Artificial intelligence (AI) is the development within computer systems of the ability to perform tasks that have traditionally required human intelligence. Examples of AI applications include visual perception, speech recognition, and decision-making.
Medicine is a natural fit for many artificial intelligence technologies under development.
It’s easy to see how AI may become useful in medicine. An AI algorithm doesn’t experience fatigue after analyzing dozens or even thousands of radiology images. Decision-making algorithms can be based upon quantities and types of data that would be impossible for even the most talented physician to take in and analyze.
AI is also expected to have a role in drug development and approval, since it can analyze data quickly and discern patterns that may be invisible to human researchers. And the FDA is taking notice, developing use cases for AI and modifying or creating regulatory frameworks that should allow AI a more prominent role in the pursuit of actual therapies following new drug patents, as well as the development of other innovative new therapies.
FDA Developing New Regulatory Framework for Developing AI Innovations
There is really no precedent for the use of AI in clinical trial design and clinical decision support, so the FDA is actively developing new regulatory “guardrails” to promote innovation while protecting patient safety. Perhaps what the FDA (and drug developers) are most excited about with AI in drug development is the fact that AI continues to “learn” and improve with use. Ultimately, this could result in allowing drug developers to make certain small changes to products without having to begin the submission process over every time. It’s critical, however, that AI technologies deliver definable, measurable benefits to patients so that they can be used with confidence.
A New Role for Real-World Evidence in Clinical Trials
Artificial intelligence should also create a place for so-called real-world evidence (RWE) in clinical innovation. This type of evidence is traditionally gathered through observational studies, and the data is typically complicated and unstructured. AI technology, however, has the power to make this unstructured data useful.
Data for clinical trials may someday include patient-generated data, perhaps collected from devices like smartwatches.
And while it may sound as if RWE might eventually homogenize treatment, in fact the opposite should be true. Collecting and analyzing RWE using AI should help shape healthcare decisions so that care is more customized, which should improve outcomes while extracting more value from healthcare spending.
Machine Learning and Deep Learning to Uncover Patterns
Machine learning is a type of AI that relies on neural networks – computer systems modeled on the human brain. It has applications in bewilderingly complex statistical techniques like multilevel probabilistic analysis, allowing computers to simulate how the human mind works, but without human constraints like “having a bad day” or forgetting something.
Deep learning is a software application that recognizes patterns that might otherwise go unnoticed by humans. During the process of pursuing a drug patent, for example, deep learning could allow computers to separate various aspects of, for example, microscopic images. Image aspects like size, shape, and color, could be individually analyzed thoroughly before integrating all the outcomes and drawing highly-informed conclusions.
The Goal: Streamlined Trials and Approvals
The era of AI in drug development is only beginning, but already researchers and FDA officials alike recognize the huge potential of AI to streamline drug trials and the approval process. When drug patents lead to actual products more efficiently, drug developers benefit, and even generic drug makers stand to gain over the long term. Consumers stand to gain too, because a more efficient drug patent approval process should theoretically help constrain drug prices. The use cases for AI in drug development are only just starting to be explored, but they hold tremendous promise for researchers, drug manufacturers, and patients alike.
A Product Manager with expertise in pharma marketing and sales operations