Amidst economic uncertainty, high interest rates, talent shortages, and rising capital costs, artificial intelligence (AI) is top of mind for many industries, governments, and academic institutions. While there is a long list of challenges that the health care sector must overcome, there is reason to believe that through AI, we are on the cusp of a great era of discovery that could fundamentally change the field of medicine.
In pharmaceutical research and development (R&D), AI is already delivering on its potential. AI and data analytics are driving breakthroughs that enable us to predict patient responses, increase the likelihood of clinical trial success, and determine individualized treatment plans for patients. With AI, we are breaking new barriers to unlock previously undruggable targets and bring forward new therapies for patients who currently have no treatment options.
At Sanofi, leveraging AI to empower drug discovery and development is having a major impact. Our key AI models in small-molecule drug discovery are achieving more than 80% prediction accuracy–and they are constantly improving with the use of active learning. Ninety percent of our disease targets are credentialed using single-cell genomics and 75% of small-molecule projects are enabled by AI and machine learning (ML) compound design. We then create virtual patients to drive in silico clinical trials and, finally, genomics-based precision medicine will help us achieve patient stratification.
We are using advanced active learning approaches, improving AI model training, and requiring less data to train our models. AI learnings are highlighting key structural elements to guide design cycles, making them shorter and cheaper, and resulting in higher new molecular success rates. We are increasing the number of clinical trials by 50% and, to date, have quadrupled our pipeline value between 2019 and 2023.
We are in constant contact with the innovation ecosystem, adopting a drug discovery “without borders” strategy. Twenty-five percent of our projects entail working with partners, which has doubled research productivity as measured by dollars spent per clinical candidate and doubled our first-in-human entries.
Additionally, the way we operate is being profoundly altered. Decisions have shifted from an annual retrospective reporting capability to a dynamic prospective decision intelligence approach, linking strategic choices with operational decisions and seeking to enhance our feedback loop.
It’s clear that we stand at the crossroads of a great expansion in medical discoveries, but to take full advantage of AI, there are several challenges that will greatly impact the pharma industry’s ability to unlock potential.
AI regulation
Regional differences in regulation will guide restrictions on where AI can be employed, standards, and what constitutes high-risk applications.
Concerns about data quality, security, privacy, and trustworthiness have all threatened to slow the uptake of AI. Alliances and organizations are emerging to help companies self-regulate.
Strong data foundations and governance will be critical to prevent vulnerabilities as many companies move to operationalize AI across their enterprises.
Unintended effects of pricing restrictions
The unintended consequences of new pricing policies could diminish investment in promising R&D candidates. For example, the Inflation Reduction Act contains what some have called a “pill penalty” as it establishes price setting after nine years for small-molecule drugs compared with 13 years for biologics. It basically eliminates incentives for pursuing new breakthroughs and uses for older medicines. The result could be greater investment in biologics and less investment in small-molecule medicines.
Both biologics and small molecules are equally valuable. Small molecules can be administered orally, making them more convenient for many patients, and they also are critical to treating many diseases.
Access to capital for biotech startups
The biotech startup environment is a rich source of innovation that complements large pharma R&D efforts. The synergy between the two spurs drug discovery.
However, startups struggle in a high-interest rate environment as revenue from product sales are often years away. Higher rates also diminish large pharma M&A intentions as costs rise.
In 2021, 111 biotechs went ahead with IPOs in the U.S. In 2023, only 20 had IPOs. At the same time, there were increased pressures toward biotechs merging or going out of business. According to EY, half of biotechs do not have the cash needed to sustain operations for more than 18 months. Creating an attractive environment for biotech is key to maintaining power in the R&D innovation machine.
Building trust with new models for clinical trial design
Patient trust benefits from decentralized clinical trial strategies that allow those in diverse regions of the world to participate. This, coupled with designs that take into consideration the representation of the patient population most likely to benefit, especially underserved patients, and garnering insights from these patients, can create greater patient acceptance of novel therapies.
Decisions and actions on each of the above will need to be taken carefully, navigating trade-offs to ensure we fully drive the strongest impact from new innovations, insights, and tools. By increasing collaboration with diverse stakeholders to identify roadblocks and formulate solutions in these uncharted territories, we can drive faster discoveries.