Add week2_summary.md to project

Author: AAdewunmi

notebooks/docs/week2_summary.md

@@ -0,0 +1,53 @@
+# Week 2 Summary — PneumoDetect Lab Series
+
+**Theme:** Model Refinement, Explainability, and Performance Review  
+**Duration:** Week 2 (Days 1–5)  
+**Project:** [AI-Assisted-Pneumonia-Detection-Project](https://github.com/AAdewunmi/AI-Assisted-Pneumonia-Detection-Project)
+
+---
+
+## Highlights
+
+| Day | Focus | Core Deliverables |
+|-----|--------|------------------|
+| W2-D1 | Model Training (Baseline) | `resnet50_baseline.pt`, initial metrics |
+| W2-D2 | Data Balancing & Weighted Sampling | Smoothed sample weights + balanced loader |
+| W2-D3 | Grad-CAM Integration | Implemented explainability via `src/gradcam.py` |
+| W2-D4 | Grad-CAM Refinement & Testing | Added `generate_cam()` + passing unit/integration tests |
+| W2-D5 | Performance Review & Clinical Report | `performance_report_v1.md`, Grad-CAM montages, ROC/PR curves |
+
+---
+
+## Model Outcomes
+
+| Variant | Accuracy | Loss | Notes |
+|----------|-----------|------|-------|
+| Baseline | 0.7518 | 0.4999 | Solid starting point |
+| Balanced | 0.8115 | 0.4128 | Improved recall on minority class |
+| Fine-Tuned | 0.9985 | 0.0049 | Excellent generalization and focus |
+
+---
+
+## Explainability Insights
+- Grad-CAM heatmaps now align with clinical regions of interest.  
+- Overlays exported to `/reports/week2_gradcam_refinement/` and `/reports/week2_performance_review/`.  
+- Fine-tuned ResNet-50 produced sharper, diagnostically meaningful activations.
+
+---
+
+## Key Learning
+- Balancing datasets significantly improves model fairness.
+- Explainability pipelines can be systematically tested and versioned.
+- Structured documentation (via performance reports) bridges ML outputs and clinical communication.
+
+---
+
+## Next Steps (Week 3)
+- Integrate Grad-CAM visualization and metrics into the **Flask-based clinician dashboard**.  
+- Develop an interactive report viewer for ROC, PR, and Grad-CAM images.  
+- Begin drafting **Clinical Report v2** and web integration prototype.
+
+---
+
+**Maintained by:** Adrian Adewunmi  
+**Repository:** [AI-Assisted-Pneumonia-Detection-Project](https://github.com/AAdewunmi/AI-Assisted-Pneumonia-Detection-Project)