Coulter Partners recently talked to CEO and Co-Founder, Francois-Henri Boissel of Novadiscovery on digital disruption and talent in healthcare.
Coulter Partners: Please share with us some background on your goals and vision for Novadiscovery.
Francois-Henri Boissel: Nova’s mission is to help establish in silico trials – the use of mathematical modelling and computer simulation of clinical trials – as the third pillar of a newly emerging drug R&D paradigm to empower and not replace more conventional in vitro and in vivo explorations. The pharmaceutical industry has been defined by these experimental approaches over the past two decades, despite the limitations of this extremely capital-intensive trial and error process in deploying funds to identify and develop promising therapeutic innovations.
Our new paradigm will foster the adoption of engineering sciences as a key component of the pharma skillset – by building computational simulations of upcoming clinical trials for biotechs and pharmaceutical companies to enable better informed decision making and improved risk management. In this way they can optimize their R&D spend to develop drugs faster and more cheaply ensuring the right drugs go to the right patients.
CP: You are at the intersection of life sciences and technology. Do you view yourself as a life science company or a technology company?
Francois: While we certainly sit at this intersection, I see those operating in our space as primarily life sciences rather than technology companies and therefore subject to more rigorous regulation. Tech will always be, however, an efficient means to an end – and that end is to help our customers develop more efficacious drugs, cheaper and faster.
Nova has been founded on expertise primarily in clinical pharmacology, biology and medicine, but our technological knowhow is invaluable in optimizing the entire process to achieve our objectives. First and foremost, we are a life sciences company helping drug developers and we truly understand how clinical development works. We use technology to deploy the services and value proposition that we want to bring to our customers.
CP: How do your customers view you?
Francois: It depends on the customer segment. Large pharmaceutical companies are typically most interested in the technology platform itself. Smaller biotechs, also very important customers for us, want to partner with us so we run those simulated trials for them. Our approach is to be a good generalist and not an expert in a specific disease area. We need to understand how each segment works and the typical hurdles they face so we can provide the right tailor-made solutions. Technology in itself is not the silver bullet to address the failings in R&D productivity of the pharma industry. Companies led by pure tech profiles, without a healthy balance of skills on the founding team, typically face huge challenges. A solid and intricate understanding of drug development is key. This needs to encompass the ethical considerations of designing clinical trials and the potential ethical costs of a poorly designed trial.
CP: There is a lot of discussion about the value of data in the drug development space. What are some of the common misconceptions on data?
Francois: There are many misconceptions around data. Transparency on the ultimate objectives of data use is key. Our goal is to run clinical trial simulation in order to predict as best we can the clinical efficacy of a drug candidate and the outcome of a future trial on human subjects before that trial takes place. This requires access to a very specific type of data that has been generated in a very controlled experimental context. We use data from randomised controlled trials to inform our models (although not to build our models from scratch – the scaffold of our disease models is built on knowledge extracted from scientific articles). We ensure the data is from highly controlled in vitro or in vivo experiments.
The ubiquitous access to patient data that is enabled by today’s technology may have substantial value in other components of Life Sciences, but to predict the efficacy of a drug they have little value. The source of most of this data is real world observations from clinical centres and not reliable to build a predictive algorithm for a clinical trial, because the datasets don’t come from controlled experiments, with randomization of patients and a control group. The key limitation of traditional AI approaches in a clinical development setting is the paucity of available data. If you try to make up for the shortage of controlled experimental data by using real world evidence, then you are resorting to raw information that is just not reliable.
CP: What change would you most like to see to accelerate the impact that technologies like yours may have? …
Francois: If you want to inform models and demonstrate the value of a new product on the market to support pricing and reimbursement discussions with payors, then having access to large cohort data will be helpful. Whether you use these to perform traditional statistical analysis or take more sophisticated machine learning approaches, the bottom line is that large data sets will provide interesting signals. The same logic applies for earlier discovery, where computational chemistry platforms are employed to sift large databases for the right molecule to target. These high intensity workflows can be performed by computers much more efficiently than by humans.
For our purposes, however, easier access to standardized datasets from other clinical trials, especially from failed clinical trials would be the ideal. If the industry came to an agreement to publish detailed information on failed clinical trials for some of the drugs that eventually make it to market, then we would create a virtuous circle of information for the benefit of all. Real world evidence has value for some market access issues but for clinical trials, the one thing that would make the most impact would be wider access to failed trials. The best-case access right now is to summary data and high-level synthesized information. In order to run one’s own tests, much more detail and pseudonymised data would be required.
CP: What about the ethical issues around all of this?
Francois: From an ethical standpoint, I would be in favour of large scale properly anonymised sharing of data – to help inform competitors’ research programs and promote healthy competition.
The ethical costs of human clinical trials are not universally understood and the collective vs. individual ethics of this can be controversial. If you are randomized in a control group or are treated but don’t have the best responder profile, you are taking a risk for the collective benefit. Using modern simulations to ensure the best responder profiles are enrolled in clinical trials reconciles the contradiction of individual vs. collective ethics and reduces the ethical cost of performing those trials.
CP: When bringing talent into your organisation, what do you look for and what are the hallmarks of people who are successful in this cross over health-tech world?
Francois: We look for hard and soft skills in combination. We want to find the brightest minds and best expertise in each field and are very exacting in our selection requirements. We are pioneering a new domain of expertise and need not only excellent tech people but also those who exhibit specific cultural attributes. Important among these are “intellectual humility”, willingness to work in a team and being comfortable with the possibility of failure, first time around. Fail fast and learn quick is a useful mantra. We look for a real ownership mentality – we want individuals who are accountable, who feel invested in the mission of the company, who understand clearly our key strategic objectives and are prepared to make some sacrifices along the way. We sometimes have to take the rough with the smooth in such a complex space. We need resilient, eager team-workers, comfortable with the ambiguity of building a new domain.
CP: How do you balance the “fail fast” mantra with your mission?
Francois: We sprinkle our standard workflow with qualitative and quantitative validation and control points along the way. In the earlier stages of developing a disease model, before it is properly calibrated and validated, these control points will mainly be qualitative in nature but will provide some valuable information. As the model matures and we run validation tests against observable data from in vitro and in vivo experiments, then we switch gradually to more quantitative validation points in a very iterative process. At every step we try, validate, benchmark, revisit and redo things that don’t seem to be working correctly.
Harvard professor, Gary Pisano’s research on innovative cultures finds that “a tolerance for failure requires an intolerance for incompetence.” We need to be super-stringent in controlling the entire process with the right stage gates, while being open-minded to failure and learning fast.
Human capital will always be our number one asset. We invest a lot of energy and resource in building a team that can reach their full potential over time. For high calibre individuals joining an innovative company, it’s not all about salary but the exposure to responsibility and day to day learning, especially for the younger generations. They are inspired by a clearly stated mission and a social impact that is tangible.
CP: And how have you coped with new working practices and scaling the business in the challenging times we are all facing?
Francois: We were already well equipped with technology for remote working, but the impact has been seen in maintaining morale without the usual buffer of collocated team dynamics. In a management context, now more than ever it is critical to take care of every individual and focus on their well-being, despite the remote constraints.
As for remote hiring, we have had adapted well! We are scaling fast and continuing to execute our business plan aggressively. Four recent hires were interviewed remotely, and onboarding has also been a novel remote process. We have implemented virtual after-work welcome drinks, where individuals on the team introduce one another. This reassures our newcomers of existing workplace bonds and reinforces cultural values very successfully!
For more information about Novadiscovery, please visit www.novadiscovery.com
Basel Healthtech Conference 2024
25 January 2024 – 26 January 2024
Five Shifts to Digital Personalized Healthcare
10 May 2023
DIGITAL LEADERSHIP SERIES 2020 - Tanja Dowe, Chief Executive Officer of Debiopharm Innovation Fund S.A.
14 July 2020
DIGITAL LEADERSHIP SERIES 2020 - Dr David Brown, Co-founder and Chairman of Healx
08 April 2020