🧬 MedCurie (Experimental Medical T5)
Overview
MedCurie is an experimental medical language model created by fine-tuning a T5-based architecture for healthcare-oriented text-to-text tasks.
The goal of this project is to explore domain adaptation of general language models into the medical and biological domain.
Supported task styles include:
- Symptom extraction
- Disease-related question answering
- Clinical-style summarization
- Prompt-based diagnostic generation
⚠️ Important Disclaimer
MedCurie is intended strictly for research and educational purposes.
It is not clinically validated, not a medical device, and must not be used for diagnosis, treatment, or real-world healthcare decision-making.
Base Model
MedCurie is built on:
T5 (Text-to-Text Transfer Transformer) treats every NLP task as a text generation problem.
Example Prompts
diagnose: fever, cough, fatigue
list symptoms: Type 2 Diabetes
cause: What causes Asthma?
summarize: Acute bacterial pneumonia in elderly patient
extract symptoms: Patient reports joint pain and morning stiffness.
medical consultation: Severe headache and blurred vision for 3 days.
Performance & Limitations
- May partially copy or repeat prompts
- May generate incomplete or inaccurate medical statements
- Limited reasoning and clinical understanding
- Sensitive to prompt wording
- Not aligned for safety-critical deployment
This model should be considered a research prototype.
Intended Use
MedCurie may be useful for:
- Academic research in medical NLP
- Educational experimentation
- Domain adaptation studies
- Portfolio demonstration projects
Future Improvements
Planned next steps:
- Improved dataset curation
- Learning rate stabilization
- Instruction-tuned base models (e.g., Flan-T5)
- Stronger evaluation benchmarks
- Improved generation control and safety alignment
License
This model is released under the CC-BY-NC-4.0 license.
Commercial use is not permitted.
Citation
If you use MedCurie in research or experimentation, please cite this repository.
