README.md

---
license: mit
language:
- en
pretty_name: "MrGO Clinical Coding"  # Example: SQuAD
task_categories:
  - question-answering
model_creator: Chime Ogbuji
size_categories:
  - "100K<n<1M"
source_datasets:
- "FMA"  # Example: wikipedia
tags:
- medical
- health  
- instruct
- clinical coding
- finetune
---

# Mr. Grammatical Ontology: Clinical Coding

This dataset was created from a motivation to train Medical Large Language Models for improved fluency in clinical 
coding, as measurable by [MedConceptsQA](https://github.com/nadavlab/MedConceptsQA), an open-source medical coding 
evaluation benchmark designed to evaluate the understanding and reasoning capabilities of LLMs on medical concepts.  It was extracted from the Centers for Medicare & Medicaid Services' 
[International Classification of Diseases, Tenth Revision, Clinical Modification](https://www.cms.gov/medicare/coding-billing/icd-10-codes) 
2025 files, for "patient encounters occurring from October 1, 2024 – September 30, 2025."

It includes code description instructions such as:

```json
{"input": "What is the description of the medical code I10 in standard medical terminology?", 
 "output": "The medical code I10 has a description of 'Essential (primary) hypertension' in ICD-10 CM."}
```

code chapter instructions such as:

```json
{"input": "What ICD-10 CM chapter is the medical code I10 in?", 
 "output": "The medical code I10 appears in the 'Hypertensive diseases' chapter of ICD-10 CM, which includes codes I10 through I1A."}
```
and synonymy instructions such as:

```json
{"input": "What are some synonyms and alternative names for the medical code I11.0 in standard medical terminology?", 
 "output": "The medical code I11.0 is also called 'Hypertensive heart failure' in ICD-10 CM."}
```