Processing Your Payment

Please do not leave this page until complete. This can take a few moments.

April 5, 2023

AI Advantage: BioXcel uses artificial intelligence to speed up drug discovery

PHOTO | Nasdaq BioXcel executives ring the Nasdaq closing bell in New York on Feb. 21.

Developing new drugs is rarely as simple as spending an amount of money and coming out with an effective and marketable product on a set schedule.

Artificial intelligence can help — sifting through millions of molecular combinations and existing drugs to find the few that can be repurposed or redesigned to cure.

Now, innovations in artificial intelligence like OpenAI’s ChatGPT chatbot are highlighting AI’s potential — and New Haven biotech BioXcel Therapeutics is ready to capitalize on those breakthroughs.

Frank Yocca

“There are a number of things out there that are showing really good promise,” said Frank Yocca, chief scientific officer at BioXcel. “It shows the power of using AI when you can use it in the right way, when it can really benefit.”

ChatGPT, launched publicly in November, harnesses the power of an AI platform that can process huge quantities of data to produce human-like responses to queries. Some of the chatbot’s responses have proven to be inaccurate or inappropriate, but the AI behind the technology is constantly learning and improving — and bioscientists are taking note.

“We look at all of the different and various new things that are applied in AI that could make our system work more effectively,” Yocca said.

The spotlight on the evolution of AI comes at a busy time for BioXcel, Yocca said. The company commercially launched its first drug late last year — Igalmi, an under-the-tongue treatment for the agitation associated with schizophrenia or bipolar disorder in adults.

BioXcel’s AI technology identified the compound that became Igalmi, a sedative previously approved for use in anesthesia called dexmedetomidine.

A second AI-discovered candidate, BXCL701, performed well in a phase two trial against a form of prostate cancer with no currently approved treatments, the company announced in January. In combination with immunotherapy drug pembrolizumab, BXCL701 helped patients with small cell neuroendocrine metastatic castration-resistant prostate cancer, an aggressive form of the disease.

BioXcel’s candidate has the potential “to redefine the standard of care for patients with advanced disease,” Dr. Rahul Aggarwal, associate director for clinical sciences at the University of California San Francisco Helen Diller Family Comprehensive Cancer Center, said in February.

“This drug now will enter a bigger phase 2B study, and hopefully head for approval,” Yocca said. “So, we're very excited about that as well.”

As part of efforts to develop cancer drugs like BXCL701, BioXcel formed a new subsidiary last year called OnkosXcel.

Growth targets

The company’s recent developments led BioXcel executives to New York on Feb. 21, where they rang the Nasdaq closing bell.

A public company traded on the Nasdaq, BioXcel plans to ramp up sales of Igalmi this year; it expanded its salesforce to 70 as of December, the company said in March when announcing its financials.

BioXcel missed its fourth-quarter revenue targets for 2022 but had $194 million in cash on hand as of Dec. 31, according to a statement. The company said it was seeking strategic financing from Oaktree and Qatar Investment Authority.

Analyst Francisco Javier Garcia, of Seeking Alpha, said BioXcel should see stronger sales of Igalmi as the drug rolls out beyond hospital settings. Smaller biotechs have been impacted by market instability due to recent bank failures impacting revenue targets, he added.

“Success in the commercial launch of a drug is determined by good revenue growth in the first few quarters, and I really think we'll see very good growth in the next few quarters,” Garcia said in March.

At the heart of BioXcel’s development in recent years is its AI platform, which sifts through terabytes of scientific data in journal articles and other sources to create knowledge graphs, which link facts and concepts and puts them in context. The networked findings are then analyzed by data scientists and neuroscientists to identify new drug compounds, or existing drugs that can be used in new ways.

The innovations on view in ChatGPT, by contrast, center around its use of “large language model” AI, a type of machine-learning model that can generate high-quality text using predictive data. In drug discovery, similar models have been proposed to generate new molecules or protein sequences.

Yocca said he sees the utility of advances like large language models in drug discovery across the value chain.

“That would be from the start of when a drug idea was formulated, when a drug was first made, all the way through its preclinical development into its phase one, phase two, phase three, and then finally, into commercialization and product development. And in each case, we're applying AI,” he said.

AI in its new forms could speed up the process of drug discovery as well, Yocca said.

“So, it’s not one part of the value chain that we focus on, it’s all of them. And if you can tie those together even more readily than we have, say, for example, with this language processing, it’ll just make this go even faster,” Yocca said.

Hiring efforts

One example of an innovative use for AI is in the process of choosing locations for a drug candidate’s clinical trials, which requires an exacting mix of demographics and expertise.

“You want to make sure you choose the best,” Yocca said. “You’ve got all these variables, and this is where the AI comes in and makes a selection that’s based on the data. Then we can overlay our personal views on top of that and make final decisions.”

The sites picked by the combination of AI data and BioXcel scientists resulted in a clinical trial that was able to enroll 750 patients within three months, an impressive feat in the bioscience world.

That mix of both machine and human intelligence is vital for successful drug discovery, Yocca said. BioXcel has been hiring both data scientists and neuroscientists as it develops its drug candidates and expects to add more jobs in New Haven in the near future as the company grows.

But Yocca stressed that BioXcel’s mission extends far beyond the latest technological advances.

“The AI is just another tool for us to find a medication that will help a patient segment, and we can’t lose focus on that. What makes the company go is finding a molecule that has promise in a patient,” Yocca said. “That always needs to be front-and-center. Can we find them faster because of these new techniques? That’s the hope.”

Sign up for Enews


Order a PDF