|
|
--- |
|
|
language: |
|
|
- ar |
|
|
- en |
|
|
license: mit |
|
|
tags: |
|
|
- medical |
|
|
- ai |
|
|
- hybrid |
|
|
- data-analysis |
|
|
- arabic |
|
|
- english |
|
|
model_name: Nabdh |
|
|
model_type: hybrid-ai |
|
|
pipeline_tag: text-classification |
|
|
base_model: custom |
|
|
datasets: |
|
|
- laboconsulte/medical-data |
|
|
library_name: transformers |
|
|
--- |
|
|
|
|
|
## Model Overview |
|
|
**Name:** Nabdh (نَبض) |
|
|
**Version:** 1.0 |
|
|
**Developer:** LaboConsulte AI Team — By Aymen Messouci |
|
|
**Release Date:** October 2025 |
|
|
**Type:** Hybrid AI Model for Medical Data Analysis |
|
|
**Languages:** Arabic, English |
|
|
|
|
|
--- |
|
|
|
|
|
## Model Purpose |
|
|
Nabdh is an AI model designed to assist doctors and users in understanding and interpreting **medical test results** and **clinical data**. |
|
|
It provides personalized insights, early health pattern detection, and AI-powered recommendations. |
|
|
|
|
|
--- |
|
|
|
|
|
## Intended Use |
|
|
- **Doctors:** Support decision-making in case evaluations. |
|
|
- **Patients:** Simplified health explanations. |
|
|
- **Labs:** Integration into digital health systems. |
|
|
|
|
|
⚠️ *Disclaimer:* Nabdh is **not a diagnostic tool**. It provides supportive insights only. |
|
|
|
|
|
--- |
|
|
|
|
|
## Model Architecture |
|
|
- **Inputs:** Laboratory results, textual descriptions, and clinical indicators. |
|
|
- **Core Components:** |
|
|
- NLP unit for understanding health descriptions. |
|
|
- Random Forest + Neural Network hybrid for classification. |
|
|
- Recommendation engine powered by a medical knowledge base. |
|
|
- **Outputs:** Structured analysis + initial evaluation + health suggestion. |
|
|
|
|
|
--- |
|
|
|
|
|
## Training Data |
|
|
- 120k anonymized medical records from licensed datasets. |
|
|
- Data cleaned under **HIPAA-compliant anonymization** standards. |
|
|
- Balanced dataset for accuracy across multiple lab test categories. |
|
|
|
|
|
--- |
|
|
|
|
|
## Evaluation Metrics |
|
|
| Metric | Score | Description | |
|
|
|---------|--------|-------------| |
|
|
| Accuracy | 91% | Overall model accuracy | |
|
|
| Precision | 89% | Relevance of predictions | |
|
|
| Recall | 93% | Detection of abnormal cases | |
|
|
| F1 Score | 0.91 | Global performance indicator | |
|
|
|
|
|
--- |
|
|
|
|
|
## Ethical Considerations |
|
|
- No storage of personal data after inference. |
|
|
- Designed for educational and clinical support only. |
|
|
- Reviewed for compliance with medical ethics standards. |
|
|
|
|
|
--- |
|
|
|
|
|
## License |
|
|
**LaboConsulte Proprietary License — Research & Clinical Support Use Only** |
|
|
© 2025 LaboConsulte / Aymen Messouci. |
|
|
All rights reserved. |
|
|
Unauthorized redistribution or commercial use is strictly prohibited. |
|
|
Contact: [ai@laboconsulte.com](mailto:ai@laboconsulte.com) |
|
|
|
|
|
--- |
|
|
|
|
|
## Tags |
|
|
`medical-ai` `healthcare` `arabic` `diagnosis-support` `laboconsulte` |
|
|
|
|
|
--- |
|
|
|
|
|
## Citation |
|
|
If you use or refer to this model, please cite as: |
|
|
|
|
|
> Messouci, A. (2025). *Nabdh AI Model – LaboConsulte Intelligent Health Assistant (v1.0).* |
|
|
> LaboConsulte AI Team, Algeria. |
