AI in Drug Development: A Glimpse Into the Future of Drug Discovery

The emerging deep learning AI techniques propose compelling advantages for drug development industry. Healthcare and Life Sciences practice experts explain how AI-based method helps companies go from research and computer design to a working molecular lead in record time.

The discovery of new drugs is an undeniably important undertaking and represents a massive global market. Statista indicates that the drug discovery market worldwide finds itself on an exponential trajectory, with the expected market value poised to reach 71 billion U.S. dollars by 2025. As of 2016, the market was valued at just 35.2 billion U.S. dollars. Of course, this comes as no surprise; the U.S. pharmaceutical industry, after all, was coined ‘Big Pharma’ for a reason. By 2021, Big Pharma profits for prescription drugs are expected to reach $610 billion and, in 2015, Americans spent $457 billion on prescription drugs. However, as any scientist or expert in the realm of drug development is acutely aware, the discovery of new medicines is an increasingly costly and time-consuming process.

DiMasi, Grabowski, and Hansen performed a new analysis, revealing that the cost to develop and win approval for a new drug has risen to $2.6 billion. Moreover, when we consider costs for post-approval activities, including new indications and long-term safety monitoring, the associated costs reach $2.8 billion for each new drug. From basic research to clinical trials, FDA review, and post-approval monitoring, it takes at ten years on average for a new medicine to complete the journey from initial discovery to widespread use. What if there was a better way?

Well, in fact, there is. Over the last decade, artificial intelligence (AI) has begun to permeate across industries with the promise of enhanced automation and data-fuelled transformation. When applied to drug discovery, AI proposes a compelling advantage: more affordable drug development and more intelligent use of data to streamline timelines associated with bringing a new drug to market, explains Anton Dolgikh, Head of AI, Healthcare and Life Sciences at DataArt. This is monumental; in the world of drug development, even the difference of one year can make an incredible difference for both the company behind its creation and the individuals who will benefit from its intended use.

The Emerging Role of AI in Drug Discovery

Traditionally, new medicine development has been rife with error, specifically, a high occurrence of failure across clinical trials. The overall probability of clinical success is estimated to be less than 12%, and less than 10% of drug candidates make it to market following Phase I trials. Oftentimes, Phase 1 trial failure can be attributed to unexpected toxicity of the leads.

Fortunately, AI implementation helps to predict toxicity at the very beginning of the process, filter out any molecules that are potentially toxic, and validate new drugs more efficiently.

Using deep learning to study compound libraries and predict molecular behavior/interaction between compounds, AI can identify patterns and insights in an increasingly accelerated time frame. Understandably, this process streamlines the initial phases of drug discovery and proactively identifies adverse reactions. New research from the Chalmers University of Technology, Sweden, has also demonstrated that artificial intelligence (AI) can generate novel, functionally active proteins. Researchers note that this discovery promises faster and more cost-efficient development of protein-based drugs.

“In simple terms, with an AI-based method, companies can finally go from research and computer design to a working molecular lead in record time,” says Dolgikh. Beyond initial discovery, AI can be used for cell target classification or diagnosis, drug target identification and validation, and more targeted drug design and discovery. Moreover, leveraging AI during preclinical development could help trials run smoothly and enable researchers to more quickly and successfully predict how a drug might interact with the animal model. However, it is important to note that similar to in vitro experiments, animal models are not always enough. Clinically successful leads can demonstrate toxicity in humans, so it is not yet possible to remove traditional methods out of the loop. To this effect, AI offers researchers the tools required to personalize the clinical trial experience and improve participant monitoring.

This technology can also be applied to analyze existing drugs, including effects on the body and side effects, to inform potential drug repurposing opportunities. Through data modeling, companies can run existing drugs through ‘AI drug repurposing platforms’ to determine new medical applications. Notably, existing drugs have already been passed through regulatory approval processes, and, therefore, the approval of repurposed drugs is streamlined. This is especially applicable to the global healthcare industry’s current challenges, as pharmaceutical companies are working to bring COVID-19 vaccines to market in record time. With the help of AI, companies can readily test and identify existing drugs to determine if they are a candidate to fight COVID-19 and other viral diseases. In fact, AI has the potential to provide over US$70 billion in savings for the drug discovery process by 2028.

Warning: Data Overload

Can too much data be a bad thing?

In the realm of drug development - absolutely. To date, drug companies, scientists, and technologists have collected an incomprehensible amount of data through their research and testing. This information is pivotal in the continued development and advancement of the medical and pharmaceutical industry, but, like any data, it is subject to human error; specifically, data silos and misinformed inference. In this case, the problem lies in human limitations and inefficiencies. We simply have too much data and information to effectively sift through and contextualize without some degree of technological intervention. And as data collection continues to grow more complex and effective over time, our ability to decipher it must match that trajectory of innovation. Fortunately, AI unlocks a degree of rapid comprehension that was previously unattainable.

With emerging deep learning AI techniques and modeling, analyzing an otherwise insurmountable amount of data can take days rather than years. AI technology can effectively comb through all publicly available datasets, statistical models, and historical information to create predictive algorithms. McKinsey reports that big data strategies deploying AI capabilities, such as predictive modeling and analysis of sensor data, could generate up to $100 billion in value annually across the U.S. healthcare system by “optimizing innovation, improving the efficiency of research and clinical trials, and building new tools for physicians, consumers, insurers, and regulators to meet the promise of more individualized approaches.” Moreover, findings from a recent Carnegie Mellon study suggest that AI could lower their discovery costs by 70%.

“Empowering a more streamlined, cost-effective, and timely approach to the development of new medicine, AI is positioned to modernize and transform our understanding of health and human disease,” shares Dolgikh. “The improved ability to analyze, understand and apply a wealth of critical data in the drug design and implementation process is made possible by AI.” Without it, industry innovation will remain at the mercy of human error and traditional time constraints. Developing new pharmaceuticals is a marathon, and AI allows companies to tap into advanced levels of efficiency to consistently decrease associated costs and the time to market.