https://github.com/basharatwali/medicine_rec_system
https://github.com/basharatwali/medicine_rec_system
Last synced: 2 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/basharatwali/medicine_rec_system
- Owner: BasharatWali
- Created: 2025-01-29T11:30:23.000Z (4 months ago)
- Default Branch: main
- Last Pushed: 2025-01-29T11:33:23.000Z (4 months ago)
- Last Synced: 2025-01-29T12:29:15.891Z (4 months ago)
- Language: Jupyter Notebook
- Size: 0 Bytes
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
This repository demonstrates a medicine recommendation system that leverages ClinicalBERT and PubMedBERT from Hugging Face for embedding generation of patient and prescription data. After creating high-quality embeddings, we apply Locality-Sensitive Hashing (LSH) to efficiently compute similarity and recommend the top three medicines.
Our approach is evaluated against a SimHash-based system on two main metrics:
Time Efficiency – How quickly top-3 recommendations are generated.
Similarity Score – How relevant the recommended medicines are, based on embedding proximity.
Results demonstrate that leveraging these LLM-based models (ClinicalBERT & PubMedBERT) alongside LSH outperforms SimHash in both speed and accuracy, making it a promising solution for real-time, high-accuracy medical recommendation scenarios.
Features:
Embedding generation with ClinicalBERT and PubMedBERT.
Locality-Sensitive Hashing (LSH) for fast, approximate nearest-neighbor retrieval.
Easy comparison of SimHash vs. LSH results on time and similarity.
Use Cases:
Personalized medicine recommendations.
Clinical decision support systems requiring efficient and accurate retrieval.
Research on embedding-based retrieval methods in healthcare