13
Examination of witness
Professor Florian Markowetz.
Q24 The Chair: Good morning. Welcome to this Select Committee on Science and Technology. We are undertaking an inquiry into innovation in the NHS, personalised medicines and AI, and we are very pleased to have as our witness for this first session Professor Florian Markowetz, who is associate director of artificial intelligence for Cancer Research UK. Professor Markowetz is coming in online from Germany this morning.
Professor Markowetz, thank you very much for joining us this morning. By way of an opening statement, could you give us a brief overview of your research? Your group’s stated mission is technologies for doctors to make better decisions faster, especially in oncology, using genomic and AI approaches. Could you set out for us, please, where AI in cancer genomics is already making a difference to patients? How close are we to this sort of approach being routinely used in the NHS? We are very keen to hear your answers.
Professor Florian Markowetz: Thank you very much. I am afraid we are not very close at all. There is a huge gap between academic innovations, where there are lots of approaches and very good ideas, and where we have prototypes that we have tried out on academic datasets, and getting those into routine patient treatment—we are very far from that. There are many reasons.
You have a very broad remit here for your inquiry, and I thought about where my own experiences are best-suited. I can talk about three particular topics that show where the obstacles lie and why we are not as far as we maybe should be. One is the clinical buy-in from the NHS and the hospital trusts, which is limited. The second is the obstacles we face with entrepreneurship and interfaces to the trusts, the genomic laboratory hubs and resources like GEL—Genomics England. Finally, I am an academic, and I can talk about all the issues we have with the academic reward system, which prioritises novelty over implementation.
The Chair: Okay. We have many more questions for you. I am sure you will enlarge more. But your main point is that we are quite a long way from implementing.
Professor Florian Markowetz: I believe so.
Lord Patel: Good morning, Professor Markowetz. Can you enlarge on the answer you just gave by giving examples where the research is already there to be implemented in patient care?
Professor Florian Markowetz: I have a couple of examples where we are actually there. My best example, where some of my own research is actually in patient care, is not genomics. It is about early detection of cancer from a device that my colleague Rebecca Fitzgerald has developed, which is a little sponge on a string. You need pathologists to analyse cells that are collected with this. There was a very clear clinical need. There was a very clear bottleneck; we simply do not have enough pathologists.
Lord Patel: Sorry to interrupt, but are you talking about cells for oesophageal cancers?
Professor Florian Markowetz: I am sure you know somebody with reflux, which means there is acid coming up from the stomach into your oesophagus and the tissue changes. The question is: how do we see whether that might lead to a cancer or not? People usually use endoscopy, but nowadays there are newer methods, which are little sponges on strings, which you swallow and they get pulled up.
The big bottleneck was that we do not have enough pathologists to look at the cells that got collected. There, AI was really a breakthrough because all these AI image analysis tools really enable this technology to scale up. It is now used across the UK and is being rolled out into the US. That is a very clear example where AI helped. In genomics, I do not see the same type of examples, even in my own research.
Lord Duncan of Springbank: There are different state systems where we have treatments, and you are saying there are bottlenecks in the ones which you are experiencing. Are there particular state systems which are better placed, whether the US model, the UK model or some of the European models, or are they all broadly in the same place?
Professor Florian Markowetz: There are many commonalities, of course, but they all have different strengths. I can give you an example. On the genomic side, I am the co-founder of a start-up called Tailor Bio, which uses technology that we have developed in my lab to understand very complex genomes. If you look at ovarian cancer, oesophageal cancer and pancreatic cancer, these are very deadly cancers with very chaotic rearranged genomes. Patients die very quickly and we do not have biomarkers.
My lab has developed an AI computational and pattern-matching approach to find these biomarkers. We had a start-up to get this into the NHS system. That was our plan. We have given up on this. The problems were that, on the clinical side, we do not have the same buy-in that we get in Spain or Germany, which is where we currently develop these methods. When we tried, even in Cambridge where we have very good contacts, the responses we got were that they valued patient safety very highly and, as a result, they were very conservative in what new technologies they would even try out.
I am sure in other countries they also value patient safety, but they also feel a much greater incentive to innovate and try out new methods. When we had the same technology in Madrid, we immediately had half a dozen to a dozen doctors who wanted to try this out, something we had never achieved in the UK.
We also faced issues with, for example, integrating this with Genomics England, which is a massive data resource and lots of trusts contribute their samples there. The major vehicle to translate my research is the start-up, so for me to do my translational job well I need interfaces. I need ways to be reimbursed for the services that the start-up provides. With the trust in Cambridge and also Genomics England, we never really figured out how this would work. There were lots of academic access points; for example, GEL has lots of programmes for academics to work with it. It is just that when you have a product that you now want to roll out for patients, nobody really knew how to do this.
This experience is maybe three or four years old, so it might be very different now, but as a result, we had to move on. Our product was not viable in the UK and now we are developing it in Spain.
The Chair: Okay, that is very interesting to hear.
Q25 Lord Willis of Knaresborough: Good morning and thank you very much indeed for joining us. I am particularly interested in the way in which we actually convert our research to actual use within our hospitals, with ordinary people and with the health service. Our Chancellor of the Exchequer is meeting today to discuss how on earth the UK can offer more resources to the research element, not mentioning in one single word what we want to do to our patients. What areas of research and development in genomic medicine and AI do you think have the biggest potential to benefit patients? What should we do to actually make that happen?
Professor Florian Markowetz: I cannot speak for all of it. These are huge fields. I can tell you that in genomics one of the big opportunities is to find new biomarkers. There are lots of drugs which are given very broadly. All the chemotherapies, which are generally just poisons, just poison the tumour quicker than the patient, hopefully. They are given often without any indication of whether they might work or not. There are no biomarkers for patient stratification.
For genomics, using genome-wide signatures of different types has a huge potential. To get this into the clinic you have to overcome incredible obstacles. You ask about academics. The academic reward system is not built for translation. My career was built on writing papers where in every paper I claim that I have developed something new; novelty was the most important thing. Novelty is not the most important thing for translation. In translation, you want well-established methods and you have to get them to work. It is much more about engineering, setting things up in the real world, rather than reinventing the wheel.
As an academic, I am rewarded, and my post-docs and students were all rewarded, by having yet another method, yet another approach, but there is very little reward for actually implementing it in the clinic. I am doing it now, and I can because my career is fully developed. I am a full professor. I am well funded. I can do this even though the whole system around me does not really reward it. That is the biggest obstacle for academics.
Lord Willis of Knaresborough: Out of interest, given that so much of the research seems to be translated in the United States after it has been done in the UK, do they have a different approach to then delivering it to patients, or do they do the same things? Do they have the same problems that we have, and if they do not, what are they doing that we should be doing?
Professor Florian Markowetz: The problem that I just described, they have too. In the US, even having a company is often seen as being out of focus, out of scope for a researcher—you should rather focus on your research. The one thing the US does much better than anybody else is that it is much more rewarding to have a start-up in the US than in the UK. The terms that we got are crippling from the university and from our main funders. We had to negotiate for years to get an exclusive licence for our own patent. So, even if as an academic I decide I want to translate, and the vehicle to translate is a start-up company, then yes, I am much better off trying to do this in the US than in the UK, because here it is just even more work and I get very little out of it.
You see, of the people who do start-ups, 90% fail, so you do not do this to get rich. You do it because you have an exciting discovery and want it to be useful for as many people as possible. The people in start-ups put lots of effort into building this, and we do not feel that we are getting any reward out of the climate here, and we have to overcome an incredible uphill battle to get anything done.
Lord Willis of Knaresborough: Finally, in the UK, given that there are so many different elements of research going on, do you think it is important that in fact we choose, even as a committee, what will be the key areas, in order to make sure that you can put resource in to deliver those to patients? If you have a lot of things going on, clearly trying to do them all is just not going to happen.
Professor Florian Markowetz: That is correct. My own view is that the strength of the UK is the discovery science—the basic science. Having very creative people come up with all kinds of new ideas—a big creative chaos—is a real strength, and I do not think you improve this by prioritising some areas over others. If you look at some of the major breakthroughs, they were serendipitous, so in a way you are best off just broadly funding all kinds of blue-sky research.
When it comes to translation, the difficulty is that there is a selection process already, which is through start-ups. If you look at VC firms that fund start-ups by business models, they make that selection that you just asked about, because they will see what might have the biggest impact, where are the big markets, and which product is even viable. As you can see, I do not have much experience with policy-making, so I am not completely sure if you are helping if you restrict this too much.
Q26 Baroness Nicholson of Winterbourne: What you are describing, Professor, is essentially a system failure, is it not? As we know very well, if you just take the examples of Germany, the USA and the UK, the systems are very different indeed. Here we have the NHS, which is our major system. Do you have any suggestions or proposals through which we could, as a committee, offer a different thought of a system bypass in a sense, which is a different channel that would pick up the research and major with it, not necessarily through individual trusts or even through NHS England? What would you suggest that we could put forward that would enable your research to flower here?
Professor Florian Markowetz: That is an excellent question; thank you very much. The key challenge is with the NHS. While it is a national system, it is very fragmented. If I talk to the Cambridge trust, I do not know if people in Peterborough or King’s Lynn will be able to profit from the same developments that I have set up in Cambridge. The big improvement to the system would be to harmonise. For example, just with data access, there are just seven genomic laboratory hubs, which is good—this is already starting to be harmonised—but you need to fund them to be able to do technology development, you need to enable them to talk to each other very simply, and you need to, in a way, force the hospitals to transfer as much data as you can.
For example, put this up as trust success criteria. Did we give some harmonised central data structure—all the data that we can? If you did this, this would really enable researchers like me to work with routine clinical data. This is where the big chance is that all these AI approaches are not being applied to some fancy academic data, but they need to be applied to the type of data that are actually currently helping patients. This interface with the clinics—not just a single one but very broadly—is currently missing, and that would be the most helpful thing I can currently think of.
Lord Duncan of Springbank: To follow up from Lord Willis’s point, our Chancellor is currently talking about what we should be able to do here, and there is a discussion that we should perhaps be aligning ourselves more closely with what is going on inside the EU. But you seem to be suggesting that we should perhaps be looking across the Atlantic at what is happening in the US. Would you characterise that as accurate?
Professor Florian Markowetz: I do not want the American health system here. I am very fond of the NHS and I do not want the American one. For the US, look at their entrepreneurship. This is what you can learn from the US. The US is also fragmented, maybe even more than we are. I think they just have much more money in their system, which helps them to solve lots of problems. If you want to steal an idea from the US, help academics to set up start-ups in the UK at terms that are easily viable and which make it easy for them to translate their research into a viable product. That is to take from the US.
On what some European countries are better at, for example, I have contacts in Germany where the research institute is part of the hospital. I have daily problems, being a university member and then interfacing with the trust, having to look at these data access points. The infrastructure is only now being built. I think I am a trailblazer in Cambridge, just trying to get to routine data. This is something other countries are much further ahead on. If I was doing this in the west of Germany—my contacts are from a hospital in Essen—I would just have this data every day without problems. The integration between research and hospital is better there. If you want to steal something from Germany, take that.
Q27 Lord Drayson: Professor, first, congratulations on founding a start-up company. I completely appreciate the challenges which you have identified, so it is terrific that you have done it. It was interesting to hear you say that the terms from Cambridge were crippling, because often Cambridge has the reputation of being one of the better universities in the terms for academics, but that clearly was not the experience that you had.
Professor Florian Markowetz: Sorry, I was not clear about this. I am in a Cancer Research UK institute and I do not want to bite the hand that feeds me; the terms were from Cancer Research UK, not the university. You are completely right; some of my colleagues argue that we should be better off just always going to the university, but we cannot if it is CRUK funding.
Lord Drayson: That is really helpful. I think you probably gathered from our questions that we are really concerned about the failure that we have in translating world-class academic research into effects for patients and growth for the UK. You mentioned in your opening statement the barriers to clinical buy-in within the hospital trusts. Could you speak to the culture in the NHS trusts towards commercialisation? If an innovation is to be widely adopted, it has to be commercialised. Can you give us some flavour of what that is currently like?
Professor Florian Markowetz: Currently, it is completely unclear how you would actually do this for like a company like mine. On getting reimbursement for our services, it is just not clear how this would happen. When we approached, for example, the NIHR to organise trials for some of the biomarkers we had, the feedback we got was very negative. It was pretty much seen as if it was not very important or nobody really seemed to care. This might be a feature of our biomarkers maybe not being very good, but—
Lord Drayson: I can assure you that it is not.
Professor Florian Markowetz: We do not know, because we could never do the trial in the end.
Lord Drayson: From the evidence we have had, that is not unusual.
Professor Florian Markowetz: After having started the company and sorted out all kinds of problems, now, when we have got to the phase where we really want to engage with the health system, I would like to see the doors a bit more open. It was very unclear how reimbursement would happen and we saw very little encouragement and excitement from the medical community.
Lord Drayson: Are you saying that there is not anyone who has it as their responsibility to affect the pull-through?
Professor Florian Markowetz: This was all three or four years ago. What happened in the last couple of years, at least in Cambridge, is that a director of innovation was hired at Cambridge University Hospitals and suddenly it set up systems where it collects routine data, which, in principle, I can access. On the hospital side, lots of things are going in the right direction. But the positive examples I currently see are all on the academic side, so I think that it is much keener to engage with academics now. It has a software company, which just appeared in the last 12 months, that takes care of the computation on the hospital side, which is excellent. So now there is a platform for me potentially to deploy my new innovations. But I am an academic, and the standards for academic code and academic software are not the same as for commercial software. So, there is still this gap where they would like to work with me but it is not clear to me how I get my own code to a viable product unless I have another start-up, and for the start-up it is not clear how they would engage with that.
Lord Drayson: Which platform or company is that?
Professor Florian Markowetz: A company called Newtons Tree. We have very good relationships with it, which is excellent. There is a second platform by GE Healthcare. Cambridge, at least, is really invested in having the data better sorted and more accessible, which is an amazing development. But there are still lots of issues. I think I am the first to try this out, which might be part of the problem as I see all the teething problems there.
Lord Drayson: Let us turn now to Genomics England. When Geonomics England was set up after the Olympics in 2012, part of its mission was to kick-start a genomics industry in the UK to create a flowering of companies such as Tailor Bio. What has been your experience of engaging with Genomics England?
Professor Florian Markowetz: On the academic side, very good. It has a trusted research environment. Currently, it is applying the technology we developed to 15,000 cancer genomes. That is an incredible resource and I am very thankful that we can do this. On the commercial side, which we have to talk about for translation, the experiences were not that positive. Although it was excited about our technology and would have liked to use it, there was just no way to engage. There was no commercial agreement possible. Lots of academic agreements were possible, but nothing on the commercial side.
Lord Drayson: Why not?
Professor Florian Markowetz: I do not think they were set up to do this. The idea we had was to use Innovate UK to show the clinical utility of our tests—because commercial access to GEL is still very muddy, that is the preferred option for most start-ups—and we got two Innovate UK grants and really did our homework. But in the end, we still did not get an agreement done with GEL.
Lord Drayson: This is very strange, because part of the reason for setting up GEL was, as I said, to kick-start the industry. It has been going since 2012, and you are saying that it still has not got its commercial pipeline?
Professor Florian Markowetz: I am saying that with the one example that I know of, we just could not get it to work.
Lord Drayson: What do you think the Government need to do to urgently fix this pipeline? We have this great resource that can improve outcomes for patients. It is not getting to patients because there is no commercial pipeline to enable people to scale. What do the Government need to do?
Professor Florian Markowetz: Tell them to build their pipeline, I guess. You were surprised, and I am surprised too. It seems like an obvious point that they should have it. They might have it now. Again, my experiences are mostly three years old.
Lord Drayson: When you say, “Tell them to set it up”, within the NHS trusts, are you talking about the management organisation of the trusts or the clinical staff within the trusts or both?
Professor Florian Markowetz: I think that it is both. In the end, you need to convince the doctors to use these things, and you need to build trust. There is a huge issue with all these AI tools. We do not trust them yet, and I do not think we should. But, to build the trust, we have to try to use them and show their clinical utility. There are two parts to this—one is to set up the infrastructure, which is where the leadership at the top comes in. Then, when we are in a position to try these tools out, the other is to build trust with the clinicians on the ground. It is about both of those things.
Lord Drayson: We have had evidence that there is a culture of risk-aversion within the NHS—
Professor Florian Markowetz: Completely correct.
Lord Drayson: —which has come as a result of problems that have happened in the NHS and which has led to policies being put in place that have disincentivised people from trying new things. As a result, this risk aversion is becoming culturally embedded in the system. You are nodding. Has that been your experience?
Professor Florian Markowetz: Completely right. There are exceptions in Cambridge. We have so much research in Cambridge, but seeing people actually interfacing with the hospital and getting things done happens only as exceptions. I could name for you, at the most, two or three colleagues who are really successful and at a level where we should all be, where we have the research and the product and we reach the patient with it. Part of that is academia, because the reward system in academia is skewed, but part of it is just that it is so hard to interact with the clinic.
Lord Drayson: There is an old cliché in business, which is now out of date: nobody gets sacked for buying IBM. I think you get the point. There is a concern that there is a tendency for people to choose the big tech companies, such as DeepMind and others, over the smaller start-ups, which British start-ups tend to be. Has that been your experience?
Professor Florian Markowetz: I think that that is correct. For Cambridge, I think GE Healthcare was its first. Newtons Tree is a British start-up, I think, and it has done really well. But in general, you are completely correct. As I said, it is very hard to get start-ups off the ground. We would be better off if we had a very lively start-up scene—a marketplace of ideas where we have all these start-ups that find it easy to connect to the hospital and we just see which of these approaches work. That would be the best.
Q28 The Chair: Professor Markowetz, can I ask you your views on the role of AI and clinical prediction models? We are very interested in this topic. Last week, we heard Professor Ewan Birney talking about clinical foundation models, which are looking to emulate the large language models. A large language model, of course, uses a large volume of text to train the model to predict the next word, but what we are talking about here is using clinical and genomic data to predict future health events. What are your views on that? Are you optimistic?
Professor Florian Markowetz: There is currently lots of research going into this area, so I think that Professor Birney is completely correct in highlighting it. This is a very promising area that lots of us are working on.
I am very confident that we will have pilot studies where we can show that these technologies work. How we show clinical utility is a much bigger issue. One of the easiest points where you can help patients with cancer using predictions is called the multidisciplinary team. When the question is, “How do we treat a patient?”, there is a whole group of people from different backgrounds—a surgeon, a nurse, a medical oncologist, maybe even a bioinformatician—and they pool all the data together and decide on the best way to deal with that patient. That is a prediction model. You want to predict the treatment that helps the patient best. Lots of AI tools nowadays are being developed based on large language models and groups of agents—they are called multi-agent systems—to try to automate this. The research is there. Even if it works—it will in the end—we face the same problem we have always faced: that it is not clear how to actually implement it in the clinic. The translation problem stays the same, even if it is an almost perfect AI prediction model.
Lord Patel: This is a follow-on to that answer you gave, actually. As I understand it, you said that there are machine models that can do the prediction that normally a clinical team undertakes. How is that machine model tested in reality to confirm that what the clinicians predict is exactly the same?
Professor Florian Markowetz: Excellent. That is a very good point. That is part of the issue at the moment. When I talk about implementation, that is part of it. The current academic models are being applied to data that people just have, so they never test them in real-world scenarios. What you would have to do is clinical trials, where you run this in parallel to human experts. In the first step, the decision would still be with the humans, but you would run the humans and the AI in parallel, and then you compare.
Lord Patel: So that work needs to be done before it could be implemented in clinical practice. Until that work is done, it cannot be implemented in practice.
Professor Florian Markowetz: Doing a trial, for many people, is already a success of translational research, even for drugs. Getting it to a trial would be a major success, and I do not think we are very good at this yet. You are completely right: to then actually implement it, we would have to do the trials, just like for a drug. There is a reason that drug development takes 20 years, because you do a sequence of different trials. In phase 1, you check: is it hurting people? In phase 2, you check: is there any effect? You find the right dose. There are all these steps, which for drugs are very well understood. There are reviews and pipelines, and everybody knows what the steps are. For these AI agents, we have to figure this out at the moment.
The Chair: Lord Patel, do you want to go on to the national cancer plan?
Q29 Lord Patel: Yes, I shall go on from there. On several occasions, you suggested that there were already results of research that could be implemented in practice in the NHS but that this is not done. Why do you think that is and what needs to be done to get it implemented? Is it workforce? Is it a structural issue? Is it guidelines? What is it?
Professor Florian Markowetz: I wish I could point at one individual thing to answer that question. There are obstacles everywhere. Just to give you an example, it starts with academics having great ideas about new technologies but, for translation, these technologies need to leave academia. Already at the moment, the support you get for doing this for AI and for software is very limited. There is a wealth of academic research, but it is not code that you would give to anybody else. It is just very bad code. It is academic code run by people who are not full-time software engineers.
Lord Patel: I think we are at cross-purposes here. I will come back to the code in a minute. My question related to the comments you made on more than one occasion, which were that results of the research that clearly indicated they would benefit the patients were not being implemented. The question I ask is: why do you think that is?
Professor Florian Markowetz: Sorry, I was actually talking about this. We are talking about AI models, so we are mostly talking about code. Part of it is that we do not have a pipeline to take these results out of academia into the clinic. The very first step is not supported, which is getting it to code that can be applied to people. There are rules about software as medical devices. They are not very transparent. It is my field, but I am always very confused about what the current regulations are and how I move forward. It is very opaque. That needs to be clarified. To scale this process up, we need the same kind of steps and understanding we have for drugs. We have to have this for software as medical devices. The whole process is not very well worked out at the moment so, unfortunately, I cannot really point to one individual thing that you have to fix.
Lord Patel: Okay. My next half of the question relates to the national cancer plan—you are familiar with it—and one of the things that was said about it: that it had an ambition that everyone who needs a genomic test for cancer can receive one. That refers to what you talked about earlier on; presumably the cancer genomic test is a biomarker. How do you think that could happen and why would it not happen?
Professor Florian Markowetz: There are many different genomic tests you can have. The simplest one gives some results about the person’s or the tumour’s genome, and you look at a single gene or a single mutation, like in BRCA—the Angelina Jolie gene—and you say, “Oh, if you have that mutation, we do this. If you do not have the mutation, we do that”. This is a very gene-centric thinking which I think you will find very easy to implement, testing individual genes for mutations. You can get this for many patients. In cancer, very often this is not enough. When you talk about genomics and AI, you do not talk about individual mutations in the genome; you talk about much more complex, broader patterns. It is at this stage, which is the one we really need, where sometimes we simply do not have the right mindset or understanding. In Genomics England, many of the people who analyse the data come from a genetics background of rare diseases. They look at individual genes. In cancer, you need a very different way to understand these genomes, and that is not there yet. You would build this through training and through hiring different types of people and investing in more cancer-centric technologies rather than the original genetic tests that were done for rare diseases.
Lord Willis of Knaresborough: Is not AI the key to resolving that? You will be able to get so many things done.
Professor Florian Markowetz: In a way, yes. This is what we do. AI now is a very broad field, right? There were these foundation models and the large language models that Professor Birney talked about. It is a huge field with many different models, but they all are pretty good at prediction. You are completely right: as a whole, this global picture of predicting something for patients from a whole genome is exactly where the AI fits in.
Q30 Lord Drayson: Professor, you said earlier that the academic rewards are skewed towards novelty production of patents and not implementation. I am involved in running an AI company, not in healthcare at the moment. You see how the explosion in papers generated around AI leads to people focusing on novelty, but the vector whereby innovation translates into application—for example, in software development—is through the practitioners themselves. The challenge that the UK has in this is that the software platforms, which are the vehicles for that implementation to take place, are all US-owned, effectively. Medical innovation in the use of AI is astonishing. You probably saw the news about the Australian man who treated his dog with a novel cancer vaccine, which he used AI to develop, and he was not a medic.
Professor Florian Markowetz: That is excellent.
Lord Drayson: Yes, it is all over the AI news at the moment. What is your advice on getting the academic clinical research community to culturally act as practitioners and to be incentivised—like coders are—to learn how to vibe code and then to use the AI to increase the productivity of their code development? What are the mechanisms—I do not know—that the colleges could use? How can we change the culture within academic clinical research such that the practitioners are the methods whereby this gets translated to patient use, and how can we create a critical mass of UK companies not relying on the American tech stack to be able to do this?
Professor Florian Markowetz: That is also a very big question. I cannot solve all of this. On the MD-PhD programmes, if you had some just-for-computational PhDs, that would help a lot. These are PhD programmes for medical doctors who want to engage with research. I think historically they mostly go to experimental labs, such as biological ones. If you had specific programmes simply for computational people, this would exactly give them the skill set you are talking about.
I think there is something from university leadership. If you look at academic careers, we depend on people getting individual fellowships. There is a variety of places: CRUK gives out many, I am sure Wellcome does, and there are the UKRI leaders’ fellowships. The way you get evaluated is by having lots of novelty in journals of a particular flavour. There is a certain type of journal we all love, and they all want to see new things. You could steer this process—the evaluation criteria—to say, “Actually, what we are judging you by is how useful this is to a patient and how much you work to get it to the patient”. If this was a feasible route—I am not saying it has to be the only one—I think you would get much more translationally minded people into the business. I give this talk on problems with translational medicine, and usually to students. At the very end I say, “It is good to see you all excited about what I just said; you need to be very careful because it will torpedo your career if you take this advice on fully. You still need to play the game, and the scientific game is currently about new stuff, novelty, reinventing the wheel”.
The Chair: That is a very interesting message.
Q31 Lord Duncan of Springbank: We are conducting an inquiry, and obviously at the end of that we look to present recommendations to Government about how we experience the problems; you have already touched on several of those. If you were looking for some of the headline issues that you would want to see in our report, what would be the top three things you would want us to be communicating to Government in the area in which you are best equipped to speak?
Professor Florian Markowetz: The top three—I am looking at my cheat sheet as we speak—are, first, that we need to harmonise data, access to routine data. It must be much easier for academics to reach this so we have the right type of models. For all the players in there—for the trusts, for GEL, for everyone—there must be clear pathways for commercial contracts and reimbursement. On the academic side, if you can help with unskewing the rewards system, that would also be great.
Lord Duncan of Springbank: Just following up on that, presumably the Government themselves should be aware of these things now, but they are not doing that. Why do you think that is?
Professor Florian Markowetz: I cannot comment on that; I do not know.
Lord Duncan of Springbank: Fair enough. Thank you.
The Chair: Professor Markowetz, you have answered a lot of our questions. Thank you very much—we are very grateful. It has been very instructive hearing your views. We are now going to suspend this session and prepare for the one following. Thank you very much indeed for joining us.
Professor Florian Markowetz: Thank you very much. It was a pleasure to be here.
