https://github.com/priyam-hub/medicare-bot
MediCare-Bot provides clear, reliable health information by combining trusted medical sources with smart search and AI. It makes medical queries easy to understand and accessible for everyone.
https://github.com/priyam-hub/medicare-bot
chatbot groq-api langchain llama3 medical pinecone-db retrieval-augmented-generation sentence-transformers
Last synced: 3 months ago
JSON representation
MediCare-Bot provides clear, reliable health information by combining trusted medical sources with smart search and AI. It makes medical queries easy to understand and accessible for everyone.
- Host: GitHub
- URL: https://github.com/priyam-hub/medicare-bot
- Owner: priyam-hub
- License: apache-2.0
- Created: 2025-05-05T09:32:56.000Z (6 months ago)
- Default Branch: main
- Last Pushed: 2025-07-16T05:28:49.000Z (3 months ago)
- Last Synced: 2025-07-17T08:06:14.003Z (3 months ago)
- Topics: chatbot, groq-api, langchain, llama3, medical, pinecone-db, retrieval-augmented-generation, sentence-transformers
- Language: Jupyter Notebook
- Homepage:
- Size: 10.7 MB
- Stars: 2
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# 🤖 MediCare-Bot — Your trusted health companion.
[](https://www.python.org/)
[](https://flask.palletsprojects.com/)
[](LICENSE)
[](https://ai.meta.com/llama/)
[](https://www.docker.com/)
*Supporting patients with trusted answers, one conversation at a time.*
[Features](#-features) • [Installation](#-installation) • [Usage](#-usage) • [Tech Stack](#-tech-stack) • [License](#-license) • [Contact](#-contact)
---
## 🌟 Overview
**MediCare-Bot** is designed to assist users with reliable and easy-to-understand health information. It learns from trusted medical books and uses intelligent search techniques to find the most relevant content. By combining smart retrieval with a powerful language model, the chatbot offers clear and accurate responses to medical queries, making healthcare information more accessible and conversational for everyone.
---
## 📚 Dataset
**Source:** [Medical Book Dataset – Kaggle](https://www.kaggle.com/datasets/abhirajmandal/medical-book)
#### 🔍 Description
This dataset contains textual content extracted from medical books, designed to provide structured and informative knowledge on a wide range of medical topics. It includes concise explanations, definitions, symptoms, causes, and treatments for various diseases and conditions. The dataset is ideal for building knowledge-driven applications like medical chatbots, as it captures essential clinical and healthcare-related information in a digestible format.
**Use Case:**
The dataset serves as the foundational knowledge base for the chatbot, allowing it to generate helpful and accurate responses to user queries by referencing real medical literature.
---
## 📌 Features
* **Patient-Centric Design**
- Focused on delivering helpful and compassionate responses tailored to user needs.
* **Trusted Medical Knowledge**
- Built using information from reliable medical books to ensure accuracy.
* **Natural Conversations**
- Understands and responds in a human-like, conversational tone.
* **Instant Answers**
- Provides quick and relevant responses to a wide range of medical questions.
* **Context-Aware**
- Remembers and considers the context of your query for better responses.
* **User-Friendly Interface**
- Simple and clean chat interface that's easy to use for everyone.
* **Secure and Private**
- No personal data is stored or shared—ensuring privacy and confidentiality.
---
## 🛠️ Installation
#### Step - 1: Repository Cloning
```bash
# Clone the repository
git clone https://github.com/priyam-hub/MediCare-Bot.git
# Navigate into the directory
cd MediCare-Bot
```
#### Step - 2: Enviornmental Setup and Dependency Installation
```bash
# Run env_setup.sh
bash env_setup.sh
# Select 1 to create Python Environment
# Select 2 to create Conda Environment
# Python Version - 3.10
# Make the Project to run as a Local Package
python setup.py
```
#### Step - 3: Create a .env file in the root directory to add Credentials
```bash
PINECONE_API_KEY = "xxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
GROQ_API_KEY = "xxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
```
#### Step - 4: Store the Vector Embeddings & Initialize RAG
```bash
# Run the Main Python Script
python main.py
```
#### Step - 5: Run the Flask Server
```bash
# Run the Web App using Flask Server
python web/app.py
```
Upon running, navigate to the provided local URL in your browser to interact with the MediCare-Bot
---
## ⚙️ Technology Stack
* **Python** – Core programming language used for building the backend.
🔗 [Install Python](https://www.python.org/downloads/)
* **PyTorch** – Deep learning framework used under the hood for embedding models and LLMs.
🔗 [Install PyTorch](https://pytorch.org/get-started/locally/)
* **Transformers (Hugging Face)** – Used to load and manage pre-trained embedding models.
🔗 [Transformers Documentation](https://huggingface.co/docs/transformers/index)
* **LangChain** – Framework for building applications with LLMs, supporting RAG and vector search.
🔗 [LangChain Installation Guide](https://docs.langchain.com/docs/get_started/installation)
* **Flask** – Lightweight web framework to deploy the chatbot as a web API.
🔗 [Flask Installation](https://flask.palletsprojects.com/en/latest/installation/)
---
## 🧠 Artificial Intelligence Models Stack
* **Vector Embedding – `sentence-transformers/all-MiniLM-L6-v2`**
Lightweight and efficient embedding model for generating semantic vector representations of medical texts.
🔗 [Model on Hugging Face](https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2)
* **Vector Database – Pinecone**
High-performance vector store to manage and search large-scale embedding data.
🔗 [Pinecone Documentation](https://docs.pinecone.io/docs/overview)
* **Information Retrieval – RAG (Retrieval-Augmented Generation)**
Augments LLM responses by retrieving relevant content from medical embeddings.
🔗 [RAG on LangChain](https://docs.langchain.com/docs/components/retrievers/)
* **Inference Engine – Groq**
Ultra-fast inference platform to run LLaMa-3 and generate real-time answers.
🔗 [Groq API](https://groq.com/)
* **Large Language Model – LLaMa-3 8B**
Advanced open-source language model from Meta used for understanding medical context and generating responses.
🔗 [LLaMa 3 (Meta)](https://ai.meta.com/llama/)
* **API Integration – Flask**
Handles user requests and chatbot interactions through a simple web API.
🔗 [Flask Official Docs](https://flask.palletsprojects.com/en/latest/)
* **Environmental Image – Docker**
Ensures consistent, portable, and isolated runtime environment for the chatbot application.
🔗 [Docker Installation](https://docs.docker.com/get-docker/)
* **Deployment – EC2 Instance**
Scalable cloud server to host and serve the medical chatbot application.
🔗 [Amazon EC2](https://aws.amazon.com/ec2/)
---
## 🧠 Large Language Model Inference Comparison Chart
| **Model ID** | **Requests/Min** | **Requests/Day** | **Tokens/Min** | **Tokens/Day** |
| ----------------------------------------------- | ---------------- | ---------------- | -------------- | -------------- |
| `allam-2-7b` | 30 | 7,000 | 6,000 | No limit |
| `compound-beta` | 15 | 200 | 70,000 | No limit |
| `deepseek-r1-distill-llama-70b` | 30 | 1,000 | 6,000 | No limit |
| `gemma2-9b-it` | 30 | 14,400 | 15,000 | 500,000 |
| `llama3-70b-8192` | 30 | 14,400 | 6,000 | 500,000 |
| `llama3-8b-8192` | 30 | 14,400 | 6,000 | 500,000 |
| `mistral-saba-24b` | 30 | 1,000 | 6,000 | 500,000 |
| `qwen-qwq-32b` | 30 | 1,000 | 6,000 | No limit |
---
## 📊 Success Rates of Different Meta-LlaMa Models
| **Category** | **Benchmark** | **LLaMA 3 8B** | **LLaMA 2 7B** | **LLaMA 2 13B** | **LLaMA 3 70B** | **LLaMA 2 70B** |
| ------------------------- | ---------------------------- | -------------- | -------------- | --------------- | --------------- | --------------- |
| **General** | MMLU (5-shot) | 66.6 | 45.7 | 53.8 | 79.5 | 69.7 |
| | AGIEval English (3–5 shot) | 45.9 | 28.8 | 38.7 | 63.0 | 54.8 |
| | CommonSenseQA (7-shot) | 72.6 | 57.6 | 67.6 | 83.8 | 78.7 |
| | Winogrande (5-shot) | 76.1 | 73.3 | 75.4 | 83.1 | 81.8 |
| | BIG-Bench Hard (3-shot, CoT) | 61.1 | 38.1 | 47.0 | 81.3 | 65.7 |
| | ARC-Challenge (25-shot) | 78.6 | 53.7 | 67.6 | 93.0 | 85.3 |
| **Knowledge Reasoning** | TriviaQA-Wiki (5-shot) | 78.5 | 72.1 | 79.6 | 89.7 | 87.5 |
| **Reading Comprehension** | SQuAD (1-shot) | 76.4 | 72.2 | 72.1 | 85.6 | 82.6 |
| | QuAC (1-shot, F1) | 44.4 | 39.6 | 44.9 | 51.1 | 49.4 |
| | BoolQ (0-shot) | 75.7 | 65.5 | 66.9 | 79.0 | 73.1 |
| | DROP (3-shot, F1) | 58.4 | 37.9 | 49.8 | 79.7 | 70.2 |
---
## 📁 Project Structure
```plaintext
MediCare-Bot/
├── .docker_ignore # Ignoring Files in Docker
├── .env # Store the Pinecone and Groq Credentials
├── .gitignore # Ignoring files for Git
├── Dockerfile # Stored the Docker Setup
├── env_setup.sh # Package installation configuration
├── folder_structure.py # Contains the Project Folder Structure
├── LICENCE # MIT License
├── main.py # Store Vector Embeddings and Initialize the RAG
├── README.md # Project documentation
├── requirements.txt # Python dependencies
├── setup.py # Create the Project as Python Package
├── config/ # Configuration files
│ ├── __init__.py
│ └── config.py/ # All Configuration Variables of Pipeline
├── data/ # Data directory
│ └── Medical_Book.pdf # Medical Book Dataset used for Vector Embedding
├── notebooks/ # Jupyter notebooks for experimentation
│ └── Medical_Chatbot.ipynb # Experimented Chatbot in Jupyter Notebook
├── src/ # Source code
│ ├── llm/ # Large Language Model Directory
│ │ ├── __init__.py
│ │ └── llm_builder.py # Python file to Build the LLM using ChatGroq
│ ├── prompts/ # System Prompt Directory
│ │ ├── __init__.py
│ │ └── prompt_builder.py # Python file Build the Prompt for LLM Inference
│ ├── text_splitting/ # Text Splitting Directory
│ │ ├── __init__.py
│ │ └── split_text.py # Python File to Split the Text
│ ├── vector_index/ # Vector Index Directory
│ │ ├── __init__.py
│ │ └── index_manager.py # Python file to create index in Vector DB
│ └── utils/ # Utility Functions Directory
│ ├── __init__.py
│ ├── download_embeddings.py # Download the Embedding Model
│ ├── load_pdf.py # Load the PDF for Embedding
│ └── logger.py # Logger Setup
└── web/
├── __init__.py
├── animations/
│ ├── chatbot.json # Chatbot Animations
│ └── doctor.json # Doctor Animations
├── static/
│ ├── style.css # Styling of the Web Page
├── templates/
│ ├── chat.html # Default Web Page
└── app.py/ # To run the flask server
```
---
## 🛠️ Future Work Roadmap
### **🔹 Phase 1: Feature Expansion (1–2 Months)**
**Goal**: Enhance user experience and improve core functionality.
* Integrate symptom checker with dynamic conversation flow.
* Add multilingual support for broader accessibility.
* Implement feedback mechanism for users to rate chatbot responses.
* Improve handling of edge cases in medical queries.
### **🔹 Phase 2: Clinical Reliability & Compliance (3–4 Months)**
**Goal**: Increase medical accuracy and build trust with regulatory alignment.
* Collaborate with healthcare professionals for validation of responses.
* Include references to verified medical literature in answers.
* Ensure HIPAA/GDPR compliance for secure data handling.
* Add a disclaimer and escalation option for complex queries.
### **🔹 Phase 3: Real-Time Integration & Monitoring (5–6 Months)**
**Goal**: Make the chatbot production-ready for deployment in real environments.
* Integrate with wearable health devices and EMR/EHR systems.
* Deploy monitoring system to track chatbot performance and anomalies.
* Introduce voice interaction using Whisper or similar model.
* Host full system with CI/CD on scalable cloud infrastructure (e.g., AWS/GCP).
---
## 📜 License
This project is licensed under the **MIT License**. See the [LICENSE](LICENSE) file for more details.
---
**Made by Priyam Pal - AI and Data Science Engineer**
[↑ Back to Top](#-affective-ai--understanding-emotions-through-text)