Artificial Intelligence, BioPharma

AI offers promise but faces barriers in drug development

Inertia is a barrier as is the traditional split between the clinical and the data-driven spheres of drug development. While smaller firms have an edge in bridging the gap, big pharma will eventually get there, said panelists at the INVEST conference session.

Artificial intelligence and machine learning promise to lower the cost of drug development in the arena of precision medicine. But it will take time for the technologies to reach their full potential, especially inside big pharma companies, according to panelists speaking during a session of the MedCity INVEST Precision Medicine conference.

Inertia is one likely factor. Pharma companies already derive significant revenue from drugs developed without the use of AI, reducing the incentive to bet on new approaches, said panelist Panna Sharma, president and CEO of Dallas-based Lantern Pharma.

presented by

“I think data-driven, machine-learning approaches are going to be part and parcel of future pharma. But how they get there is unclear,” Sharma said in a conference session Thursday on AI and drug development. The session was moderated by Dr. David A. Shaywitz, founder of Astounding HealthTech Advisory Services.

A key step, though, involves blending clinical and technical expertise, Sharma added. “I think some aspects of pharma are doing digital really well. But the challenge really is for them to integrate it into their product development and early research work. And that is not being done. It’s still very siloed.”

Lantern Pharma has had little choice but to adopt AI and machine learning approaches in drug development, Sharma said. The company crunches data on existing therapies that may have failed in clinical trials to see if they may succeed in new indications. Its clinical portfolio includes drug candidates for forms of prostate and lung cancers.

“We’re purpose-built around specific programs and specific compounds,” Sharma said. “I can’t go spend $5 million on pre-clinical work and so I’ve got to get it out of machine learning.”

It has plenty of company. More than 200 startups are leaning on AI to identify and develop new drugs, according to one count.

Smaller startups have some advantages in balancing technology and clinical science, said panelist Lina Nilsson, vice president of product for Recursion Pharmaceuticals, based in Salt Lake City, Utah. But she expected big pharma to get there eventually. “Over time, they’re going to learn, too,” she said.

At Recursion – which uses AI to develop drugs for rare, genetic diseases – biologists, engineers and data scientists work together in sub-teams to achieve goals that aren’t easily divided between the science and the technology, Nilsson said. Professionals in different disciplines also work side by side, at least when they can work in an office.

“It’s a hard problem everywhere and something you can never stop focusing on,” she said.

The need for adopting new technology is driven, in part, by the persistent demand for lower drug costs. Precision, personalized therapies are often highly effective but they are expensive.

“The only way to change it is isn’t just to reduce drug pricing … but really you have to go back and change the risk and change the cost of development,” Sharma said, noting that AI and data-driven approaches have reshaped other industries. “They’ve crushed product-cycle development timelines and changed costs in everything from transportation to finance.”

Featured image: Yuuji, Getty Images