Advanced data sciences such as artificial intelligence (AI) and machine learning (ML) are increasingly appealing to pharmaceutical professionals. Use of such technologies offer the chance to accelerate the time to take a drug from idea to market, improve return on investment and other benefits.
Aurelio Arias (AA), senior consultant for thought leadership at IQVIA, will be talking AI and machine learning (ML) during a panel entitled How to Meet the Increasing Global Demand for Data Science, AI and ML and its Potential, scheduled October 8, 3:30 to 5 pm CET. He recently gave a preview of the topic in a discussion with Outsourcing-Pharma (OSP), including AI opportunities and challenges.
OSP: Could you please provide an overview on how use of AI and ML has evolved in pharma R&D?
AA: Over the past few years there has been a large leap in the way AI can assist the hunt for new drugs. AI can now search catalogues of known compounds to assist in retrosynthesis, give a novelty score on a drug and even identify new indications for old medicines.
All these domains have one thing in common: they rely on large amounts of historical data to inform AI algorithms; truly de novo R&D lacks this data trove and so AI is relegated to support functions in searching through literature and reducing the candidate search space. However, AI is making important steps in computational biology, where greater pattern recognition can be leveraged for enhanced protein folding models.
OSP: What are some of the primary challenges in preventing the pharma industry from further tapping into the potential of such technologies?
AA: The technology is simply not mature enough to handle the complexities of drug discovery in a one-size-fits-all model. AI favors the lucky few who have access to structured data, talent, funding and leadership to implement successful use of AI in a heavily regulated industry.
For many, the largest barrier to entry remains the huge complex challenges around implementing flexible regulations with uncertainty around ethics, custodianship, privacy, security, and data sharing to name a few.
OSP: Similarly, how have AI/ML tech providers and pharma firms worked together to help increase understanding and adoption?
AA: Successful AI providers have had to show promising use cases as a way to convince what is a careful and slow-moving industry. These have been focused and began with increasing efficiencies across narrow business processes, such as distribution, before expanding into the healthcare domain such as sorting through electronic medical records and clinical trial literature.
More lately, we have begun to see a shift from cost reduction to value addition, in effect bringing direct benefits to patients. This is an exciting step for AI and digital technologies as a whole as it reframes the way industry sees AI’s value proposition.
OSP: Could you please provide an overview of some of the benefits increased use of AI, ML and advanced analytics has helped pharma firms reap?
AA: The largest benefits to date come from cost and time reduction across the value chain. From optimizing chemical synthesis to finding the shortest distribution routes, AI is already helping companies cut costs. Harder to measure, but increasingly important, is AI’s ability to predict disease progression, target physicians and engage with patients. AI is breaking into softer benefits that have the potential of bridging the gap between pharma and patients more than ever.
OSP: Any advice for pharma firms and other stakeholders on how to increase their use and understanding of AI and ML in their organizations?
AA: Adoption of AI must be aligned to the wider company vision and strategy. Successful implementations have come from identifying processes that can be solved by existing AI solutions, such as sifting through trial data related to the company’s therapy area focus. Novel implementations must be approached with care and access to experts is key to temper expectations and avoid common pitfalls.
The best practice for companies moving forwards is to modernise their data feeds first, structuring and enriching where possible will make clear the possibilities and limitations. A solid data platform is required for a successful AI launch.
The CPhI Festival of Pharma takes place digitally October 5 to 16. For more information about the program or to register, go to https://www.cphi.com/festival-of-pharma/en/home.html.