Optimizing Rare Disease Patient Matching with Large Language Models
We present RepLLaMA, a neural ranking model for optimizing patient matching in rare disease communities. Using data from Unrare.me consisting of over two thousand profiles and over ten thousand ratings, our bi-encoder architecture maps profiles to 4096-dimensional vectors, enabling efficient similarity computations. The system processes unstructured symptom descriptions and structured responses, incorporating expert-guided LLM enhancements. Results show Top-10 Recall of 49.36% <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\pm 2.03)$</tex>, surpassing baselines while maintaining generalization. The implementation provides a scalable solution for rare disease patient matching, addressing computational complexity challenges.
Authors
- Tobias Deußer (ORCID: https://orcid.org/0000-0003-4685-0847)
- Ali Hamza Bashir
- Lorenz Grigull (ORCID: https://orcid.org/0000-0001-8807-2874)
- David Berghaus (ORCID: https://orcid.org/0000-0002-0740-154X)
- Max Lübbering (ORCID: https://orcid.org/0000-0001-6291-9459)
- Hubert Högl
- Rafet Sifa (ORCID: https://orcid.org/0009-0004-6680-8210)
- Armin Berger (ORCID: https://orcid.org/0009-0003-1924-8401)
- Ralf Schmidt (ORCID: https://orcid.org/0000-0002-1224-2678)
- Gundula Ernst (ORCID: https://orcid.org/0000-0002-2483-7206)
- Thiago Bell
- David Bascom
- Lara Fendrich (ORCID: https://orcid.org/0000-0002-8063-0373)
- Tom Anglim Lagones
- Mowmita
- Nazia Afsan
Institutions
- Queensland Health (AU)
- University of Bonn (DE)
- University Hospital Bonn (DE)
- Government of Western Australia Department of Health (AU)
- Medizinische Hochschule Hannover (DE)
Publication Details
- Published
- 2024-12-15
- DOI
- https://doi.org/10.1109/bigdata62323.2024.10910113
- Citations
- 1
- Primary Topic
- Biomedical Text Mining and Ontologies
- Type
- article
- Field-Weighted Citation Impact
- 0.23
Funders
- Bundesministerium für Bildung und Forschung