Recent years have seen far-reaching changes in medical research, thanks to the advent of data processing tools using artificial intelligence (AI). This technology paves the way to considerable time saved in drug discovery and to more personalized medicine.
At Servier, we are exploring a wide range of AI initiatives with the goal of speeding up certain R&D steps, such as the search for molecules of interest and therapeutic targets, always for the benefit of patients.
Today’s pharmaceutical industry faces a major challenge. While research costs continue to rise, the rate of new developments is slowing. This can be explained by Eroom’s law, which observes that every nine years, for every billion dollars invested in R&D, the number of medicines authorized is halved.
Comes this question : since medicines are increasingly more complex to develop, and the regulatory context is becoming stricter as time passes, how can laboratory research efficiency be improved?
Part of the answer may lie in digital technologies, and artificial intelligence in particular. In the healthcare sector, the digital revolution can be linked to the explosion in data and to our ability to collect, store and process it, thanks to computers that are capable of analyzing millions of bits of information at considerable speed.
This treasure trove medical data is an invaluable resource for predicting diseases, diagnosing pathologies and improving patient follow-up.
When combined with human expertise, the development in AI is very promising for therapeutic innovation. This is particularly true with respect to earlier diagnosis tools that rely on AI to help detect rare diseases.
AI in support of Servier research
At Servier, improving our capacity to innovate, accelerating it and making it more efficient at serving patient needs are our priority.
In 2020, a data factory was created within the Group to support development of solutions and services that rely on artificial intelligence as a means of leveraging performance in order to boost a wide range of therapeutic projects.
“Our goal is to significantly increase the probability of success for our drug candidates. There is a huge amount of therapeutic data generated by medical and pharmacological research — thanks to AI, we finally have the opportunity to make full use of it.”François-Xavier Blaudin de Thé, AI/Data Expert in neuroscience and immuno-inflammation
That is why Servier has defined an AI-centric data strategy that encompasses all of our research projects.
To serve our entire R&D chain, we have identified Use Cases; each designed to provide therapeutic projects with a set of usable functionalities and services.
This strategy must bring together all the driving forces involving data and AI that are already present in our organization to produce an agile operational model based on co-development. This will stack the odds in favor of our Use Cases and give them the best chance to come to fruition.
We need all our R&D teams to be involved, therefore we have chosen to keep a significant proportion of our data processing and AI tool development in-house.
“Outsourcing can create distance. On the contrary, keeping things in-house makes it easier for employees to adapt to a new way of doing things. It also allows us to maintain control over our digital transformation; in simple terms, this means making sure everyone speaks the same language and is on board with the project.”François-Xavier Blaudin de Thé, AI/Data Expert in neuroscience and immuno-inflammation
Speeding things up for better care
“It is still a bit early to measure the impact of artificial intelligence on real research experiments (In Vitro and In Vivo)», explains Alban. « According to several pharma experts, the full potential of AI will be used in 7 years(1), so its influence will be more precisely measurable. But we can already say that the use of AI on our data is a catalyst: it positively impacts the probability of success of our therapeutic projects, thus accelerating the availability of our drugs to patients. This is why we wanted to invest and involve Servier teams quickly in these disruptive approaches.”
Lastly, AI is a great accelerator in the development of precision medicine. Massive data processing enables us to take account of the heterogeneity and particularities of each of our patients, as Philippe Moingeon, head of the immuno-inflammation pipeline, explains:
“Until now, we were treating all patients affected by a certain pathology with the same medicine. The principle of precision medicine is defining subgroups of patients with shared pathological characteristics in order to provide them with targeted treatments that are better suited to them — and therefore more effective.”
Assisting researchers, not taking their place
In addition to the scale of the AI challenge, and its place in laboratory organization, there are justifiable concerns about seeing, in the near future, medicines created by AI alone.
So, let’s try to put these doubts to rest right away. AI is, and will remain, a digital assistant whose role is to increase the power of our work by performing tasks and calculations that were inaccessible before. Not to replace the researcher.
In other words, it is “a way to make our data ‘smart’ and to harness its value,”Philippe Moingeon said.
By processing millions of medical data using predictive models, we will be able to bring to light unknown and unexpected correlations, a source of fresh impetus in terms of experimentation.
Renan Andrade, lead data scientist, said:
“The in silico approach, i.e., using computer processing, provides additional resources. For example, modeling combinations of molecules or predicting how several medicines will circulate within the body and their effects. This approach also reduces the number of in vivo tests and consequently limits reliance on animal experimentation.”
This is, in fact, the value of AI and its impact on the drug discovery phase according to Philippe Moingeon: “first, choosing the right therapeutic target, then selecting the right candidate drug, and finally, targeting the most suitable patient.”
It’s combining scientific innovation with the power of technology to come up with innovative new treatments for patients with rare, hard-to-treat diseases.
Patrimony : la plateforme IA « made in » Servier
Used in the exploratory stage, the Patrimony platform launched in 2018 helps researchers use patient data to identify key therapeutic targets for the pathologies we are studying.
This is all the more important as we have decided to concentrate our efforts on complex diseases for unmet medical needs that are very serious.
These diseases are still very poorly understood, and the first step is finding therapeutic targets of interest (genes or proteins) that will enable us to have a powerful impact on the disease and/or its symptoms. To date, we have incorporated more than 70 internal and external databases on autoimmune diseases, neurological diseases and cancer.
Our first successful outcome: Within the framework of a profiling study on patients with various autoimmune diseases, Patrimony allowed us to identify and prioritize several innovative therapeutic targets, enabling a first drug discovery project to be launched in 2021.
To accelerate the drug discovery process using artificial intelligence, at the end of 2022 Servier and Oncodesign Precision Medicine (OPM) launched FederAidd (Federation for Artificial Intelligence in Drug Discovery), an international virtual campus for open innovation.
The goal of FederAidd is to create a place for encouraging and centralizing exchanges revolving around AI projetcts applied to the discovery of innovative drugs. Launched in both France and in Canada, it will host healthcare stakeholders from all over the world!
1 Source : Pharmaceutical technology It will take years for AI use to peak in drug discovery and development process; June 2022