# Automated Medical Coding

## Overview
Automated Medical Coding is an AI-driven model designed to streamline the process of extracting and assigning medical codes from clinical notes. This model leverages natural language processing (NLP) to predict **ICD (International Classification of Diseases)** and **CPT (Current Procedural Terminology)** codes based on unstructured text data, such as physician notes or medical documentation.

Medical coding is a critical step in healthcare, facilitating accurate billing, claims processing, and statistical tracking. By automating this process, our model reduces manual effort, enhances accuracy, and saves time for healthcare providers.

## Features
- Predicts **ICD codes**, which categorize diagnoses and medical conditions.
- Predicts **CPT codes**, which detail medical services and procedures.
- Designed to handle clinical notes with complex, unstructured language.

## Base Model
This model builds upon the **[Microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract]**, a pretrained transformer model fine-tuned for medical text understanding. BiomedBERT's capability to process medical jargon makes it an ideal foundation for this task.

## How It Works
1. **Input:** Clinical notes or medical documentation in textual format.
2. **Processing:** The input text is tokenized and passed through BiomedBERT for feature extraction. Additional fully connected layers process these features to predict corresponding ICD and CPT codes.
3. **Output:** A list of ICD and CPT codes relevant to the input clinical notes.

## Benefits
- **Improved Efficiency:** Reduces manual coding time for medical professionals.
- **Increased Accuracy:** Minimizes errors in coding and improves billing accuracy.
- **Scalability:** Can process large volumes of clinical notes effectively.

## Sample Model Prediction


![image/png](https://cdn-uploads.huggingface.co/production/uploads/67647d81585fbf029d3abfcf/5qQJbYt-bocC0wRHcwAfA.png)