to provide in getting treatments to patients much quicker. One example was in the multiple myeloma space. We’ve built profiles both from a newly diagnosed profile standpoint and also those who have relapsed or refractory and looking to switch treatments that weren’t currently working. In that space across the profiles, we were able to identify over 800 new patients, either getting a new diagnosis based on the earlier detection with the lab data or switching to a more effective therapy for their specific diagnosis. So that’s just one example. We hear it day in and day out across our clients that we’re serving about the data that we’re giving them and the impact it’s making.
Absolutely. Patrick, any examples from your side?
Yeah. I think as Mattie indicated across the continuum of where lab data can be leveraged, I think about one example of a customer that we worked with over several years that really it was across several years and across continuum, we started working with them to identify prevalence, disease prevalence, right? So there was an underdiagnosed condition and the lab data can help you see that, right? And then as they decided to move forward, move something forward out of their pipeline into development and we started helping them recruit patients for early stag trials and then we helped to identify investigators. We helped them with understanding where potentially to place trial sites, again, leveraging the lab data to do that. And then as they were ready to move into later stage trials and into commercialization, we were able to continue working with them and leveraging that relationship and the data that we’ve had, they saw value in it through the entire process and just continued to work with us.
So whether that’s helping to understand prevalence, as I said, the investigators setting up trigger program, there are different places across the continuum where can drive value from that lab data.
I mean, if I had to guess, that sounds like it’s a rare disease patient journey you might be describing, certainly where you don’t have a handle on the epidemiology to start with.
Yeah. And I think we might talk a little bit more about that, but Maddie mentioned rare disease already. I think about another example where our team, even ultra rare disease, right? I think we had been working with someone that was trying to … Two examples, right? We have a program where for a more rare neurological condition we offer testing for that condition, right? And we are surfacing again, we’re talking about handfuls of patients, right, but the breadth and scope of Quest really helps get a number of patients tested. So what ends up getting surfaced each month are maybe a handful of patients for a therapy incredibly valuable, incredibly important. Earlier on in the development continuum on the trial side, we started working with someone that had spent, I think it was about eight months and they had only identified one patient for their trial. Now it’s not a huge number, but I think we got six or seven patients for them within a two month period.
I just think there’s a lot that can be done with the lab data and the breadth and scope of what’s available.
Sounds like you’re taking perhaps treatment settings which wouldn’t be clinically or commercially viable without access to information and actually then supporting these companies to, I guess, forge paths they couldn’t do before. I want to talk about how lab data sits in the wider pantheon or universe of real world data, because obviously this is just one source of information and there is a huge amount out there. Certainly within North Stellar Link, we curate a range of different data sets. Perhaps both of you could comment and I’ll start with you, Maddie, first. How are you integrating labs with other data sources?
Yeah. So this is the key part in what I do every day. So in NorstellaLinQ, we have closed claims, open claims, lab data, and then EMR data. And when we speak of EMR data that we have the structured EMR data, so we have labs within our EMR as well, diagnoses and then we also have the unstructured clinical notes. So if you go to a physician and you get a clinical workup, you would see that unstructured clinical text from that physician’s visit. That’s what myself and my team are doing every day. We’re looking into the data, we’re analyzing the data. When we’re building our clinical profiles, we’re using all of our data assets across and to appropriate utilize it to answer the business question or what is the end goal that the client is looking for. When we’re using our claims data, we’re looking at primarily diagnoses codes.
So they have had a diagnosis, if they have comorbidities, there are different therapies. Labs, we’re looking at some of the key lab data points, whether it’s for a diagnostic purpose or treatment progression or monitoring. And then we’re using our EMR data as well. If they’ve had labs drawn within a hospital setting versus like a reference lab setting, if they’ve had them locally drawn, if they’ve had some of that biomarker profiling done within the hospital networks and then any imaging test results and we’re bringing it all together to get that full view of the patient. So what NorstellaLinQ allows us to do is some of the gaps that typical RWD data assets may have. We get to fill those gaps and really see that entire view. So if we want to identify physicians and the appropriate physician treating the patient, if we want to look in it to find the appropriate patient or for patient tracking and patient monitoring, again, everything’s de- identified.
It’s tokenized on a patient ID, so we don’t know any of the patient information, but we can identify what’s happening in that clinical journey of the patient and map that out very, very detailed with our linked data assets. So really bringing the whole picture together. Now lab remains a key component of that, looking at their labs again for earlier detection. We have a heavy focus when we’re building those clinical profile on labs, but being able to add the procedures, the diagnoses, the whole picture of the patient really allows us for endless possibilities within our data asset.
And Patrick, any comment from yourself? I mean, you mentioned how you were able to use labs across the drug life cycle as it were and there’s the richness of the lab data in itself. I mean, how are you able to connect it with other data sources to help your customers?
Yeah. So primarily and historically we’ve leveraged and built solutions off of the lab data because we see so much value in it. We’re beginning to ourselves work with other types of data beyond the expanse of the lab data that we have. But when we think about supporting customers, investigators with the use of lab data, it’s important for us to make sure that we have well structured data that’s available, whether it’s for licensing or for input into an analytic, an analytic product. So we spend a good amount of time on our side trying to standardizing and normalizing data, turning unstructured lab data into structured lab data so that it’s easier to work with and so that as an end user of the data, we are providing that to you in an easier to use form. And then another way that we’ve been supporting people is that at Quest we have about 400 clinicians and PhDs and MDs on staff that are experts in what we call different clinical franchises.
So if you’re interested in, or if your portfolio is neuro, we can connect you with experts around lab testing and neuro. If it’s cardiometabolic, same thing there, women’s health, et cetera. And so it’s that combination of bringing our expertise around the lab data and making that available to the users of the data, making the data itself easier to use. And then as Maddie said, leveraging tools that allow us to link the identified data sources together at the patient level to provide that full view of the patient.
I mean, I think you may have answered my question already through some of the processes you described, but when it comes to implementing these data to your customers, normalization, perhaps the clinical context, are there any other challenges that come to mind?
I think Patrick covered it really well. We like to say at Norstella, it’s not just our data assets, but it’s our people that make our data unique. I mean, we’ve put together such a unique team of clinicians, AI experts, data science experts. It really does take a team effort to put this data into a usable ingestible data asset for our clients. So as Patrick noted, when we’re implementing data assets like this, especially across all of our data assets, it really is a partnership among our teams within Northstella to … Our clinical team is directing on how we should be querying what clinical time points, insights are we looking for. Our data science team is really in the data, querying that data, coding the data for us and our AI team if we’re looking at NLP for unstructured data or we’re looking at machine learning for trending and purposes.
So it’s really, again, it’s not just the data asset, it’s the people behind the data that makes the data assets unique.
Questions come in and just a reminder to the audience, please do continue to fire your questions, but it’s just come in. It’s really relevant to what we’ve been speaking about. So I’d rather come to it now than later, but are life science organizations able to link your lab data to other real world evidence such as social determinants of health or genomic data? Do you have any perspectives on those particular data sources linking your labs to the Patrick, perhaps before we start with you?
Sure. So yes, with the appropriate privacy preserving tools and techniques in place, right? So we support a number of different tokenization engines within the Quest environment. So as long as the organization, right, whether it’s your bringing your own data, you as a customer have your data asset or a data set that perhaps you’ve derived yourself and you want to link that to lab data. We can work together to say, “Well, which de- identification engine or tokenization software are you using?” And we can tokenize that data. We’ll make sure that if that aggregated data set needs an expert determination on it, we require all of that to be in place in order to link that data at the patient level. We also, as this transition over here too to Maddie, to Dr. The way we work with them is there is a, we provide data to them in a de- identified way with a variety of engines, but there’s one that they use and I think they’re linking their data assets across that tokenization engine and have that expert opinion that that data remains de- identified for the end customer.
Maddie, something to add?
No, you covered it very well. So we have our data on a Datavant token as our primary token we use at Norstella. However, we have been able to interlink with other partners given the appropriate de- identification and appropriate processes in place. So short answer is yes, we are able to interlink with other data sources, whatever interest source you were wanting to link that with, we would just work closely with that group to make sure we’re doing it in taking the appropriate steps to do so and keeping everything de- identified, tokenized, and the privacy intact.
Okay, great. So a slight change of direction, but Patrick and Maddie, you’ve both spoken about the clinical teams that you have to help, I suppose, understand the disease context and you might structure it along therapeutic area lines, whether that’s in neuro or cardio. Are you finding there are particular areas where the lab data has more rich applications or clients have been more successful in applying labs? Patrick, if I can start with you
Unsurprisingly we have a lot of inquiries for data in the oncology space and of course, as we all know, our kind of driver, precision medicine and understanding who’s indicated for which therapy and the genetic information around that in the oncology space we see that frequently, probably more than half of the type of requests that we have neuros very prevalent, cardiology is very, very prevalent as well, but the great thing across Quest, right there are things that we see most common when we’re talking about what do we see most commonly, but with the compendium as broad as ours we have testing and testing for
Pretty
Much any indication that you are involved in and then have team members to talk about what other things to consider, why is this test relevant? Is there anything new that’s being developed in that space? So the most common as I’ve said, but really across the board there’s testing available and we see that commonly as well. Maddie?
And Maddie, how does your workload reflect, I suppose, the ability to apply labs across the spectrum? Are you finding that again, oncology is the most mature area or are you finding any kind of interesting kind of hot areas of growth that the lab data are enabling?
Yeah. So I get all therapeutic areas across my desk. I will say I agree oncology is continuing to be the largest area of interest, both from the biomarker signaling part and targeting from protocol development and Norstella, our motto is pipeline to patients. So we support across all aspects of the development and implementation. So oncology still is the primary driver for that. Now I will say we’re seeing a pickup in neuro lately in the recent months. My background’s primarily in neurology. So I spent 10 years in clinic space primarily in the clinical research space in neuro, but we’ve seen a huge uptick in neuro interest across the board autoimmune diseases as well and rare disease. And I know rare disease could be different diseases, but especially in these really rare or misdiagnosed disease states, more and more companies are interested in looking at the lab data, seeing what led to the misdiagnoses, what the workups are now in the current state.
So we’re seeing that more as well. But agree with Patrick oncology is still the primary of interest, but we’re seeing others quickly pick up in this space.
I mean, on a brief word on the neuro space then, are we beginning to see whether you want to call it precision neurology? Is this more of an investigational setting or are we … Yeah, is it patient identification for new therapies that have just been approved?
Primarily new therapies and we’re seeing Alzheimer’s as an area of interest, Parkinson’s as an area of interest. So looking more at physician targeting and therapies, those would be the two primary, but again, across neuro, we’re seeing different interests in neuro in general.
Yeah. Of my notes that I have from our prep before this webinar was just around physician targeting and identifying NPIs. Patrick, can you just, I guess, walk through if you’re launching a new drug and you’re trying to find new prescribers for this, how does the lab data enable that?
Yeah, sure. So first starting with who’s indicated for the therapy, right? So any of the biomarkers that are relevant to that particular indication can be and obviously be seen in the lab data. And then there’s the direct kind of descriptive component of that. What have you seen in the past? We were looking for somebody with a result above or below a certain level, that’s pretty straightforward, right? A great thing that we also have that we’re able to do because of the breadth And the longitudinal nature of the data is look at trends. So you can see as someone’s levels are changing over time, that’s an important component that’s been part of the cohort definitions for us in identifying the patients who are being treated by those NPIs.
I think the other maybe other thing just to reinforce is that in the lab data, the order comes in generally with the ordering provider information, with the NPI and also the facility that they’re attached to. So just a way to focus on use of the lab data where I think sometimes or in the past it was claims being a driver for trigger or targeting programs, that information certainly comes from that a test was ordered on the claim, but you don’t have the result information on the claim. And then Mattie alluded to the availability of the data. So at Quest, a basic chemistry test might be submitted. So it’s drawn on a Tuesday and it reaches a Quest lab Tuesday night is processed overnight and is resulted back to the provider. I started with a Tuesday on Wednesday morning and then available almost real time in our database.
So when we talk about leveraging the data, the Quest data, it can be interrogated and updated like a trigger program can be updated on a weekly basis. It could be updated on a daily basis as well. So the speed of the lab data is also of availability of the lab data and having access to the NPI really can accelerate that outreach, those outreach activities.
Yeah. Patient trigger sounds like a really powerful use case that perhaps wouldn’t exist without access to lab data, that speed of a patient and the test results institution. So again, a slight change of pace. It’s probably difficult to run a webinar in 2024 without talking about artificial intelligence. I’m wondering if Maddie, you could comment on how AI is disrupting or enabling what everything we’re talking about here with labs, whether that’s patient triggers, whether that’s indeed diagnoses that weren’t achievable previously. How is I guess the surgeon interest in AI affecting the work that’s coming across your desk?
Yeah. So we were on an onsite not too long ago and the comment was made that AI is the future of healthcare and just seeing the progression and the acceleration that we’re able to have within our data space using AI technology within our AI team, it is actually pretty incredible, especially in such a short period of time. We’re able to design models that have helped with … When we say earlier identification, we’ve been able to identify trends in patients that led to a diagnosis and take those trends and apply them across our broader data assets and identify those patients even before they’re getting diagnoses. We’re able to use our NLP models. We have this large data lake. We have almost two billion records of EMR data. We have this large lab data. We have claims data. We’re able to use our NLP models to help us identify and target the appropriate patients and build these clinically significant deliverables for our clients.
We’re using AI NLP across the board in many of our data assets and really transforming the data and the offerings that we’re able to use. I think there’s going to be a continued interest and uptick in the AI and wanting more and how we could use it for our use cases. I think we’re kind of just scratching the surface right now as we are integrating all of our data assets and harmonizing and standardizing as Patrick was noting. But the possibilities we’re able to take our clinical expertise, our AI, our data science expertise and combine all that with the large data asset. And I think we’re going to have see a lot of impact across the healthcare industry with this.
That’s really exciting. Patrick, love to hear the Quest perspective on AI as well.
Certainly. So certainly I agree with everything that Maddie said. I think there’s a lot of opportunity in the future. At Quest, I do want to say are cautious with our use of applying any type of modeling or AI across our data. We do not do that in a broad general sense. We do leverage AI tools and ML to help standardize and normalize the data and some projects that have been identified and we’ve collaborated with sponsors. I think of one around the use of ML and with the AP anatomic pathology data and then understanding this, what can we learn from the anatomic pathology result, which is generally unstructured text, right? And then understanding and looking at the rest of the biomarkers that we have available on patients, right? We’ve done some projects and learned some learnings from that, which I think is an interesting thing going forward.
So you’ve got that arm of leveraging these tools to help make the data usable from an everyday perspective for all of us and then applying AI tools, whether it is trying to see maybe who qualifies for a trial or perhaps who’s those providers who may have a growing patient population for a certain condition by looking at the patients that they’re treating, applying a model to that to understand the progression of a disease and using more and more of the data that’s available now to help inform those models. I think there’s just so much as we’ve all seen the tremendous expansion and the use of AI tools just over the last year, that’s only going to continue.
And are we seeing- Yeah, please, Maddie.
Sorry. I was just going to say, I think Patrick brings up an excellent point though. It’s not just using the AI, but making you sure you’re using it in an appropriate manner, working with your teams, the clinical teams, the data science teams and using it to get to the use case or interest you want and not just applying it generally across your data assets. So I think that’s a key point that Patrick did highlight.
And are we seeing AI, I guess, upstream of some of these processes that we’re describing? So actually in the genesis of the lab data itself or enriching the kind of diagnoses that perhaps wouldn’t have been available prior, earlier disease states, that kind of thing, is that something that Quest is involved in?
So when we talk about upstream, right? So the order that comes in is the order that comes in. There’s nothing about kind of changing what the provider wants to order from a testing perspective, right? Now, are there insights that could be learned over time for understanding even that patient journey around lab testing or that’s a whole kind of different area that Quest is involved in around something called lab stewardship programs of making sure that the right test is ordered at the right time and the right location. And I think that’s outside of kind of the scope of this conversation, but it is something that we are partnering with health systems on to help them manage cost and care, right whether it’s under overutilization of lab testing, then when that reaches our repository, right, there are some ML tools that are used in the refinement of the data, but in no case are we modifying or changing the clinical content that we’ve received from the provider, but more preparing that data for use and utility and then other projects downstream.
I know you had mentioned and asked about upstream, but I think more of that kind of downstream, is there an opportunity to garner insights around speed of diagnosis, right? I know that we’ve talked about internally, how can lab be used to help accelerate diagnosis? Think about something like lupus and I think the average time to diagnosis is north of six years and I’m not remembering off the top of my head, but we’ve done some analysis on this and the number of providers that a patient sees and the number of diagnoses that they have over time. So are there things like that where we can leverage the breadth of data that we have along with the sponsor and learn from that so that we can help the broader patient, just help the community.
Yeah. Got it. Thanks for the clarification. I think what I was mentally juggling with in my mind was, I think you started off by saying that your universe of lab tests is something in the region of 3,500 different types of tests and I’m wondering whether AI is going to unlock a whole new type of test. So in theory, what’s coming in through the pipeline of data or tests that are being ordered might be that much richer as well. Yeah, lots of things downstream where we can sharpen up how AI is going to accelerate time to insight, but potentially more data might be enabled because of
AI.That’s a great question and so the Quest research and development team sure are looking at trends and what’s happening to understand which tests we need to bring online because we do launch new tests every year and some of that, just circuiting back to what Maddie said about areas of interest in neurology and I think about and Alzheimer’s, so the testing that we’ve brought online over the last two to three years to support the evolution of therapies and diagnosis around some neurological conditions I can’t speak to though kind of how AI was involved in that I suspect our team is using some of them.
Okay. I have one final question and then we’re going to transition into full on audience Q&A. So again, the inbox is filling, so that’s great. Please keep them coming. As we transition, perhaps if you could speak directly to the audience and if you had to put yourself in the shoes of someone that wants to get their feet wet with using lab data and they’re quite excited by what they’ve heard so far, what are the steps best practices for these companies to kind of get started? Maddie, perhaps Sophie you can go first
Yeah, it’s a bit of a complex question, right? There’s a lot that goes into making not just lab data but lab data usable as Patrick was saying, lab data that we could query and make into a usable data asset for our clients. A lot has gone into the Norstella side and the Quest side to make sure we are providing the best lab data asset. I do encourage anyone who’s interested in lab data in this space to anytime reach out to Patrick or myself we could talk from a clinical perspective from Patrick who’s been in this space for many, many years as he noted because we can’t stress enough the impact that lab data is having on the patients and finding patients, finding the right providers in the RWD space in general. I think Patrick would ever be best to speak on how to get involved in like a Quest specifically.
For Norstella, we’re happy to support … We’re always happy to talk with clients, talk through use cases, how our data assets can support their use case. Again, we have an integrated data asset. We have the claims, the lab, the EMR data, and taking all of those data points and really supporting that pipeline to patient use cases. So always happy to dive deeper into client. We build our clinical profiles, we build our deliverables based on the specific needs of a client. So if anyone is looking into lab data or data just in general to help answer those business questions or structure profiles, whether it be for clinical targeting or research or commercial targeting, always willing and happy to have those conversations.
So it starts with quite a, I guess a simple question really. So what is it you’re trying to solve? What are the pain points you’re finding? Is it recruiting for clinical trial? Is it understanding patient numbers? Is it finding physicians? Yeah. Patrick, would you add anything to Madeline’s comments?
I wouldn’t add, I would just echo, right? I think having the dialogue, right? I know that both teams are passionate, whether it’s like we work closely with the Nursella team as well. I know that they’re passionate about helping advance studies and projects as well as our own Quest team. And I would encourage people, it doesn’t matter what it is, right? I think of an example that we have where we actually link some census and income data to some testing data to help identify underserved hepatitis C zip codes, if you will. And we were able to find that in certain zips, the prevalence of hep C is higher than in others, right? I mean, that’s not shocking, but kind of using income levels and things like that, which probably wouldn’t be a thought or maybe initially, maybe it is for some, but maybe not for others, that how does lab data help me understand where there are other market opportunities and what can I link it to?
And that just came out of a conversation that we had. So I would just encourage that dialogue, I’m sure we’ll have contact information, have a conversation of something we can add to, something we can make easier for you, we’re happy to discuss.
And how data savvy, if that’s the right word, do you need to be in order to, I suppose, start working with lab data? What’s the kind of hurdles that we’re talking about? Maddie, I suppose you’re far more technical than I am, so that might be a very dumbed down question. So apologies if you take offense. But yeah, do you have conversations just about education and what you can do with data or are you working with quite knowledgeable data scientists?
Yeah. So our data science team is incredible. I can’t say enough great things about our data science team and what they do with this data asset, how they bring it into our environment. As Patrick noted, their team does quite a bit as well to standardize, harmonize this data, make it a very clean data asset. When you’re looking at lab datas, as Patrick noted, I mean, there’s different records, there’s different type of labs. When we’re thinking of labs, we’re thinking of blood, we’re thinking of urine, CSF, biomarker data. So all this data can also come in with different names, standardizing, different thresholds, different units. So of our data sets, lab data is before standardization, harmonization, everything that the Quest team does, everything the Norstella team does to make it usable. I mean, there’s a lot of work that puts into it to make it a data asset to pull out those key data points for it.
Now once you have it into that manner, I mean, again, we can’t stress enough the value of it, but it is a team effort to making sure the clinical team and the data science team is working very closely to pull out exactly what’s needed for our clients. So I’m not say it’s impossible for anyone to use, but there are nuances that are very important. So I think Quest or North Stella are great to kind of teach and dive into those types of things when if interested in using a lab data asset.
Also it’s a good piece.
Oh, go ahead Maddie. I
Was going to say it’s also a great piece when you are getting deliverables or from like Norstella or Quest that we’re doing that, the work of harmonizing, standardizing and getting a clean data asset into a deliverable for the client.
Yeah. There’s that technical component of standardizing, harmonizing the data, but then as Maddie said, the specimen source or type, right? So where lab can be, I don’t want to say complex, but why there’s value in having a dialogue when you’re scoping a project beyond the use case of just saying, “I’m interested in text test A, right?”
Well,
What type are you interested in serum, CSF, if that’s possible, right? Or do you want all of it? Do you want that test that was ordered by itself or do you want if that test was ordered in a panel of things, right?
Yeah. Do you
Want that test? There’s something called a reflex, right? So if you had one result, then it reflexes to another and there are different reasons why that happens. And our commercial team will frequently talk with end customers to help clarify that. And as we work with customers over time, they learn more about the lab data and kind of those nuances that Maddie talked about, because again, we want it to be easy for you to work with. We want you to be knowledgeable about these different nuances because then it also opens up some other potential use cases even ordering pattern behaviors. Why do providers order a standalone versus a panel? Are they ordering if there’s a new test out? An example that we did, there was a genetic test that was offered a next gen sequencing test that was more advanced than something that we used to do through cyber sequencing in different pockets of the country there was different uptake in that and the result itself gave you different information that impacted prescribing behaviors.
So it was kind of interesting how you could get all of that just out of an ordering pattern and knowing which questions to ask. Was it just that result that you wanted or was it the test methodology that you’re also interested in? So just a lot of different things that you can derive and learn from the lab data beyond the result itself.
I feel like we could have a whole nother webinar on just digging into lab data and structuring it and what it means and the type of data and the naming conventions and thresholds. And it’s pretty fascinating, but it’s also complex as well.
And we unfortunately or fortunately, depending on your perspective, our 10 minute warnings just come on up so we’ll have to take a rein check on the technicalities. So what do you do with the results and findings? This is a question. Are you delivering these insights to the HCP directly? My inference is from the conversation so far, you’re working primarily with sponsors who then might engage HCPs, but are you also going directly to HCPs yourself? I’ll put that to the floor and I’ll let each side who wants to answer.
Do you want to go first, Patrick, or I can take that one from Anor still aside?
Yeah. It’s not the HCPs directly on the Quest side, but Maddie, if you want to cover that.
Yeah. So primarily we’re working sponsored pharmaceutical companies in the pharmaceutical companies, whether it be for like clinical trials or clinical site identification or appropriate patient targeting, it’s primarily straight to the pharmaceutical company. Same thing with our commercial targeting. Our commercial targeting, we call them our HCP triggers are a big brand for us. It’s also one where we’ve had a lot of great success in. When we send those HCP reports, we’re sending them to the pharmaceutical companies of who’s the appropriate HCP making those diagnostic and treatment decisions. Then they’re using that to help inform their field reps on where to go where they have those conversations for their drug or brand of interest. That’s our primary ways of targeting.
Okay. Another question, I’ll put this to Patrick then in the interest of balance, but again, Maddie, feel free to jump in. I’m going to paraphrase. So the way diseases are diagnosed and defined can be quite rapidly changing, particularly in oncology when you might have a new addressable mutation or a new patient subgroup. How do you reconcile this kind of shifting in how diseases are classified with then how you would be working with your customers and enabling the rapid patient identification?
Yeah. So we at Quest do not diagnose patients. So we receive those diagnosis codes from, they come in on the order. So those diagnosis codes come from the providers themselves. We do keep abreast of all the clinical guidelines and as those evolve, our testing, if there’s a new test coming back to our R&D on the Quest side, if there’s a new test that we want to launch or develop because of change in guideline, or if there are changes in reference ranges and updates that need to be made, we keep abreast of those and then make that available on our website and as we publish the results to the providers, the reference ranges get updated. And then on the genetic side, as there is new information learned, the reference databases are updated to include that information and then as we result that genetic information back to the providers, we reference the most current available information.
We don’t try it. We do stay closely aligned to any evolving guidelines.
Yeah. Just I suppose summarize for my own benefit, you’re on the cutting edge. So if a test is being ordered, that is being captured. So if it is a new patient subgroup that hasn’t even been approved yet or that treatment, but someone is ordering that test, that data is coming into the system. So you would be able to identify when a positive result for those patients has come through. So I guess by very definition, the information is on the cutting edge of science. Right. We spoke about an example of Tuesday morning, Tuesday afternoon, and then Wednesday actioning the results. Maddi, I’ll put this to you because I know you spoke a little bit about the privacy and tokenization. How do you reconcile the speed at which this all can take place with the ability to kind of preserve, well, act within the right data protection legislation and maintain anonymity?
Yeah. So when any data comes into our environment, no matter the source, it’s all de- identified already. So we’re not doing the de- identification of the data. So if we’re getting EMR data, for example, and we’re looking at like a clinical note, an unstructured clinical text, it’s all already de- identified in our environment. With that, we have proces in place. So it takes us by the time … So we get typically a weekly refresh from our data sources. As Patrick noted, some labs we could do daily refreshes for some of the lab data assets. We have processes in place. Then when we structure our deliverables, it takes us about a 24 hour turnaround time. Once the clinical algorithm’s already built, we know what labs of interest or the clinical points of interest, I should say, whether it’s diagnoses, codes, imaging, labs of interest that we’re using.
It takes us about 24 hours from the refresh that we get into our environment to then put into a deliverable for the patient. In that 24 hours, we also consider the QC process. We have several QC steps in place, have a process set up on the backend for once we get the data in our environment, depending on what the clinical profile is, about 24 to 48 hours to then get it to the client in their deliverable depending on their refresh schedule as well. As we’re building these deliverables, we set a lot of processes up. So it takes about four to five weeks to get a first deliverable in hand on average. We get a historical deliverable, many of our clients, and then either a weekly or daily refresh or monthly depending on use case. Once they’re getting to that refresh cadence, we already have all those processes set up on the backend to make sure we’re ensuring accuracy, but also timeliness because it is key, especially when you’re informing treatment decisions for our HCP targeting, the earlier you could identify those patients and get to the physician to inform treatment decisions.
I mean, that is a key component to keep in mind. So about 24 to 48 hours for QC and timeliness for us.
Okay. One final question, Patrick, over to you. There’s a question coming about misdiagnosis and I understand, I suppose this might be quite difficult considering the data is the data, as you said previously, you’re not kind of manipulating, but is there any way in which you might be able to help your customers or your studies account for a test result and then perhaps being erroneous or understanding better the context of that test result? Are you able to perhaps predict when a test result has come in and it’s actually not one that should be actioned on?
Yeah. So I would say there are a couple things about that when one licenses the data. So again, from a provider perspective, we deliver the result back to the provider and at Quest we are not making any medical decisions. We provide that result back to the provider for them to review. There are processes in place around when we call providers with certain results, et cetera, things like that to have that consultation with them around the lab tests. But when we think about a patient and a journey for that patient, the longitudinal information that’s available for that patient, I think is that best source of that. So again, because we capture diagnosis and we have diagnosis over time, right? There’s the diagnosis that came on that particular order. Some things that we do with customers is perhaps you’ve got a, everybody’s easiest example, a diabetic patient, right, that perhaps today’s order for a flu test doesn’t have an A1 doesn’t have the diabetes diagnosis code on it, but testing from six months ago May.
So we can look across the patient’s history at all of their diagnoses and see how that has changed or been consistent over time and then it would be up to investigator to study and analyze that information.
So yeah, it does sound like you would be able to account for patients being misdiagnosed or diagnosed and then that diagnosis changing. Okay. So just to wrap up, our time is up, but thank you to the audience for those questions. You put me to shame there. As I mentioned, if any other questions want to come in, quickly send them in now and our speakers will get back in touch with you. But let’s thank Patrick, let’s thank Madeline for their insights today. Some really powerful examples of how drug companies are already using lab data alongside other sources to accelerate time to treatment, whether that’s in the clinical setting or indeed for approved therapies. Thanks again to Fias for hosting us today. The full recording of the webinar will be in your inbox in the next 24 hours. But yeah, thanks again for your participation and have a good day everyone.
