Clinical trials play a crucial role in advancing medical treatments and therapies, but the process of screening patients for eligibility can be time-consuming and inefficient. A recent study published in the journal Machine Learning: Health has explored the use of artificial intelligence, specifically ChatGPT, to streamline the patient screening process for clinical trials.
Researchers at UT Southwestern Medical Center conducted a study using ChatGPT to assess whether patients were suitable candidates for a head and neck cancer trial. By analyzing patient data with the help of GPT-3.5 and GPT-4, researchers were able to identify eligible patients within minutes, significantly reducing the screening time compared to manual methods.
The traditional process of screening patients for clinical trials involves reviewing medical records to determine if individuals meet the eligibility criteria. However, this manual process is time-consuming, taking around 40 minutes per patient, and often leads to delays in trial enrollment. With the help of AI like ChatGPT, researchers were able to expedite the screening process and identify suitable candidates more efficiently.
The study tested three different prompting methods to interact with the AI: Structured Output (SO), Chain of Thought (CoT), and Self-Discover (SD). Results showed that GPT-4 was more accurate than GPT-3.5 in identifying eligible patients, although it was slightly slower and more expensive. Screening times ranged from 1.4 to 12.4 minutes per patient, with costs varying between $0.02 and $0.27.
Dr. Mike Dohopolski, the lead author of the study, highlighted the potential of large language models like GPT-4 in accelerating the patient screening process for clinical trials. While AI models are not perfect and may require human oversight, they can significantly save time and support human reviewers in identifying suitable candidates for trials.
The research underscores the importance of AI in supporting faster and more efficient clinical trials, ultimately bringing new treatments to patients more quickly. The study is part of IOP Publishing’s Machine Learning series, the first open-access journal series dedicated to the application and development of machine learning and artificial intelligence for the sciences.
In addition to patient screening for clinical trials, the same research team has developed a method called GeoDL, which allows surgeons to adjust patients’ radiation therapy in real time. This deep learning system provides precise 3D dose estimates from CT scans and treatment data within milliseconds, making adaptive radiotherapy faster and more efficient in clinical settings.
Overall, the use of AI like ChatGPT shows great promise in revolutionizing the clinical trial process by accelerating patient screening and improving trial success rates. With continued advancements in AI technology, the future of healthcare and medical research looks brighter than ever.
