Clinical ICD-10 Extractor

Clinical coding application that extracts ICD-10 diagnoses from clinical documentation and maps supported conditions to Hierarchical Condition Categories (HCCs).

🚀 Live Demo💻 View Source Code

Key Features

  • ✅ Extracts ICD-10 diagnosis codes from free-text clinical documentation
  • ✅ Maps supported diagnoses to Hierarchical Condition Categories (HCCs)
  • ✅ Accepts both narrative notes and ICD-10 code input
  • ✅ Supports multiple-condition extraction
  • ✅ Interactive web interface
  • ✅ Publicly deployed on Hugging Face Spaces

Supported Conditions

  • • Type 2 Diabetes Mellitus
  • • Chronic Kidney Disease
  • • HIV Disease
  • • Acute Myeloid Leukemia
  • • Multiple-condition detection
  • • Direct ICD-10 code input

Application Preview

Clinical ICD-10 Extractor Architecture

Technology Stack

  • 🐍 Python
  • ⚡ Streamlit
  • 🤗 Hugging Face Spaces
  • 🐙 GitHub
  • 📘 ICD-10-CM
  • 🏥 CMS HCC Mapping

Future Roadmap

  • ☐ Expand ICD-10 condition coverage
  • ☐ Improve clinical rule engine
  • ☐ Negation detection
  • ☐ Context-aware extraction
  • ☐ FHIR-compatible outputs
  • ☐ Export results as CSV and JSON

Case Study

A detailed technical and clinical case study describing the architecture, design decisions, implementation, evaluation, and future development of this project is currently being prepared. The project focuses on a hybrid NLP and rule-based pipeline that extracts ICD-10 diagnosis codes from unstructured clinical narratives and maps them to Hierarchical Condition Categories (HCCs) to support healthcare risk adjustment and downstream analytics.

📌 Detailed technical documentation and case study coming soon.