Back to podcast

Pamela Spence

Hi, this is Pamela Spence, EY Global Health Sciences and Wellness Leader. Welcome to our podcast. Today we'll be discussing the importance of data and how it can help redesign the health experience. I'm excited to be in person together today with our guests in-spite of the latest Coronavirus variant, all socially distanced of course.

So, let me first introduce Jackie Hunter, Board Director of Benevolent AI and Chair of the Brainomix Board and the Stevenage Bioscience Park. And Ridwaan Jhetam. Ridwaan has just left GSK where he was VP on Oncology Medical Fairs in Europe and has just started at BMS as VP, Global Medical for Hematology.

So, let's get stuck in with our first question: who will you partner with to access the data you don't own, but need?

From my perspective the health industry generates an unprecedented volume of data each year which is only getting bigger and bigger for data sits in silos and because of it, it's very challenging to actually extract value from all the various different data sets generated. No company, of course, can own all the data that's generated and a discussion I've been having with some has been actually data ownership isn't important, it's actually access to data, whether it's yours or someone else's and the ability to interrogate that to provide insight.

So, how can a company access the data they need but don't own? We've seen recently, and perhaps we'll touch on it, a growth entrusted research environment, federated data systems and arrangements etc. but, let me start with you, Jackie. Both Benevolent and Brainomix are focused on using AI, what's your response to the, the question about how will you partner to access the data you don't own but need?

Jackie Hunter

I think, you're absolutely right, we don't need to own the data, we need to access the data and be able to interrogate it and use it in an effective way. I would say that from a BenevolentAI point of view, we make a lot of use of public data but then we partner, for example, in our collaboration with AstraZeneca where we bring aspects of our knowledge graph and incorporate AstraZeneca's knowledge graph, knowledge and material and experimental data and clinical data and really create a specific knowledge graph for them, which we can then both interrogate in our shared, shared programs of work.

A knowledge graph is actually what Netflix really uses to surface the films it thinks you are most likely to see, it's a way of representing complex, huge amounts of information and relationships.

In healthcare, for example for the BenevolentAI knowledge graph, essentially, it's really a systems biology map of all the biomedical entities, genes, proteins etc. and the relationships between them that are known and then populating it with relationships that we are inferring from the knowledge that we already have.

Spence

That's fascinating, Jackie. We hear so much about structured data, easier to interrogate, unstructured date, more difficult to interrogate. Does this knowledge graph thing actually work for unstructured data as well as structured data?

Hunter

Yes, we have, essentially, two pipelines. We have a pipeline for structured data and a pipeline for unstructured data and we trained the system using natural language processing to be able to recognize both the entities, so it knows that ALS is the same as motor neuron disease, is the same as Lou Gehrig's disease but also the context, so that AD in one paper is atopic dermatitis and AD in another paper is Alzheimer's Disease. And you can then map the relationships between them and extract them and bring them into a common file format that then populates the knowledge graph. So, we can deal with both structured and unstructured data.

Spence

That's fascinating.

Spence

Ridwaan, coming from a bio-pharma background, I would love to hear your perspective on how big pharma is viewing a more data and collaborative environment?

Ridwaan Jhetam

I would categorize the data into two brackets: internal data that pharma collects over the life-cycle for product from the time it's being discovered, researched, developed through clinical trials and then real world evidence associate with the product. There's also a barrage of data that pharma collects internally via different functions within the pharma organization itself. There is internal data, that's the one category and the other category had got to do with external data, uh that's collected through, through electronic medical records and other sources where, again, we don't have the ownership, we don't have the ownership thereof. But certainly, and that doesn't have to be the case at pharma has to have the ownership thereof, as long as we're able to access this and our concerns of course with IP and data protection. But that opportunity to access that data for pharma, I think there is an increasing understanding now that that can only improve that whole cycle in terms of new data being made available, for that to be taken back into the R&D engine.

In terms of data privacy, of course it's of paramount importance that patient data, held data itself is actually protected. And as an organization this has been an area where pharma has held back pharma to an extent that we haven't been able to fully partner to the-- to the extent we'd like to with, with external stakeholders and academic institutions. The kind of concern that big pharma would have access to patient data, uh, has held back that interaction. But we're increasingly seeing academic institutions and, uh, c-- cooperative groups, for example, in the area of gynae-oncology, to use that as a tangible example, are now reaching out to pharma to say, can we partner with you in terms of sharing information, sharing selected parts of information, from electronic medical records, to actually go beyond, to setting up registries, but to leverage the data, to look at it more carefully in the interest of better patient outcomes, and, and benefit of society at large.

Spence

Something that has really fascinated me is if I look across other industries, Airbnb, Uber. those companies have excelled in, you know, characteristics such as convenience, seamless exchange, predictive and personalized, high consumer choice and transparency all seem to be very prevalent common themes. So I'm really intrigued about how we can build that same level of leadership and user experience with those characteristics in healthcare, how do you see health experience and data driving a health experience, with some of those characteristics as we move forward in developing healthcare models?

Hunter

Firstly, thinking about the experience of the user, from a Brainomix point of view, they've developed this really neat stroke app which was developed with the clinicians. So really looking at what the clinicians wanted and their experience as opposed to presenting a solution so that -- and I think, you know, within Benevolent we've been doing the same with the scientists and the technologists to really co-create what we need. So, I think first of all actually really involving whether its patients or nurses or clinicians in actually shaping the experience, what is it that they really need and would make them really adopt the technology? Because actually, adoption is a big issue so, a lot of this is as much about a social science as it is about technology.

Spence

I guess you can invent anything but unless it's adopted, it's a worthless invention.

Hunter

Exactly. Exactly. I think the other thing about these technologies and the ability to really personalize the data to you, if we look at how you can categorize patients through mining their characteristics, not just genetic, but actually the power is being able to bring together genomic, pathology, disease mechanisms etc. And really then tailor the treatments that you're developing to the particular subgroup of patients that will be more likely to respond to that treatment. And that's important, not just because it's better ethically but also you're not exposing the cost of the health service or, or the patients to, to medicines that won't likely work.

Spence

And, turning to you Ridwaan, the user experience for big pharma, I've worked with big pharma for, for many years, that seems to be flexing a bit, notwithstanding the consumer arms but in the pharma arms but, who do you see as the user? And once you define the user of course then we can define the user experience.

Jhetam

I think across pharma there's the realization that one has to look at the organization have to look at beyond the pulled solutions. And these solutions require access to data, it requires healthcare professionals to have data accessible to them on tap or on demand. In addition, it's got to do what the, the final end key user being the patients themselves being empowered to be able to take the right decisions, to have the right information at the right time. And the opportunity to leverage technology also comes into play here where this can actually play an important role for patients to be, able to be um able to, to kind of access health solutions online at any time, and also to be able to access healthcare professionals through this mechanism as well.

Spence

I sense that there is a little bit more of a shift to actually the patient, the consumer, as well as the physician. I think it's fascinating about the difference that, that data can make in that personalization journey. How, how is big pharma responding to that?

Jhetam

I think there's definitely a lot of work ongoing and much more to be done in that space. We certainly are tapping into, into the two that are out there, we also looking at those partnerships with key players. I can speak and say that from a geo-scape perspective it's no bit trade secret, the same with other pharma companies that are engaging with third party providers to be able to actually have more information available throughout that life-cycle of a particular product rather than to be only have information that comes through an adverse event, for example, that's reported through the traditional mechanisms. Through social media we're able to now access information literally continually in terms of being able to make adjustments, for example, our clinical trials, to adjustment to doses of medications, and ultimately in terms of being able to actually take decisions about newer opportunities for drugs to be developed for other indications coming out of this exercise.

Spence

Which data are the most important to your business model? Clearly one can't have access to every single bit of health data everywhere all of the time. 

Spence

Jackie, coming to you then, how do the organizations that you spend most of your time with actually decide which data you actually want?

Hunter

Well, you hit on something, Pamela, that's really important because actually it's not necessarily the volume of data, but it's how you're going to process it, how you're going to use it and how applicable it is for the task at hand. From a Benevolent point of view, I think the most important data that we can access is really matched patient level data and by that I mean, that you've got the data for an individual from their genetics, maybe their clinical data, etc. so that you need to have a reasonable size of data to be able to interrogate it. Although there are technologies available that can work with less data but, I think, for us, it's really around focusing on genetics, transcript tone.

We are always interested in working with partners who have large scale data sets. UK Biobank, we really like working with UK Biobank, and it's driven a lot of our precision medicine workflows and also public data. We are lucky that public data, in the main, is actually quite well curated especially in some of the more kind of um structured data sources. And it's-- what we really are trying to do in our business model is putting priority views on that data. And I mentioned how we've worked with AstraZeneca in the collaboration that we have there. So, actually from a Brainomix point of view, having access to good training data for developing the algorithms for looking at screening and surfacing it to the physicians is really important. And one of the issues I've, I've found talking to SME's is that certainly within the UK system, there's a lot of difference in terms of how access to the data is allowed, there's different cloud requirements across different health trusts etc. So, wherever possible being able to, to streamline that but, it's really about having that high quality multi-omics, transcript omics and clinical data from a patient point of view and then-- and good data that you can process because quite often, a lot of the time you're spending is actually in processing the data to make it useable.

Spence

And Ridwaan big pharmas data strategy, the maturity of it, I'd love your perspectives.

Jhetam

So, again, I think that from a pharma perspective it's that pre-clinical and clinical data that we collect internally through our clinical trail that's most imperative for us to actually bring those products to the market to have them submit it to the health authorities for, for registration. But, then, we also need to make sure that these products are then available so, having the real world data to compliment pre-clinical and clinical data to ensure there's reimbursement and broad patient access is most important across, across the continuum. That said, the external sources are most critical. Again, so it's kind of a combination of both internal/external source that's going to be most important for pharma to put as part of it's digital or data strategy for each asset and across the portfolio itself, whether you are working in oncology therapy area or a vaccine area for example.

Spence

I'm just conscious that we've had a fantastic discussion this morning, I think it's unlocking the power of data to really drive a better health outcome, health experience for patients and physicians alike is going to be a fascinating topic, and area ripe for significant innovation in the future. So, thank you very, very much to my guests today, Jackie and Ridwaan. It has been amazing to be able to record this in person, in the same space, of course, socially distanced so, l thank you very much, a great discussion this morning. It's been brilliant insights. Thank you.

Back to podcast