AI R&D

Over the last three years, artificial intelligence (AI) has become a major trend and area of focus in pharmaceutical research and development, with machine learning expected to deliver greater insights and understanding of diseases and patients.

However, AI is not just a technological step-change; it has the potential to radically transform core drug discovery and development models—and power the journey toward intelligent therapy solutions. This perspective was confirmed when I recently talked about “revolutionizing clinical development” during this year’s Clinical Innovation Partnership conference in Zurich, Switzerland. The vast majority of clinical operations leaders there understood the change AI can unleash in the way we design and execute clinical research. It is now about demonstrating that value and powering ahead!

Historically, pharma R&D has grappled with three fundamental questions:

1. How to drive the most compounds through to market launch.
2. How to ensure these compounds have a superior performance compared to current standards of care.
3. How to do so in the most productive way.

How can AI help to answer them? Clearly, AI can help pharma companies win the productivity race by increasing speed and cost efficiency while maintaining high levels of compliance and reducing “human error.” But—more importantly—it can help drive companies toward a more personalized and outcome-driven value proposition.

Industry executives already see AI as essential to a successful R&D operating model both today and in the future. In a recent article I wrote for the Harvard Business Review about “Intelligence-Driven Therapy Solutions,” Badhri Srinivasan, Head of Global Development Operations at Novartis, contributed his thoughts. He mentioned that “…60 percent of all the development cost resides in the design and execution of clinical trials where implementation of AI and machine learning brings a considerable opportunity on productivity and cost to saving….” The Accenture survey, “AI: the momentum mindset” also stated that 51 percent of respondents identified productivity as a benefit of deploying AI within their organization; in addition, they pointed to AI’s value in improving patient and clinical outcomes.

To seize the truly transformative power of AI, R&D leaders need to see the bigger picture. Taking a view that’s too narrow and focuses on simply optimizing R&D processes might result in missed opportunities for competitive advantage.

Across the industry, we see three stages of AI awareness and adoption in pharma R&D organizations:

1. 1. Accelerate R&D productivity. Most companies are currently at this stage. They recognize the value of AI for driving efficiency and productivity, and they mainly apply intelligent automation to their processes and infuse new technologies into existing R&D processes. Pharma companies that choose not to advance beyond this stage risk:
      • Being stuck in the role of product supplier in a value chain that is rapidly moving toward integrated treatment innovations
      • Leaving room for other players to lead the transformation and capture the highest share of value

1. 1. From compound to therapy. A few companies are starting to move toward this stage. They understand that AI can enable them to improve outcomes. If they strategically leverage data, they can discover and develop breakthrough therapies targeted to patient sub-populations or leverage metadata to optimize clinical protocol design. In this stage, the overall purpose of R&D remains substantially unchanged, but pharma companies will begin collaborating within their ecosystem and leveraging AI to evolve from “simple” compound developers to developers of therapies targeting patient sub-populations.

1. From therapy to living services. In the future, pharma companies will leverage big data and AI to empower human study and decision-making, enabling the identification of patient-specific needs and treatments. Moreover, together with markers and devices, AI will allow the evaluation of disease evolution and patient response to therapies to adapt accordingly. AI data analysis will also enable companies to identify patterns and common patient behaviors to develop disease prevention recommendations.

To prepare for this transformation, we recommend companies focus on three main pillars:

1. 1. Data. Companies should build strong data integration and analytics capabilities to combine structured raw data with unstructured, personalized and interpreted data from different sources, including internal and external researchers, clinical trials, genomics data, devices, and real-world evidence. At the same time, they must consider data ownership, privacy and ethics. The most successful transformation journeys, however, build a data focus in a very targeted, strategic manner. I have often seen holistic data integration endeavors fail miserably. Ideally, you need to have a good understanding of what data you require, which use cases you want to enable and how this data will be gathered.

1. 1. Operating model. Companies should evolve their innovation operating model, breaking functional silos across discovery and development to allow translational teams at the intersection of biomedical, clinical and data science to advance assets across preclinical and clinical stages. Moreover, they should build innovation platforms to enable communication and collaboration within translational teams and others in the larger ecosystem. From what we see, a great way of building these new capabilities is by applying a venture capitalist perspective: stage-gating investments and tying funding closely to outcome, progress and milestone achievements.

1. Skills. For companies wanting to obtain data and AI skills, it’s not as simple as adding data scientists. Everyone in the organization, from researchers to managers, will have to improve technology skills and adopt a data-driven mindset. Especially in drug development, where statistical modeling has been part of the capability set for decades, I’ve seen successful clients actively drive a change agenda that clearly distinguishes between biostatistics and the design of machine learning algorithms.

The transformation from accelerated productivity, to compound, to therapy, to living services is one that will require years to reach maturity, which is why I recommend organizations get a head start now. As one of my clients recently said, “You want to have your surf board in the water when the wave is coming.” I deeply believe that R&D departments of innovative pharma companies are uniquely positioned pioneer this broader enterprise transformation. The time to act is now.