AI-based-Medical-transcription-analysis-and-prediction

Physician Notetaker

📌 Overview

This project is a Physician Notetaker application that leverages Natural Language Processing (NLP) to extract medical entities, generate SOAP notes, summarize conversations, and analyze sentiments from physician-patient dialogues.

🚀 Features

• Medical Entity Extraction: Identifies symptoms, diagnosis, treatment, and prognosis using SpaCy's en_core_sci_md model.
• Text Summarization: Uses facebook/bart-large-cnn transformer model to summarize conversations.
• Sentiment Analysis: Determines if the patient's tone is "Anxious" or "Reassured" using TextBlob.
• SOAP Note Generation: Automatically structures patient interactions into Subjective, Objective, Assessment, and Plan format.
• Streamlit UI: Provides an interactive web interface.

---

🛠 Installation & Setup

1️⃣ Install Dependencies

Ensure you have Python installed. Then, install the required libraries:

pip install --pre torch torchvision torchaudio -i https://download.pytorch.org/whl/nightly/cu118
pip install spacy transformers textblob streamlit

2️⃣ Download the Biomedical NLP Model

Download the en_core_sci_md model from SciSpaCy:

pip install "path_to_downloaded_file"

3️⃣ Run the Application

streamlit run app.py

---

🔍 Methodologies Used

1️⃣ Medical Entity Extraction

• Model Used: en_core_sci_md (SciSpaCy)
• Approach: Identifies and categorizes entities such as symptoms, diagnosis, treatments, etc.

2️⃣ Text Summarization

• Model Used: facebook/bart-large-cnn
• Approach: Summarizes long physician-patient dialogues into concise text.

3️⃣ Sentiment Analysis

• Library Used: TextBlob
• Approach: Determines sentiment polarity and classifies it as "Anxious" or "Reassured."

4️⃣ SOAP Note Generation

• Custom implementation: Converts extracted entities into a structured SOAP note format.

---

• Live Demo (if applicable): https://drive.google.com/file/d/1Q30wE9wIcv0MmXwEaaptb6alkksf-GYM/view?usp=sharing

---

🤝 Contributors

Developed by Syed Abdul Kareem Ahmed as an assignment for Emitrr.