Check Eligibility for Participation in a Clinical Study
In the realm of healthcare, traditional patient screening for clinical trials can be a time-consuming process, often taking an average of 40 minutes per patient. This demanding pace poses a challenge for already burdened clinical teams.
However, a groundbreaking study published in the journal Machine Learning: Health, led by Dr. Mike Dohopolski, has shown that ChatGPT, a leading large language model (LLM), can significantly accelerate this process. The research, which involved analysing data from 74 patients being considered for a clinical trial focused on head and neck cancer, used three distinct prompting techniques - Structured Output (SO), Chain of Thought (CoT), and Self-Discover (SD).
The study, which was published by Dr. Dohopolski and his team, demonstrates that ChatGPT can parse patient records and evaluate their eligibility for clinical trials much faster than traditional methods, without compromising accuracy. In fact, GPT-4, the model used in the study, outperformed its predecessor GPT-3.5, with screening per patient ranging from approximately 1.4 minutes to over 12 minutes, a substantial improvement over the conventional 40-minute manual review.
The associated costs per screening varied from $0.02 to $0.27, making this an economically viable tool for healthcare providers seeking to scale up their trial recruitment efforts.
This study is the inaugural article in IOP Publishing's Machine Learning seriesTM, the world's first open access journal series dedicated to machine learning and AI applications in the sciences.
The benefits observed in the use of AI at UT Southwestern can be generalized broadly within medicine, ultimately transforming patient outcomes on a global scale. The integration of AI has broader implications for the entire clinical trial ecosystem, potentially reducing the time and cost of trials, improving the diversity and representativeness of participant pools, and hastening the introduction of novel therapies to market.
Recognizing the chronic bottleneck in patient enrollment for clinical trials, which can prevent up to 20% of National Cancer Institute (NCI)-affiliated trials from reaching necessary enrollment thresholds, researchers turned to the capabilities of cutting-edge language models like GPT-3.5 and GPT-4.
The use of ChatGPT for clinical trial screening is set to expand the role of AI in healthcare, pushing the boundaries of what's possible in diagnostics, treatment, and research. However, it's important to note that large language models are not infallible and should be viewed as a powerful adjunct to human reviewers, offering scalability and consistency that can enhance clinical workflow and decision-making.
Further studies will be essential to validate and refine AI models across diverse patient populations and trial types. The demonstrated ability of large language models to handle complex, unstructured medical data rapidly and accurately signals a future where clinicians are empowered by intelligent tools like GeoDL, a deep learning system for real-time adjustment of radiation therapy during treatment, exemplifying how deep learning can enhance the accuracy and efficiency of adaptive radiotherapy in clinical settings.
These advancements in AI are poised to inspire new research, integrating with electronic health record platforms and creating hybrid AI-human workflows, ultimately revolutionizing the way we approach healthcare and clinical trials.
