How AI reduces the cost and time of drug discovery and development

According to the Tufts Center for the Study of Drug Development, the average cost of bringing a new drug to market almost doubled between 2003 and 2013 to $2.6 billion. The lab-to-market timeline increased to reach 12 years, and 90% of drugs wash out in one of the phases of human trials.

Pharma companies are spending more on drug discovery and development than ever, yet developing less successful drugs. Ten years ago, one dollar invested in R&D generated a return of 10 cents. Today, it yields less than two cents.

There is no question that drug discovery has become increasingly competitive and expensive, driving pharmaceutical companies to look into new methods for reducing R&D costs. Artificial Intelligence promises to solve key industry challenges and significantly speed up the discovery process to help companies reclaim a sizeable chunk of their costs.

Whether it’s cost savings or faster drug development, AI approaches such as machine learning create tangible value for companies. The McKinsey Global Institute estimated that AI solutions applied in the pharma industry could bring almost $100 billion annually - and that’s across the healthcare system only in the United States.

What problems does AI solve?

• Machine learning algorithms can help predict the success of molecules that are used in clinical trials.
• They also let scientists identify compounds that could be applied to target other diseases.
• Researchers can use ML tools to draw insights rapidly from vast data sets that previously would have taken large teams years to analyze.
• AI offers access to new biology (functional genomics insights), as well as improved or novel chemistry.
• It generates better success rates by increasing the number of drugs that make it through clinical trials and gain regulatory approval, as well as better patient stratification.
• AI can address many challenges and constraints of traditional R&D.
• All in all, AI technologies make discovery processes faster and more cost-efficient.

AI in drug discovery and development - real-life examples

• Bristol-Myers Squibb deployed an ML program trained to find patterns in data that correlate with CYP450 inhibition. The program increased the accuracy of CYP450 predictions to 95%, a 6x reduction in failure rate compared with conventional methods. These results help researchers quickly screen out potentially toxic drugs and focus instead on candidates that are more likely to make it all the way through multiple human trials to FDA approval.
• Merck and Bayer partnered with Cyclica to improve the identification of promising drug candidates. They succeeded in identifying a target protein connected to already FDA-approved drugs for systemic scleroderma and the Ebola virus.
• GlaxoSmithKline partnered with Exscientia and already reported on a promising molecule targeting a novel pathway to treat chronic obstructive pulmonary disease.
• Exscientia also partnered with a German biomedical company Evotec to develop a new cancer treatment - human clinical trials of the A2a receptor antagonist began in 2021. The candidate drug was discovered within eight months of the project launch.
• Berg software's recent trial of a drug called BPM31510 indicated that the drug may be a promising treatment for pancreatic cancer. The trial was conducted in patients with advanced pancreatic cancer, which is extremely aggressive and difficult to treat. Results from previous trials against other types of cancer verified Berg’s ability to predict which patients would respond best to the drug and those more likely to experience adverse reactions.
• BenevolentAI used AI to identify baricitinib - a drug developed by Elli Lily for treating rheumatoid arthritis - as a potential COVID-19 treatment. The medication has been approved for use against the coronavirus in the U.S. and Japan and is being evaluated by the European Medicines Agency.
• In a joint research project, Taisho Pharmaceutical and Insilico Medicine began to identify compounds that may slow the cellular effects of aging. Insilico uses an AI network to identify therapeutic targets and find druglike molecules that target senescent cells thought to be behind a variety of diseases when accumulated.

What do these examples show us about the role of AI in drug discovery?

Drug discovery is a process that involves sorting and cross-referencing millions of compounds and molecular designs. The effort is time-consuming, even with machine learning support.

AI tools can help by sorting and cross-referencing data to deliver targeted results to speed up the discovery process. To unleash the full power of AI analytics, it is crucial to optimize models on vast amounts of the so-called prediction-first datasets, which entails a continuous feedback loop between the results of wet laboratory and in silico experiments. A good example of this approach is NaturalAntibody’s partnership with Icosagen Cell Factory. Computational solutions trained in this way enable precise and more effective verification of biological hypotheses, allowing the data to be translated into real therapeutic strategies.

Many companies use AI to mine historical information to predict clinical design pitfalls and target drugs for specific disease categories. For example, Microsoft and Eagle Genomics are developing an enterprise research platform that processes large amounts of data on how bacteria, fungi, and viruses play a role in disease.

Process workflows, prioritization, and pipeline management also present a data challenge. Pharma companies typically work on many potential new drugs at once, using multiple complex workflows, including sequencing, molecular engineering, validation, and mapping.

AI can help standardize and streamline data integration across disparate processes; it also improves the speed and efficiency of drug development by reducing costs.

Take the first step

NaturalAntibody solutions have supported many organizations in finding suitable candidates faster thanks to an extensive antibody database and AI-driven antibody analytics.

Learn how AI can help from our latest publication published in Briefings in Bioinformatics: Machine-designed biotherapeutics: opportunities, feasibility, and advantages of Deep Learning in antibody discovery

And if you work with antibodies, see how AI can help you streamline the drug discovery process for yourself - get in touch with us and book a demo to see how our platform works.