Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# 🧠 Model Card — Nabdh AI Model
|
| 2 |
+
|
| 3 |
+
## Model Overview
|
| 4 |
+
**Name:** Nabdh (نَبض)
|
| 5 |
+
**Version:** 1.0
|
| 6 |
+
**Developer:** LaboConsulte AI Team — By Aymen Messouci
|
| 7 |
+
**Release Date:** October 2025
|
| 8 |
+
**Type:** Hybrid AI Model for Medical Data Analysis
|
| 9 |
+
**Languages:** Arabic, English
|
| 10 |
+
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
## Model Purpose
|
| 14 |
+
Nabdh is an AI model designed to assist doctors and users in understanding and interpreting **medical test results** and **clinical data**.
|
| 15 |
+
It provides personalized insights, early health pattern detection, and AI-powered recommendations.
|
| 16 |
+
|
| 17 |
+
---
|
| 18 |
+
|
| 19 |
+
## Intended Use
|
| 20 |
+
- **Doctors:** Support decision-making in case evaluations.
|
| 21 |
+
- **Patients:** Simplified health explanations.
|
| 22 |
+
- **Labs:** Integration into digital health systems.
|
| 23 |
+
|
| 24 |
+
⚠️ *Disclaimer:* Nabdh is **not a diagnostic tool**. It provides supportive insights only.
|
| 25 |
+
|
| 26 |
+
---
|
| 27 |
+
|
| 28 |
+
## Model Architecture
|
| 29 |
+
- **Inputs:** Laboratory results, textual descriptions, and clinical indicators.
|
| 30 |
+
- **Core Components:**
|
| 31 |
+
- NLP unit for understanding health descriptions.
|
| 32 |
+
- Random Forest + Neural Network hybrid for classification.
|
| 33 |
+
- Recommendation engine powered by a medical knowledge base.
|
| 34 |
+
- **Outputs:** Structured analysis + initial evaluation + health suggestion.
|
| 35 |
+
|
| 36 |
+
---
|
| 37 |
+
|
| 38 |
+
## Training Data
|
| 39 |
+
- 120k anonymized medical records from licensed datasets.
|
| 40 |
+
- Data cleaned under **HIPAA-compliant anonymization** standards.
|
| 41 |
+
- Balanced dataset for accuracy across multiple lab test categories.
|
| 42 |
+
|
| 43 |
+
---
|
| 44 |
+
|
| 45 |
+
## Evaluation Metrics
|
| 46 |
+
| Metric | Score | Description |
|
| 47 |
+
|---------|--------|-------------|
|
| 48 |
+
| Accuracy | 91% | Overall model accuracy |
|
| 49 |
+
| Precision | 89% | Relevance of predictions |
|
| 50 |
+
| Recall | 93% | Detection of abnormal cases |
|
| 51 |
+
| F1 Score | 0.91 | Global performance indicator |
|
| 52 |
+
|
| 53 |
+
---
|
| 54 |
+
|
| 55 |
+
## Ethical Considerations
|
| 56 |
+
- No storage of personal data after inference.
|
| 57 |
+
- Designed for educational and clinical support only.
|
| 58 |
+
- Reviewed for compliance with medical ethics standards.
|
| 59 |
+
|
| 60 |
+
---
|
| 61 |
+
|
| 62 |
+
## License
|
| 63 |
+
**LaboConsulte Proprietary License — Research & Clinical Support Use Only**
|
| 64 |
+
© 2025 LaboConsulte / Aymen Messouci.
|
| 65 |
+
All rights reserved.
|
| 66 |
+
Unauthorized redistribution or commercial use is strictly prohibited.
|
| 67 |
+
Contact: [ai@laboconsulte.com](mailto:ai@laboconsulte.com)
|
| 68 |
+
|
| 69 |
+
---
|
| 70 |
+
|
| 71 |
+
## Tags
|
| 72 |
+
`medical-ai` `healthcare` `arabic` `diagnosis-support` `laboconsulte`
|
| 73 |
+
|
| 74 |
+
---
|
| 75 |
+
|
| 76 |
+
## Citation
|
| 77 |
+
If you use or refer to this model, please cite as:
|
| 78 |
+
|
| 79 |
+
> Messouci, A. (2025). *Nabdh AI Model – LaboConsulte Intelligent Health Assistant (v1.0).*
|
| 80 |
+
> LaboConsulte AI Team, Algeria.
|