Watch the video summary for The Rise of Precision Medicine.
Summary of key points from our panel discussion.
Celine Pallaud, VP & Global Head of Biomarkers & Diagnostics, Novartis
Celine is the global head of biomarkers and diagnostics at Novartis and is responsible for its Precision Medicine strategy across areas, which include investing in early diagnostic solutions and biomarker activities in key programs in order to maximize the collection of patient data.
Dmitry Korkin, Director of Bioinformatics & Computational Biology & Professor of Computer Science, WPI
Dmitry’s focus is interdisciplinary research which lies between Machine Learning (ML) and life sciences: developing new ML and bioinformatic approaches to better understand complex genetic disorders and gain insight into infectious diseases. His work leverages various types of biological data to understand molecular mechanisms behind disease and give insights to clinicians for better decision-making and new therapies.
Moderated by: Craig Steger, SVP of Life Sciences, Outcome Capital
Craig leads the diagnostics in life sciences practice at Outcome Capital. He has 20 years of experience within the diagnostic industry, running business units with emerging technologies within large companies, and on the M&A side, with large companies pursuing a growth strategy and startups commercializing to exit.
Can you comment on how biology and technology are being bridged, especially over the last few years? Is it about measuring the data or the biology?
- Precision medicine has been undergoing an evolution and transformation as we have been following the evolution of the technology to help us understand diseases. For example, with cancer – through the technology, we have been able to identify different types of cancers by understanding the molecular essences of different types of lung cancers. As technology evolves, we are gathering more data and we now have huge amounts of data. In the past we had hypotheses and testing. Now we can ask – how do we better understand and maximize this huge amount of data – and the impact and potential insights from this is important for patients.
- Precision Medicine is now a multidisciplinary science. We must understand the technology, analysis, and processing of data to get to new hypotheses from this. Life sciences are on the verge of making breakthroughs on high throughput experimental data. Both components (technology and biology) are Interconnected, complementary to each other, and inform each other. They lead to improvements based on feedback. We’ve started to see this in imaging research driven by the data. In the future we’ll see this strengthening, with the feedback loop in these disciplines.
What types of technological advancements have driven the building of the bridges between technology and biology?
- We are making tremendous technological advances, not just in computational methods development, which is key, but also in deep learning, AI, and ML to improve predictions in various areas of medical research. We now have more streamlined computing technology. In addition to common technologies, we also now have less common technologies coming into play – such as visualization technology – mixed reality technology. We can go beyond a monitor and visualize large data sets and work with data sets more intuitively using mixed reality devices. This is an Important contribution to the classical components of software and hardware development for better computing.
How has the evolution of biology helped build these bridges?
- We are closing the loop with technology to generate knowledge. The collective technology brings it together – looking from the cells, to the organs, to the body. From the patients’ perspectives – by bringing the technology to the patient we can get better diagnoses with technology and monitor treatments. In the past this was not possible and this is a revolution. It is technology generating knowledge, generating insight, and having impact.
With digital pathology platforms and the AI used with those systems, and the ability to multiplex at a higher level, can we now dig into other disease states in addition to common ones?
- There are multiple applications. An obvious application is oncology. These approaches, combined with ML and AI, are helping with more accurate diagnoses. The benefit to patients is that we can provide better treatment. This applies also to neurological disorders. The combination of imaging approaches and these functional activities helps us to better understand disease.
- Algorithms helping generate faster diagnoses is something that we should look carefully at. We are already using a multidisciplinary approach with genomics, imaging, proteomics, etc. – these are all tied together. We need the combination of these multiple factors and apply many lenses for a better understanding. This is relevant for the subtypes of cancer within one type of cancer that I talked about earlier. We are able to get to more and more small patient populations within a specific definition of a disease. We’ve been able to look into subsegments of patient populations.
How has the global pandemic affected precision medicine?
- There are multiple aspects. This is because we need data. And what is the source of our data? We need samples from patients, and if a patient can’t get to a hospital to give a sample, we have been challenged to rethink how we are approaching precision medicine and the tools that we are using. How can we bring precision medicine to the patient rather than the patient to the hospital? We have learned that we need collaboration and joint efforts to accelerate things. We have existing technology that can be leveraged by bringing collective knowledge together to be more innovative. We have acquired fabulous knowledge during the pandemic.
We’re seeing the accumulation of data in EHR records and we’re able to mine the data – how are you seeing this happen now and how are you using that data?
- Accumulation of data from EHR records and data mining data is one of the bridges we’ve been talking about, with a goal of using data in a patient-centric way. We are merging cell specific data and tissue specific data with imaging which is organ specific. And then merging it with EHR clinical data to look at patients from the cell to the entire organism perspective. The EHR is one of the missing links. It’s a great example of being a catalyst for the medical and research communities to unite.
- We are now working on understanding COVID, which has a complex pathogen interaction, at a personalized level. Why is this related to precision medicine? Now we can look at people who are potentially infected or are infected and see how they respond to treatment. Mutations of the virus differ from one patient to another and can be looked at through the lens of Precision Medicine, not just at the patient level, but also at disease level.
With EHR information, there are many ways to gather the data and we have a lot of data, but do we have a lot of dirty data?
- There is noise in the data as data is generated by humans and there are some natural errors. We are developing data analytics and ML designed to minimize and detect errors and correct them. Whether scans of notes or direct input into computers, the more digitized data that we have, the more accurate the ML becomes.
Can you tell us more about how precision medicine is treating various other diseases (aside from cancer)?
- We are studying many diseases and getting the feedback loop from data mining to ML methods, which has a direct impact because they use common principles in addressing genetic disorders. Neurological disorders, cancers, autism spectrum disorders, depression, type 2 diabetes – these complex diseases often are yet to be accurately diagnosed and cured, but we do see common themes for computional approaches. Of course, there are specific differences, but overall, the whole field is moving similarly and sharing knowledge across different disciplines.
- In the case of neurodegenerative disorders, it’s a slow onset disease and the field is still evolving. We are making use of emerging technologies to understand more at the level of the function. And we have clinical assessments – when we combine the two together, and follow the patient, this allows us a better understanding of the disease. Oncology has been an area where we’ve had to do precision medicine, but now we need precision medicine in all other disease areas. We may not be fully ready to embrace it. We must early on identify disease and do early biomarker research and combine these. ML and AI will really help.
- The more data we get, the better biomarker discovery we have, no matter the disease. Experiment, gather data, generate insight, drive impact.
If you were starting of starting a company in precision medicine, is there an area that you would start in or not?
- No one path is easier than any other, and there is still a lot to discover and there is a lot to be done. With oncology, for example, there is still a lot to be done even though we have tools, because we are now moving to different treatments based on gene therapy. Follow your heart, it is hard work. Don’t underestimate the importance of collaboration with academic institutions, pharma and diagnostics companies.
We have data from patients and physicians and we’re trying to marrying it with what we have in labs. It’s not always an easy link from lab to clinic – is it just a question of refining the data?
- Depending on the disease, we may or may not have an animal model. But if we have an animal model, usually we are curing them, but with patients it is a different reality. What we see in clinics versus what we see in labs – it is always difficult to model a disease. Animal model versus clinical studies are different. For compounds entering clinical studies, only about 11% make it to market. We need to demonstrate a large benefit, and that is the nature of drug development. There is a large attrition rate from bench to bed.
We’re using wearables to collect data. How do you see us collecting that data and being able to mine the data slowly over time? Does that add noise into the system as you add a large number of people?
- Generally, the more data you have, the easier it is to clean out the noise. The data is not random and therefore you are better able to see trends and patterns. The use of wearables will greatly improve our understanding of many conditions. Wearables are worn daily and for extended periods of time. It’s difficult to gather this type of data with diagnostic tools. The challenge is to marry a diagnostic snapshot with longer term and more variable data from wearables. This is why we are still developing ML & AI methods – which are facing challenges of mixing data that is typically not mixable. It is the big challenge that scientists are addressing.
- We are entering a time with many issues of data security and privacy. We’re developing secure protocols and identification protocols and data sharing policies. We see a rise in this research. A lot of groups are working on this problem. As we evolve, the data mining and data processing methods will drive the evolution of data sharing and privacy policies.
How do you see Precision Medicine impacting people’s lives day to day in the near and long term?
- Its impact is in multiple aspects. Personalizing early diagnostics that allow us to identify the condition and the extent to which the condition is worsening and be able to intervene early. Also, the personalized approach to treatment so that we can identify the treatment that the patient will be most responsive to. In complex diseases, there are multiple genes and sets of genes being affected by the disease. Between patients, differences can be quite drastic. Being able to determine which treatment is the best for the patient by looking the patient’s progress or lack of progress through time is helpful and allows us to adjust the treatment.
Treatments and tests over time are often expensive. How do we control or manage these costs?
- Treatments can be expensive. Cancer is an example which can be applied to other disease areas. By being able to monitor disease with very specific tests, when we are able to identify that the disease is at a certain threshold for the patient, we can get them off treatment. This benefits the patient and the system. What you pay for monitoring can shorten total treatment time. We aim to do this in every disease. It’s about the sensitivity. If we can look at one cell to change treatment, we can shorten hospital time, which is a cost to the healthcare system. It’s a long-term approach that drives impact.
- From a bioinformatics and computational perspective, we are coming up with more and more intelligent diagnostics systems. These systems can say that in the long run, if you spend x dollars to do this, it is good enough. It allows us to manage the length, quality, and cost of treatment and diagnostics. We are merging experimental with computational approaches because computational is much cheaper to do. The evolution in technology is becoming cheaper and cheaper to allow longer term diagnostics to be cheaper.
Can precision medicine speed up the process of drug development to help small biotech and big pharma evolve in their thinking?
- We now have the agility to screen much more quickly with a large library of potential compounds, and the abilities of our computing systems to understand the potential of compounds for a specific disease helps us go fast. In terms of animal models, we can go quickly to translational activities by using EHR and the biobench. Many countries and institutions are collecting samples from patients over time that is combined with the EHR. From the biobench, we are asking – if I want a patient with these specific characteristics can you identify this patient – so we can recruit our patients much more quickly through collaboration. We are collaborating a lot with diagnostic companies, institutions and startups to help accelerate things.
Speaking of startups – there’s been big evolution in Precision Medicine in last 3 to 5 years. For young entrepreneurs, are there controversial areas that you might recommend that they stay away from?
- Not quite. We are very curious about what they can bring. There may be some steps that need to happen before making their claim. When we start to work with them and give directional guidance, it’s up to them to follow it or not. Most of the time they come back after getting feedback, which is extremely valuable. We can partner to maximize the value of samples and generate data that can advance the field.
What are the trends that you see in the next 12 to 18 months?
- We see a lot of enthusiasm within research groups because scientific communities are eager to help us get back to regular life and prevent future pandemics. We see a boost in developments in the computational sphere – which happened because everyone was at home. We should reframe our thinking – to think about areas we should be looking at. Especially in the area of diagnostics – using blood and liquid biopsies. The case of Theranos has greatly impacted science and businesses – people have become skeptical. But there is a huge potential in using these types of diagnostics and going a level deeper by looking at epigenetic biomarkers, things that dynamically change in our bodies, not just our genetic background. We are seeing how much these additional levels can inform us, and we can streamline our analyses which is possible with new technologies.
Any last comments or thoughts?
This is a field that has not finished discovering new things. Precision Medicine needs more collaboration, joint partnerships, and co-development. We talk a lot about data, and I believe that the data should not be owned by an entity, but on the contrary be shared because it is for the improvement of humankind and our healthcare systems.