Nearly every industry on the planet has embraced artificial intelligence and machine learning technologies — it’s high time the clinical R&D community followed suit.
On our blog, Medical Marketing Insights, we write a great deal about the impact of big data, artificial intelligence (AI), and machine learning on the medical industry. From Google’s various high-tech ventures, to AI-driven chatbots, to telehealth, it’s clear that healthcare is on the brink of a smart tech transformation.
And while I’ve written a number of articles about the impact of new technologies on clinical trials (e.g., mHealth, virtual reality, data analytics, remote patient monitoring, etc.), I have yet to explore AI and machine learning’s impact on the drug development and approval process.
A few weeks ago, Jeanne-Francoise Williamson and Pablo Lubroth published an article on Applied Clinical Trials that caught my attention. The piece, “Can We Predict Drug Efficacy with Artificial Intelligence?” explored several aspects of clinical research that could be transformed by AI and machine learning tech. Here are a few of the key takeaways.
An Innovative Approach to Drug Discovery
The bulk of the article focuses on BenevolentAI, a UK-based company using AI to increase efficiency in both diagnostics and the drug discovery process. Recently, its system was used to identify biomarkers in Amyotrophic Lateral Sclerosis (ALS) in an innovative way.
They started by analyzing “billions of sentences and paragraphs from millions of scientific research papers and abstracts.” The AI then identified direct relationships among various datasets and organized it into “known facts.” From these known facts, they were able to develop a set of hypotheses based on qualified criteria. After assessing the validity of these hypotheses, BenevolentAI’s team settled on “20 triaged biomarkers they thought were worth exploring further.”
They narrowed this list even further to the five most promising compounds, which were then tested on ALS patient cells. With clinical trials planned for later this year, the idea is that for the drugs already tested by BenevolentAI’s technology, the time to market will be significantly faster.
As Williamson and Lubroth point out, the idea of making connections between data from separate scientific papers and abstracts is a novel one. According to Ken Mulvany, Chairman of BenevolentAI, “The data [from scientific articles] might show that a protein up regulates a particular gene which is not directly related, leading researchers to look for drugs in a completely different area.” This is a great example of researchers combing through vast troves of data to identify new targets in a way that wouldn’t have been possible without the help of AI.
AI Increases Clinical Trial Efficiency
It’s not just the drug discovery process that stands to benefit from these technologies — clinical recruitment and approval timelines will likely see a positive impact as well. In their article, Williamson and Lubroth suggest that AI may eventually “revolutionize the way pharmaceutical companies perform screening.”
Researchers are already analyzing thousands of patients’ molecules to identify more promising trial candidates. This, combined with more effective treatments (based upon the approach outlined above) and AI-driven trial design and data analysis, will make trial outcomes more predictable and speed up the overall time to market. In turn, we’ll likely see a significant reduction in drug development costs — especially considering, as Williamson and Lubroth point out, a 10% improvement in the ability to predict a drug’s efficacy pre-trial could save upwards of “$100M in development costs per drug.”
While these two examples represent just a small sliver of AI’s current and future applications, the implications are clear: such technologies have the potential to not only identify previously unidentifiable treatments for serious chronic and/or genetic conditions, but also to deliver those treatments in a more cost- and time-effective manner.
However, it’s early days for these technologies and they’re not yet being implemented on a wide scale. In fact, Williamson and Lubron suggest that there are only about 173 companies in the bioinformatics space, worldwide. If we want to see a real tech disruption in clinical R&D, these technologies and approaches will need to embraced by larger biotechnology and pharmaceutical organizations. But when you consider the potential benefits for all parties involved it really begs the question, what exactly are we waiting for?