Identifying symptom etiologies using syntactic patterns and large language models.

Journal: Scientific reports
Published Date:

Abstract

Differential diagnosis is a crucial aspect of medical practice, as it guides clinicians to accurate diagnoses and effective treatment plans. Traditional resources, such as medical books and services like UpToDate, are constrained by manual curation, potentially missing out on novel or less common findings. This paper introduces and analyzes two novel methods to mine etiologies from scientific literature. The first method employs a traditional Natural Language Processing (NLP) approach based on syntactic patterns. By using a novel application of human-guided pattern bootstrapping patterns are derived quickly, and symptom etiologies are extracted with significant coverage. The second method utilizes generative models, specifically GPT-4, coupled with a fact verification pipeline, marking a pioneering application of generative techniques in etiology extraction. Analyzing this second method shows that while it is highly precise, it offers lesser coverage compared to the syntactic approach. Importantly, combining both methodologies yields synergistic outcomes, enhancing the depth and reliability of etiology mining.

Authors

  • Hillel Taub-Tabib
    Allen Institute for AI, Seattle, USA.
  • Yosi Shamay
    Faculty of Biomedical Engineering, Technion, Haifa, Israel.
  • Micah Shlain
    Allen Institute for AI, Seattle, USA.
  • Menny Pinhasov
    Allen Institute for AI, Seattle, USA.
  • Mark Polak
    Allen Institute for AI, Seattle, USA.
  • Aryeh Tiktinsky
    Allen Institute for AI, Seattle, USA.
  • Sigal Rahamimov
    Allen Institute for AI, Seattle, USA.
  • Dan Bareket
    Allen Institute for AI, Seattle, USA.
  • Ben Eyal
    Allen Institute for AI, Seattle, USA.
  • Moriya Kassis
    Allen Institute for AI, Seattle, USA.
  • Yoav Goldberg
    Department of Computer Science, Bar Ilan University, Tel Aviv, Israel.
  • Tal Kaminski Rosenberg
    Technion Faculty of Medicine Library and Rambam Health Campus Library, Haifa, Israel.
  • Simon Vulfsons
    Institute for Pain Medicine, Rambam Health Campus, Haifa, Israel.
  • Maayan Ben Sasson
    Institute for Pain Medicine, Rambam Health Campus, Haifa, Israel. maayanbensasson@gmail.com.