The artificial intelligence (AI) revolution seems to pick up pace from one week to the next, but what does it mean for the pharmaceutical sector?
Since its inception, the pharmaceutical industry has been at the forefront of human knowledge and innovation, occasionally driving the development of new technologies. Naturally, the sector has been an early adopter of AI and machine learning.
How are these new technologies being deployed? What might it mean for the future of medicine?
Drug Development
Technology has been supporting drug discovery for decades. The application of technology in drug discovery is known as computer-aided drug design (CADD) and is considered a pillar in the pharmaceutical discovery pipeline.
What is new are the latest advances in computer science as well as the application of machine learning and AI to discover novel drugs and medicines. An Artificial Intelligence in the Life Sciences article notes that the advances in AI algorithms in recent years have helped the pharmaceutical industry navigate how to create new chemical structures.
Algorithms can process vast troves of data to determine how a new drug might work, including identifying which molecule might be best suited to a particular process. The advancements are enabling researchers to identify disease-associated targets and predict their interactions with potential drug candidates, according to the Journal of Advanced Research.
This makes the drug discovery process much easier, allowing for a more targeted strategy. That means a greater chance for us to see successful drug approvals. These approved drugs will provide insights into how the process can be used (and improved) in the future.
Clinical Trials
Pharmaceutical companies and medical device manufacturers run clinical trials regularly. However, I’ve seen these organizations miss out on streamlining their drug discovery journey simply because they couldn’t document the process.
This is important, especially since pharmaceutical companies need to present their drug approval cases in front of FDA advisory committees. Companies need to show they can back up their findings.
Several years back, we worked with a large pharmaceutical company to tackle this challenge at a time when AI was not nearly as popular as it is now. We developed a design memory tool that helped the client with capturing and organizing all of the documentation, learning and management events associated with the development of a clinical trial.
Back then, we had to gather the data manually. Now, AI makes it much quicker and faster to connect events, documents, decisions and other data to gain insights into how the process can be improved in the future.
In another project we were involved in, we helped a large IT health and clinical research company reach diverse populations in clinical trials. This was also at a time when AI was not nearly as ubiquitous as it is today. We helped the client expand their geographical reach and get access to diverse demographics and difficult-to-recruit candidates all through virtual clinical trials. This was a game changer for the organization, which continues to use the tool to this day.
Still, I can’t help but wonder how AI would have made the pivot from in-person to virtual clinical trials much more efficient—especially with ample patient data available today.
Drug Repurposing
The pharmaceutical industry is among the most data-rich in the world, yet much of that data can end up effectively discarded after the drug discovery process.
While drugs may not have passed through the FDA process, the active molecules can still be repurposed for other uses. To date, that repurposing has typically depended on human intervention. With machine learning, the process can be much speedier and easier.
To give a particularly exciting example, AI can be used to catalyze something called polypharmacology—the use of multiple pharmaceutical agents to act on different targets and pathways. This can help to map the relationships between genes, diseases and drugs, and, from there, to discover new and novel connections that may lead to major pharmaceutical breakthroughs.
A recent example of repurposing occurred during the pandemic when AI was used to explore the potential for existing drugs and treatments to be harnessed in the fight against Covid-19.
Precision Medicine
As we revolutionize our understanding of genomics, we also expand the potential for something called precision medicine—treatments that are fully personalized to the patient’s genomic profile, thus greatly increasing their chance of success.
For example, take a patient who has been diagnosed with leukemia. By sequencing the patient’s DNA, doctors can know exactly what variant of leukemia they are dealing with and how to treat it.
Once again, the function of AI is to greatly expedite the research process. The speed at which researchers can work on DNA and protein structures has increased exponentially, making these once-moonshot goals a genuine possibility. The theory and practice of this work have been of significant interest to researchers.
Challenges
For all of AI’s exceptional potential, there remain some barriers to its uptake.
Algorithms require very large datasets to work and learn from. In practice, however, much of the data relating to clinical trials and drug research—to take one example—is typically siloed between different researchers and pharmaceutical companies.
The highly specialized nature of medical data also creates hurdles when it comes to data labeling—the process by which researchers provide context with which algorithms can work their magic. The pharmaceutical industry will need to improve and standardize processes around data and develop a shared understanding of what constitutes a useful dataset.
This may prove easier said than done. Given the highly regulated nature of the global pharmaceutical industry, with strict rules around disclosure and transparency, the process of preparing data inputs is likely to be even lengthier and more costly than it would be elsewhere.
Pharmaceutical companies should be proactively engaging with regulators, patient groups and policymakers to support the development of practices that will enable the effective processing of data while maintaining the vital trust of patients and other stakeholders.
Conclusion
As this work begins, we shouldn’t lose sight of what’s at stake. AI will not only help to deliver new and cheaper drugs but also expand the possibilities of what medicines can do. As we navigate these advancements, the future of medicine looks brighter than ever—with AI at the forefront of it all.