Introduction:
The article “How Far Can AI Take Medicine Research?” briefly covered AI intervention with clinical trials. Today we’ll go more in-depth on how AI can revolutionize clinical trials.
Clinical trials arguably play the most important role of medical intervention. They are a cornerstone in evidence gathering. They are used to determine if an intervention is suitable for the public or not. Clinical trials are split up into four phases, where each phase includes an increasing number of people trying the treatment.
What role does artificial intelligence (AI) play into this? With all other sectors of healthcare, AI is used to optimize workflow efficiency. This can be through many parts of the clinical trial process. Processing data, minimizing error, participant management, and more.
Current trends:
Current trends in artificial intelligence are noteworthy. One company, Intelligent Medical Objects based in Rosemont, Illinois, developed what is known as SEETrials. The purpose: to extract efficacy and safety information from other clinical trials. In this case, AI is being used to support the trial design process.
One notable AI developed for clinical trials: Trial Pathfinder. A group at Stanford, led by James Zou, invented a system with the purpose of adjusting certain criteria from completed clinical trials, and noting its effects on participation. Various large companies utilize Trial Pathfinder, including Genentech, Roche, and AstraZeneca. There are a wide range of other AI intervention startups that deal with clinical trials, but I don’t want to waste time covering those.
Limitations:
With the rapid rise of AI applications, there must be a regulating body for AI, especially in clinical trials. However, as of right now, there is no standard regulatory body for artificial intelligence in clinical trials. Even so, the European Commision proposed what's called the Artificial Intelligence Act (AIA). Essentially, high-risk AI systems must meet specific requirements and undergo a performance test.
With almost everything in life, there are limitations. AI isn’t excluded from this fact. The largest limitation is concerning what the AI is given. For data analyzing AI, it is critical that the data given is accurate and representative of the whole population. There must not be any inconsistencies, flaws, or anything else of the matter. Another limitation can be lack of human judgment. What a trained professional has that AI doesn’t, is intuition, and even common sense. A third limitation is the very implementation of AI into clinical trials. The prior AI systems mentioned deal with the very design of the AI. For some clinical trials, scientists have developed them without AI in mind, and implementing them will force scientists to potentially alter their whole trial.
Conclusion:
To conclude, there are various aspects of AI that go into clinical trials. The specific application (Patient management, data analysis, etc.), to ethical considerations, to the numerous limitations. It will take some time before we can fully integrate artificial intelligence into the clinical trial space effectively and safely.
References:
Chopra, Hitesh, et al. “Revolutionizing Clinical Trials: The Role of AI in Accelerating Medical Breakthroughs.” International Journal of Surgery, vol. 109, no. 12, Elsevier BV, Oct. 2023, https://doi.org/10.1097/js9.0000000000000705.
Hutson, Matthew. “How AI Is Being Used to Accelerate Clinical Trials.” Nature, vol. 627, no. 8003, Mar. 2024, pp. S2–5, https://doi.org/10.1038/d41586-024-00753-x.