How AI Is Transforming Drug Discovery

In the medical, biotech, and pharma industries, a quiet revolution is under way. Artificial Intelligence (AI) is transforming the drug discovery process, with an increasing number of companies seeking regulatory approval for drugs to treat conditions including glioblastoma, ulcerative colitis, and some cancers.

A recent paper in Nature reported that the integration of AI into the drug discovery and development pipeline increased by almost 40% in 2022. New regulations are being introduced to keep pace with these advancements and several drugs that were discovered via AI are being tested in a clinical environment.

So what does this mean for the future of drug discovery? Here, we discuss the benefits of AI-supported drug discovery, the obstacles to FDA approval, and the companies leading the charge.

The Benefits of AI Drug Discovery

The high costs associated with the discovery and development of new drugs are well documented. On average, it takes between 9.5 to 15 years to develop a new drug and costs up to $2.6 billion. In addition, only a fraction of the drugs put into development actually make it to market.

AI applications can analyze vast swathes of biological and chemical data to quickly identify high-potential drug candidates, predict their effectiveness and safety, and optimize their design. As a result, the discovery period for a new drug candidate can be reduced to an average of just seven months. Not only does this serve to reduce costs and drive efficiencies, but it provides patients with quicker access to groundbreaking treatments.

In addition, these applications can quickly identify secondary uses for drugs that have already been approved by a regulatory body, such as the U.S. Food and Drug Administration (FDA). This expedites the process – and lowers the associated costs – of repurposing a drug to treat alternate conditions.

Securing Regulatory Approval and Other Barriers to Success

AI promises to accelerate the process of drug development, but the journey to securing FDA approval remains a long one.

The four steps in the drug development process are:

• Step 1: Discovery and Development
• Step 2: Preclinical Research
• Step 3: Clinical Research
• Step 4: FDA Drug Review

As per FDA requirements, the Clinical Research step typically includes four phases:

• Phase 1: A group of 20-100 healthy volunteers or people with the disease or condition are tested for the treatment’s safety. Dosage and side effects are closely monitored.
• Phase 2: A larger group (several hundred) of people with the disease or condition are tested over a longer period to evaluate effectiveness.
• Phase 3: Between 300 and 3,000 volunteers with the disease or condition are studied for 1-4 years to measure efficacy and adverse reactions. Around 25-30% of drugs pass this stage and progress to Phase 4.
• Phase 4: Several thousand volunteers with the disease or condition are studied to measure efficacy and adverse reactions.

The good news is that drugs developed with the help of AI are more likely to advance through these Clinical Research phase steps and secure regulatory approval.

Another challenge drug development companies face is a lack of opacity during the discovery and development stage. AI applications often use gray or black box systems, which fail to provide insights as to how a conclusion was reached. To ensure that processes are safe and robust, these companies must invest in the tools and the skills needed to verify the analyses of their AI systems.

As drug development companies put more faith in AI, they must also consider organizational structure. While these applications promise to augment, and not replace, human workers, they will still demand an overhaul of existing systems and processes, something that requires time and effort.

Who Are the Biggest Players in AI Drug Discovery?

Researchers are developing AI models to optimize protein structure, carry out drug binding, and map complex multifactorial diseases.

Several exciting drug discovery projects are under way, and a handful of companies have advanced to the point where they are conducting clinical trials. Among them are Verge Genomics, BenevolentAI, Exscientia, and InSilico Medicine.

1. Verge Genomics

Verge Genomics has designed an all-in-human, AI-powered platform, known as CONVERGE, which it hopes will solve some of the biggest medical challenges of our time.

The first drug discovered by CONVERGE is VRG50635, which has entered the clinic for amyotrophic lateral sclerosis (ALS).The company is also working on treatments for Parkinson’s Disease and Frontotemporal Dementia.

2. BenevolentAI

London-based BenevolentAI has several drug development programs in its pipeline, including treatments for Atopic Dermatitis, Ulcerative Colitis, Glioblastoma Multiforme, Inflammatory Bowel Disease, and ALS.Its platform is designed to optimize the earliest and most critical phase of the drug discovery process; target identification and validation.

The company’s Atopic Dermatitis drug, BEN2293 has entered Phase 2 of clinical trials and its Ulcerative Colitis drug, BEN8744, is expected to enter a Phase 1 clinical trial early this year.

3. Exscientia

In 2020, Exscientia was responsible for developing the first-ever AI-designed drug candidate to enter clinical trials. The drug, EXS21546, is an immuno-oncology treatment designed for use in adult patients with advanced solid tumors.

More recently, in February 2023, the company revealed an immunology & inflammation candidate, EXS4318, would be entering Phase 1 clinical trials in the United States. Bristol Myers Squibb will oversee the clinical and commercial development of the drug.

4. InSilico Medicine

Hong Kong-based InSilico Medicine’s AI software seeks to significantly reduce the time and cost associated with bringing life-saving medications to patients. Its first drug, INS018_55, will treat idiopathic pulmonary fibrosis (IPF), a chronic lung disease, and has entered Phase 1 clinical trials in New Zealand.The company also has several oncology and immunology candidates in its drug development pipeline.