DoctorIA-MED-R-8B

DoctorIA Medical Solutions: Revolutionizing Healthcare with AI-Powered Diagnostics

[Read the Paper] [Demo] [Hugging Face Model]

DoctorIA is an innovative AI-powered solution designed to enhance radiological diagnostics, improve healthcare access, and reduce disparities in underserved areas of Morocco. Leveraging state-of-the-art machine learning techniques, DoctorIA provides accurate, efficient, and scalable diagnostic support for medical professionals.

---

Overview

DoctorIA is built to assist healthcare providers with:

• Automated Medical Image Analysis: Accurate interpretation of X-rays, MRIs, and other medical imaging technologies.
• Clinical Reasoning Support: Advanced reasoning capabilities to assist in diagnosis, treatment planning, and risk assessment.
• Healthcare Accessibility Initiatives: Bridging gaps in healthcare access by offering scalable solutions to underserved populations.

Our mission is to empower healthcare professionals and patients alike by providing cutting-edge AI-driven diagnostic tools.

---

Key Features

• AI-Driven Diagnostic Tools: Supports clinical reasoning and treatment planning by providing insights derived from advanced AI algorithms.
• Radiology Assistance: Assists radiologists with preliminary analysis of medical images.
• Patient Education: Provides clear explanations of medical procedures and technologies.
• Multilingual Support: Available in Arabic, French, English, and Spanish to cater to diverse populations.
• Scalable Deployment: Optimized for deployment in resource-constrained environments, ensuring accessibility even in underserved areas.

---

Models

We release two versions of the DoctorIA model:

1. DoctorIA-ClinicalReasoning

   • Purpose: Clinical reasoning and diagnostic support.
   • Tasks: Symptom-to-diagnosis mapping, treatment planning, and evidence-based recommendations.
   • Quantization: Available in 4-bit precision for reduced memory usage.
   • Hugging Face Repository: DoctorIA-ClinicalReasoning

2. DoctorIA-MedicalImageAnalysis

   • Purpose: Automated analysis of medical images (X-rays, MRIs, etc.).
   • Tasks: Disease detection, lesion segmentation, and abnormality classification.
   • Quantization: Available in 4-bit precision for reduced memory usage.
   • Hugging Face Repository: DoctorIA-MedicalImageAnalysis

All models are released under the Apache 2.0 License.

---

Organisation of the Repository

The repository is structured as follows:

• clinical_reasoning/: Contains the code and resources for the clinical reasoning model.
• medical_image_analysis/: Contains the code and resources for the medical image analysis model.
• examples/: Example scripts for inference, fine-tuning, and integration.
• datasets/: Links to datasets used for training and evaluation.
• notebooks/: Jupyter notebooks for experimentation and visualization.
• docs/: Additional documentation and tutorials.

---

Requirements

To use DoctorIA, you will need:

• Python 3.8 or higher (Python 3.10 recommended).
• PyTorch (torch) and Transformers (transformers) libraries.
• GPU with at least 16GB of memory (for full-precision models).

Install dependencies using:

pip install -r requirements.txt

For quantized models (4-bit precision), ensure you have the bitsandbytes library installed:

pip install bitsandbytes

---

Usage

1. Clinical Reasoning Model

Load the model and tokenizer:

from transformers import AutoTokenizer, AutoModelForSequenceClassification

model_name = "your-username/doctoria-clinical-reasoning"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)

# Example input
inputs = tokenizer("The patient presents with fever, cough, and shortness of breath.", return_tensors="pt")
outputs = model(**inputs)
predicted_diagnosis = outputs.logits.argmax().item()
print(f"Predicted Diagnosis: {predicted_diagnosis}")

2. Medical Image Analysis Model

Load the model and feature extractor:

from transformers import AutoFeatureExtractor, AutoModelForImageClassification
from PIL import Image
import requests

model_name = "your-username/doctoria-medical-image-analysis"
feature_extractor = AutoFeatureExtractor.from_pretrained(model_name)
model = AutoModelForImageClassification.from_pretrained(model_name)

# Example input
url = "https://example.com/chest-xray.jpg"
image = Image.open(requests.get(url, stream=True).raw)
inputs = feature_extractor(images=image, return_tensors="pt")
outputs = model(**inputs)
predicted_class = outputs.logits.argmax().item()
print(f"Predicted Class: {predicted_class}")

---

Benchmarks

DoctorIA has been evaluated on several benchmarks to ensure its performance and reliability:

• Clinical Reasoning: Achieved X% accuracy on the DR.BENCH benchmark for clinical diagnostic reasoning.
• Medical Image Analysis: Achieved Y% sensitivity and Z% specificity on the CheXpert benchmark for chest X-ray analysis.

For more details, refer to our paper.

---

Development

If you wish to contribute to DoctorIA or modify it for your needs:

1. Clone the repository:

   git clone https://github.com/your-username/doctoria.git
   cd doctoria
   

2. Install dependencies:

   pip install -e '.[dev]'
   

3. Run tests:

   pytest
   

---

FAQ

Check out the Frequently Asked Questions section before opening an issue.

---

License

The codebase is released under the Apache 2.0 License.
The model weights are released under the CC-BY 4.0 License.

---

Citation

If you use DoctorIA in your research or projects, please cite our work:

@techreport{doctoria2025,
      title={DoctorIA: Enhancing Radiological Diagnostics with AI},
      author={Jad Tounsi El Azzoiani and Team DoctorIA},
      year={2025},
      eprint={XXXX.XXXXX},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://example.com/paper},
}