Matching algorithm using the scispacy language models.
Please use file - scispacy_loinc_mapping_algorithm.ipynb for using the matchng algorithm
This project implements a robust pipeline to map lab terms (e.g., "MCH", "heart rate") to their most appropriate LOINC codes, using a combination of:
- ✅ Axis-based matching (
COMPONENT
,PROPERTY
,SYSTEM
,METHOD
) - 📊 LOINC's official
COMMON_TEST_RANK
- 🧮 Custom composite scoring
- ⚙️ Batch-based processing for large-scale performance
- 🛠️ Hardcoded fallbacks for known ambiguous terms
The algorithm returns the best match candidates for each input term.
Make sure the following files are present in the run environment:
loinc.csv
LoincPartLink_Primary.csv
These files are available for download from https://loinc.org/downloads/loinc/.