Holding On Tight to AI Wave

Throughout 2024, artificial intelligence (AI) continued to be a key talking point in the life sciences industry. As is the case with any new technology, the conversation centered around two questions. First, the industry asked how the technology can be incorporated into its processes. The second question is whether or not AI will be around for the long haul.

Regarding the former, Pharmaceutical Executive spoke with Panos Karelis, director of customer experience and insights at Intelligencia AI, about the key ways that AI and machine learning (ML) technology are being applied to build a variety of digital platforms across the industry. Over the past year, many pharma companies either announced the development of a new digital platform or a partnership with a company that develops these platforms.

Panos Karelis

In the coming year, Karelis sees both pros and cons on whether pharma companies will focus on developing their own platforms or seek out existing platforms to partner with.

“I can argue it both ways,” he says. “Understandably so, there is all the sensitivity around data, both proprietary and internal, non-publicly available data. Pharma companies hold onto their data very tightly. They don’t easily trust it with external sources. There are, certainly, use cases that will be developed internally by Big Pharma. These would likely focus on drug discovery, target identification, and other applications in the preclinical space, where information about molecular structures, target interactions, and early-stage research could expose competitive advantages or intellectual property.”

He continues, “There is also the other aspect of the complexity of building such AI solutions. The complexity is not only around the algorithms and the machine learning aspects, but how companies build their data cubes, and how they collect, structure, and harmonize that data so it becomes AI-ready and can feed the algorithms. This must be done to ensure that they generate meaningful and trustworthy output.”

Another area where Karelis sees uncertainty is the cost and complexity of building these platforms. At the moment, companies may not realize just how expensive the task is, not just to develop a platform, but also to run and maintain it. There are also additional costs related to hiring experts, whereas external vendors will already have these factors built in to their models.

One thing that he is certain about, however, is that AI and these platforms will not just continue to be a presence in the pharma industry, but they will become more important and more integrated in the coming year.

“What we see right now in pharma spans widely across the value chain, from molecular design, targeted identification, clinical trial design, drug efficacy monitoring, and workflow optimization,” explains Karelis. “There are many companies and applications in the works generating a lot of interest and hype. Some caution is needed, however."

Another trend he sees ahead that will impact pharma is the rise of generative AI (GenAI) applications and large language models (LLMs). "The fascinating aspect there is when AI interactions mimic how humans converse with each other,” Karelis tells Pharm Exec. “There’s a lot of excitement there because of the ways [LLMs] may be utilized to improve workflow and services. We've heard that from our customers. They already applied GenAI for smarter services within their internal documents and it even helps them synthesize materials and information.”

Karelis describes the integration of AI and ML as a “new era” in healthcare and pharma. The technology is being used in decision-making, various processes, and workflows in general, and while he sees AI as an opportunity at the moment, he believes that it will become a necessity in the future.

“AI/ML is a somewhat nascent technology, especially when it comes to real applications and impact in pharma, and it needs maturing, which will happen over time. It’s also not one-size-fits-all, so companies need to take the time to understand the technology and find the right use cases, which is critical and can take some time," says Karelis. "I believe that over the next years, partnering with external data and AI solution providers will be a necessity. It's not replacing humans. It's augmenting human processes, thinking, and decision-making, and it improves, optimizes, and maximizes workflows and efficiency.”

Karelis reminds of the fact that AI is still a relatively new technology. It’s neither good nor bad—it’s just new and uncharted territory. The impact of AI centers around how people react to and use it.

“There seems to be also some fear around the unknown,” he adds. “I think that the greatest risk that AI-driven technology faces right now is that people may lose faith in it. And this may happen because they try immature solutions and lack the patience to wait for the technology to mature enough and demonstrate its full potential and benefit. Any company that is working with data and AI must be transparent with their approaches to alleviate the fear of an AI black box.”