Summary of Zero-shot Clinical Trial Patient Matching with Llms, by Michael Wornow et al.
Zero-Shot Clinical Trial Patient Matching with LLMs
by Michael Wornow, Alejandro Lozano, Dev Dash, Jenelle Jindal, Kenneth W. Mahaffey, Nigam H. Shah
First submitted to arxiv on: 5 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper explores the application of large language models (LLMs) in matching patients to clinical trials, aiming to automate the process of identifying patients who meet eligibility criteria. The authors design an LLM-based system that evaluates patient medical histories as unstructured text and compares it with inclusion criteria specified as free text. The system achieves state-of-the-art scores on the n2c2 2018 cohort selection benchmark in a zero-shot setting. To improve efficiency, the authors identify a prompting strategy that accelerates matching by an order of magnitude and develop a two-stage retrieval pipeline that reduces processing tokens while maintaining performance. Additionally, they evaluate the interpretability of their system by having clinicians review natural language justifications generated for each eligibility decision, demonstrating coherent explanations for 97% of correct decisions and 75% of incorrect ones. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper uses special computers to help match patients with clinical trials. Right now, this process takes a long time because it’s done manually. The researchers created a system that can do this job much faster using special computer models called large language models. These models are good at understanding text and can read patient medical histories to see if they qualify for a trial. The new system is very accurate and can even explain why it thinks a patient is or isn’t eligible. This is important because it could help bring new treatments to people faster. |
Keywords
» Artificial intelligence » Prompting » Zero shot
