ICD-10-Code-Prediction
| Property | Value |
|---|---|
| Author | AkshatSurolia |
| License | Apache 2.0 |
| Downloads | 2,216 |
| Framework | PyTorch |
What is ICD-10-Code-Prediction?
This is a specialized Clinical BERT model designed to automatically predict ICD-10 codes from clinical text. It leverages advanced transformer architecture and can be initialized with either BERT-Base or BioBERT, trained on MIMIC clinical notes or discharge summaries specifically.
Implementation Details
The model is implemented using the transformers library and PyTorch framework. It utilizes BERT's sequence classification capabilities with specialized clinical training. The model can process clinical diagnosis text and output the most likely ICD-10 codes, making it valuable for medical coding automation.
- Built on Clinical BERT embeddings with multiple initialization options
- Trained on comprehensive MIMIC clinical notes database
- Supports both BERT-Base and BioBERT foundations
- Provides top-5 ICD-10 code predictions
Core Capabilities
- Automated ICD-10 code prediction from clinical text
- Processing of both general clinical notes and discharge summaries
- Easy integration with the transformers library
- Batch processing of clinical documents
- Top-k prediction support for multiple code suggestions
Frequently Asked Questions
Q: What makes this model unique?
This model combines clinical domain expertise with state-of-the-art transformer architecture, specifically trained on medical text to accurately predict ICD-10 codes, which is a crucial task in healthcare administration and billing.
Q: What are the recommended use cases?
The model is ideal for healthcare facilities looking to automate medical coding processes, research institutions studying clinical text analysis, and developers building healthcare automation tools. It's particularly useful for processing clinical notes and discharge summaries.
