This project is a content-based filtering system that recommends medicines based on user-reported symptoms. It uses TF-IDF vectorization and cosine similarity to match symptoms with relevant medicines from a curated dataset.
- 🧾 Objective: Suggest appropriate medicines based on user-input symptoms
- 🔍 Approach: NLP + content-based filtering (no ML training required)
- 🧰 Tech Stack: Python, Pandas, Scikit-learn (TF-IDF), Streamlit (optional for UI)
- 📁 Dataset: Contains medicines with associated symptoms and usage descriptions
- Preprocessing: Text cleaning and normalization of symptom data
- TF-IDF Vectorization: Transform symptom descriptions into numerical vectors
- Cosine Similarity: Compare user-input symptoms against dataset entries
- Top-N Recommendations: Return most relevant medicines sorted by similarity
User Input:
"fever, body ache, chills"
Top Recommendations:
- Paracetamol — Commonly used for fever and mild pain
- Ibuprofen — Pain reliever and anti-inflammatory
- Dolo 650 — Effective for body pain and high fever
-
Add spell correction and fuzzy matching for symptoms
-
Incorporate severity and dosage recommendations
-
Integrate with a chatbot or voice assistant
-
Add multilingual support (e.g., Hindi, regional languages)