meghamkpatel / physicaltherapyassistant

PhysioPhrame is a specialized tool for physical therapists that enhances patient care by enabling quick access to extensive physical therapy knowledge through natural language queries and a Pinecone vector database for fast and accurate information retrieval.

PhysioPhrame

Link to the app: https://physicaltherapyassistant-bbxcbhlrxm9de3yd7szjnz.streamlit.app/

Description

PhysioPhrame is a domain-specific application designed to assist physical therapists and healthcare professionals by providing quick access to information through natural language queries. Leveraging the power of Large Language Models (LLM) for semantic understanding and a vector database for efficient data storage and retrieval, PhysioPhrame offers an innovative way to access a wealth of physical therapy knowledge. Whether you're looking for possible diagnoses, appropriate tests, or key considerations during patient assessment, PhysioPhrame is your go-to assistant.

Features

• Semantic Search: Utilizes a vector database to index and retrieve data based on semantic similarity, ensuring that responses are relevant and accurate.
• Natural Language Interface: Powered by OpenAI's LLM, the application processes natural language queries, allowing for intuitive interaction.
• Real-time Response Generation: Generates responses in real-time, offering quick access to information without navigating through traditional databases.
• Chat History: Maintains a session-based chat history for the continuity of interaction, enhancing the user experience by keeping track of the conversation flow.

Supplementary Scripts and Their Roles

PDF to Text Conversion (pdftotxt.py)

To ensure the application's semantic search engine has access to a broad range of resources, the pdftotxt.py script plays a pivotal role in preprocessing. It transforms PDF documents into text files, making the content available for further processing. This step is vital for extracting meaningful information from PDFs, a common format in medical documentation and research papers.

Data Indexing (indexing.py)

The indexing.py script is at the heart of populating the vector database, setting the stage for the application's semantic search capabilities. It processes and chunks text documents into digestible segments, generates embeddings using LLM, and indexes them along with metadata in Pinecone. This meticulous organization facilitates the efficient retrieval of semantically similar content, underpinning the application's quick and relevant responses.

Enhancing Interactivity with PTAssistant (PTAssistant.py)

Building on the foundation laid by the core application, PTAssistant.py explores the capabilities of the OpenAI Assistant API to further refine user interactions. This script exemplifies how advanced AI models can be leveraged to not only understand but also anticipate user queries, offering nuanced and context-aware responses. It marks a significant advancement in making the application not just a tool but a conversational partner, capable of guiding users through complex information with ease.

How to Run

Prerequisites

• Python 3.8 or higher
• Streamlit
• OpenAI API Key
• Pinecone API Key
• dotenv package for managing environment variables

Setup

1. Clone the Repository: Start by cloning the repository where PhysioPhrame is stored to your local machine.

2. Install Dependencies: Install the required Python packages by running the following command in your terminal:

    pip install streamlit openai pinecone-client python-dotenv
   

3. Set Up Environment Variables: Create a .env file in the root directory of the project and add your OpenAI and Pinecone API keys as follows:

   OPENAI_API_KEY='your_openai_api_key'
   PINECONE_API_KEY='your_pinecone_api_key'
   

Running the Application

1. Navigate to the project directory in your terminal.

2. Run the Streamlit application using the command: streamlit run PTRAG.py

3. Streamlit will start the application and provide you with a local URL to access PhysioPhrame.

4. Open the provided URL in your web browser to interact with the application. Input your queries related to physical therapy, and PhysioPhrame will assist you with relevant information.

Testing and Evaluation

PhysioPhrame Testing and Evaluation Notebook

To ensure the reliability and effectiveness of PhysioPhrame, a comprehensive testing and evaluation process has been conducted using the PhysioPhrame_Testing_and_Evaluation.ipynb Colab notebook. This notebook focuses on implementing Retrieval-Augmented Generation (RAG) with Langchain and OpenAI to assess the application's performance in answering questions based on a provided transcription.

Retrieval-Augmented Generation (RAG) with Langchain and OpenAI

The notebook is structured into various sections, each serving a specific purpose in the evaluation process:

Introduction: The notebook sets the context by explaining the objective of creating a chatbot capable of answering questions based on a transcription.

Section 1: Load and Split Transcription

• Imports necessary libraries.
• Loads environment variables, including the OpenAI API key.
• Reads and splits the transcription text into manageable chunks for processing.

Section 2: Compute Similarity and Find Relevant Chunks

• Initializes OpenAI embeddings.
• Defines a function to compute embeddings for a given question and find the most similar chunk from the transcription.

Section 3: Create a Knowledge Base and Generate Test Set

• Constructs a knowledge base from the transcription chunks.
• Generates a test set comprising 20 questions to evaluate the chatbot's performance.

Section 4: Evaluate the Model on the Test Set

• Initializes the OpenAI model for evaluation.
• Defines an answer function to generate responses based on the context and question.
• Evaluates the model's performance on the generated test set and knowledge base.

Summary and Results

• Displays the evaluation report to provide insights into the model's performance, highlighting its strengths and areas for improvement.

Summary and Results

The evaluation report provides a detailed analysis of how well the model performed on the test set. It offers valuable insights into the chatbot's ability to comprehend and respond to queries based on the provided transcription.

print(report)

By examining this report, stakeholders can gauge the effectiveness of PhysioPhrame in handling natural language queries related to physical therapy, ensuring that the application meets the intended objectives and delivers accurate and relevant information to its users.

Through rigorous testing and evaluation, PhysioPhrame aims to establish itself as a reliable and efficient tool for physical therapists and healthcare professionals, enhancing their decision-making process and improving patient care outcomes.

---

Note: Replace placeholders (e.g., 'your_openai_api_key', 'your_pinecone_api_key') with your actual API keys. Ensure that your Pinecone account and index are correctly configured to match the script's requirements.

For any issues or further customization, refer to the official documentation of the used APIs and libraries.

License

MIT License

Copyright (c) 2024 Megha Patel

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.