[AI IN ACTION] Pharmas seek efficient AI-driven drug development

Pharmaceutical companies are turning to AI technology to speed up and de-risk the drug development process.

[SHUTTERSTOCK]

Developing a new drug is a long, arduous journey that spans over a decade and requires hundreds of millions of dollars, not to mention that about nine out of 10 drug candidates fail to get approval.

Pharmaceutical companies, therefore, have been seeking a shortcut to speed up and de-risk the process. This is where AI technology comes in.

With AI swiftly revolutionizing every facet of new drug development, from discovery to clinical trials, pharma companies have been scrambling to establish their own AI-powered drug development tools or forge ties with AI developers.

Daewoong Pharmaceutical established a team dedicated to AI drug development in 2021, the first among the country’s traditional pharmaceutical companies.

According to Shin Seung-woo, head of Daewoong’s AI drug discovery team, the company managed to condense a previously three-year-long process into just one month, using generative AI.

“[The team] used generative AI to discover a substance with better effects than a rival company’s substance in only a month, and verified it in vitro,” a process that would have previously taken three years, Shin told the Korea JoongAng Daily.

Shin Seung-woo, head of Daewoong Pharmaceutical’s AI drug discovery team [DAEWOONG PHARAMCEUTICAL]

Daewoong has developed its own virtual screening system using generative AI and cheminformatics and also secured a customized virtual library, all of which are expected to significantly reduce the time and cost spent on new drug discovery.

“The short-term goal is to discover a lead compound within a year of the project launch,” said Shin.

The CPHI Annual Report 2023, published in September, predicts that over half of new drugs approved by the U.S. Food and Drug Administration (FDA) by 2030 will involve AI during development or manufacturing.

In Korea, the number of pharmaceutical companies that have set up an AI drug development team or begun to cooperate with an AI company for research projects reached 40 in 2023, a significant jump from five in 2019, according to a recent report by the Korea Pharmaceutical and Bio-Pharma Manufacturers Association (Kpbma).

Out of 104 pipelines involving AI in Korea, 71 cases utilized the technology in drug candidate discovery, 26 in preclinical, and 7 in clinical trials.

HK inno.N has also developed and utilized a proprietary AI platform, inno-Sun, which predicts the activity, properties, and toxicity of small molecule structures.

“It is used for a variety of purposes, from providing useful databases for problems that occurred during the drug development to designing drug structures,” explained Kim Hye-Jeong, head of HK inno.N's Innovative Drug Discovery Center.

Meanwhile, CHA Vaccine Institute, which specializes in vaccines and cell and gene therapies, is working with AI companies to develop new cancer vaccines and immunotherapies.

“It has been about a year since we started utilizing AI technology,” said Yum Jung-sun, CEO of CHA Vaccine Institute.

CHA Vaccine Institute has signed a set of memorandums of understanding since last year with a research institute, a medical facility and a biotech startup for AI drug development.

The latest agreement with Pharos iBio, signed in August, is aimed at jointly developing AI-based immunotherapy against cancer.

Yum Jung-sun, left, CEO of CHA Vaccine Institute, poses for a photo during a MOU signing ceremony with Pharos iBio, in August. [CHA VACCINE INSTITUTE]

“We are developing immuno-oncology drugs using our proprietary immunization platform, L-pampo and Lipo-pam, said the CEO, adding that “by using AI, we expect to gain a deeper understanding of the mode of action of these cancer immunotherapies.”

Immunotherapy is considered the primary domain where AI technology is expected to lead significant advances.

“Immunotherapy has many advantages over conventional cancer treatments, and therefore it is important to find more suitable drug candidates,” said Yum. “As there are many unexplored mechanisms in immuno-oncology, using AI to understand them can lead to new candidates.”

Yet the fast-evolving AI technology is posing as many questions as answers surrounding risks of possible hallucination, lack of regulatory guidelines and more.

“There needs to be institutional guidelines on who is responsible for problems caused by generative AI's own flaws, errors in training data, and user misuse,” Yum pointed out

Shin of Daewoong suggested that there is a generational and cultural gap among researchers in the use of AI, saying that “some may feel pressured to change their previous paradigm.”

Shin continued, “we need to improve the understanding of AI-based drug discovery among conservative researchers, and the resulting synergy will enhance the efficiency during the process.”

BY SHIN HA-NEE [shin.hanee@joongang.co.kr]