No matter where we turn, we’re never far from the reality of how artificial intelligence (AI) is impacting our daily lives. Whether it’s asking Alexa to give today’s news headlines, or our thermostats knowing more about our whereabouts and travel plans than we do, it’s clear that AI will fundamentally change our world in ways that we may not yet fully anticipate. Which is why it’s not surprising that AI is one of the most common topics we’ve been discussing with our life sciences clients. They’re looking for insights into how to apply AI in their businesses—when and where they should invest, what they need to do to prepare, and how to tell what is real and what is just marketing hype.
The reality is that there are no simple answers, as AI is a constellation of capabilities with a broad range of maturity. Some have already shown value as proven applications within life sciences and health care. Image analytics are being used by pharma researchers to characterize features of cell lines as they seek to understand disease mechanisms and drug candidate efficacy; text analytics are being used to help with earlier identification of adverse events; and a number of organizations are using innovative techniques to find ways to repurpose existing drugs. While some complex or niche applications are further behind on the maturity curve, the competency of these tools is growing rapidly. Against this landscape, market analysts are projecting investment in AI tools will grow at an annual rate of 35 to 45 percent over the next few years.
This spectrum of capability and rapid pace of change may make it difficult to predict when and where to take the plunge with AI, yet the clear potential is impossible to ignore. For example, from a sheer efficiency perspective, AI could assist scientists with navigating the staggering breadth and complexity of data within life sciences, such as assessing the full range of possible chemical compounds in order to discover the next life-changing drug candidates. AI could also help match patients to clinical trials and then identify those patients who need extra support to remain engaged throughout clinical trial execution.
However, the greatest potential for impacting biopharmaceutical R&D, regardless of whether AI or other tools are applied, could be through translational science. The ability to select the right drug targets to pursue has the potential to dwarf efficiency improvements. As more data is collected and made available—such as clinical, electronic medical records (EMR), and wearable data—AI’s real promise in drug discovery is the potential to make distant and non-obvious connections, to in turn uncover the links between disease and underlying disease mechanisms. Evidence for this is promising, with clinical trials involving genetically validated targets showing to be twice as likely to be successful as those without that validation.
Last month, I joined a FierceBiotech panel discussion at the annual JPM Healthcare conference, titled “Rewiring R&D: The Promises of Digital, AI, and Machine Learning for Biotech Research.” I spoke on the panel with life sciences experts from Astellas, GSK, Bristol-Myers Squibb, Third Rock Ventures, Insitro, Gritstone Oncology and Novartis, and learned along with the audience about the potential that AI could bring to our industry. Here are some key points from the discussion that stood out for me:
We’re not always talking about the same thing. AI is a term that is interpreted differently depending on your perspective, and the terms AI, machine learning (ML), and deep learning (DL) are often used interchangeably—and often as marketing hype. The broad and inconsistent use of the term presents a communication challenge for the people seeking to truly understand and take advantage of AI capabilities.
You need early wins to gain momentum. With any change, generating early success is key to driving adoption. Picking “good” AI use cases—ones that are well-defined, with measurable outcomes and the potential to demonstrate value in a relatively short period of time—is critical. With AI, the added complication is the learning curve associated with training and validating the AI solution, so timelines also need to be carefully managed to ensure that deliverables meet or exceed expectations, rather than disappoint.
AI will require access to uniquely skilled talent. To seize the full potential of AI, companies will need to find talent that not only has the therapeutic and functional expertise required to drive sound research, but also the knowledge of data science and the ability to quickly adapt to changing paradigms. Such multi-lingual talent will be in short supply until university curricula and corporate training programs begin to catch up with the upskilling that is required. Even with the promise of automated machine learning platforms—those that provide user-intuitive functionality, reducing the need for technical expertise—will be best executed by those who can fully grasp their potential.
We need to overcome the risks of data sharing to reap the rewards. During a lively part of our discussion, the theme of industry-wide data sharing was raised. The effectiveness of AI and machine learning depends on access to large quantities of well-understood data. While some companies struggle to assemble enough data to individually scale their use of AI, use cases exist that show how pooling data from across the industry could accelerate the maturity of AI. While this still represents a line that cannot be crossed for some companies, others, including GSK, are actively participating in AI-focused initiatives such as the ATOM Consortium that are dedicated to finding ways to transform drug discovery. Time will tell whether biopharmaceutical companies will come to see the benefits of this approach on their own, or with encouragement from regulators and payors.
It was an honor to participate in the FierceBiotech panel discussion, and the experience made me even more hopeful for the promise of AI within biopharmaceutical R&D. The stakes are high in the life sciences industry—these companies are looking for the next life-changing discovery—and AI could accelerate their ability to reach success or failure, and ultimately help bring more targeted, effective drugs and therapies to patients faster.