Previously submitted to: Journal of Medical Internet Research (no longer under consideration since Nov 13, 2024)
Date Submitted: Sep 4, 2024
Open Peer Review Period: Sep 5, 2024 - Nov 5, 2024
(closed for review but you can still tweet)
NOTE: This is an unreviewed Preprint
Warning: This is a unreviewed preprint (What is a preprint?). Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn (a note "no longer under consideration" will appear above).
Peer review me: Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period (in this case, a "Peer Review Me" button to sign up as reviewer is displayed above). All preprints currently open for review are listed here. Outside of the formal open peer-review period we encourage you to tweet about the preprint.
Citation: Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author).
Final version: If our system detects a final peer-reviewed "version of record" (VoR) published in any journal, a link to that VoR will appear below. Readers are then encourage to cite the VoR instead of this preprint.
Settings: If you are the author, you can login and change the preprint display settings, but the preprint URL/DOI is supposed to be stable and citable, so it should not be removed once posted.
Submit: To post your own preprint, simply submit to any JMIR journal, and choose the appropriate settings to expose your submitted version as preprint.
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Medical Ontology can significantly enhance the performance of mainstream text embedding models in retrieving medical information: Exploration research
ABSTRACT
Background:
In recent years, text embedding models and their associated vector search techniques have seen significant advancements. These technologies have become crucial for deploying large language models in specialized fields like medicine. However, the inherent complexity of domain-specific knowledge still poses challenges for mainstream text embedding models. These models often struggle to accurately interpret and represent specialized terminology. The medical field, with its extensive and diverse terminology, highlights a gap in systematically evaluating how this terminological diversity affects the precision of vector-based information retrieval for text embedding models.
Objective:
This study seeks to evaluate how the diversity of medical terminology affects vector-based information retrieval for text embedding models. Additionally, it aims to identify and propose potential strategies to mitigate these effects.
Methods:
We initially developed a medical knowledge base comprising 9715 sentences, utilizing phenotypic terms and their definitions from the Human Phenotype Ontology. This knowledge base was then vectorized using 19 representative text embedding models and stored in a vector database separately. Following this, we conducted a series of information retrieval experiments to assess the efficacy of vector-based retrieval with both preferred and non-preferred phenotypic terms across the different text embedding models. This approach allowed us to thoroughly analyze how the diversity of medical terminology affects retrieval accuracy.
Results:
Our findings reveal that the diversity of medical terminology substantially affects the vector-based information retrieval accuracy of text embedding models for medical texts. Besides, it is found that the greater the literal or semantic differences between preferred and non-preferred terms for a given medical concept, the more significant the impact on vector-based information retrieval accuracy. Furthermore, current text embedding models demonstrate markedly weaker performance in retrieving cross-lingual medical texts compared to texts within the same language. Utilizing medical ontologies to standardize concepts in medical texts can significantly improve the vector-based information retrieval accuracy of these models.
Conclusions:
The diversity of medical terminology has a profound impact on the vector-based information retrieval accuracy of text embedding models for medical texts. Employing medical ontologies to standardize the concepts within these texts represents a highly effective strategy for enhancing the retrieval accuracy of existing text embedding models.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.