# Notebooks to recreate figures

**Figures:**\
    - Figure 1. Overview\
    - [Figure 2A. Comparison of predictions & cluster numbers](/notebooks/plot_examples_prediction.ipynb)\
    - [Figure 2B. Comparison cluster numbers](/notebooks/compare_results_MLP.ipynb)\
    - [Figure 3. Examples of interpretable patient reports](/notebooks/plot_examples_reports.ipynb)

**Tables:**\
    - [Table 1. results MELD graph model vs MLP model](/notebooks/compare_results_MLP.ipynb)\
    - [Table 2. breakdown performances](/notebooks/compare_results_MLP.ipynb)

**eTables:**\
    - [eTable 1. Demographics](/notebooks/demographics_table.ipynb)\
    - [eTable 2. Reduction in false positive clusters](/notebooks/compare_results_MLP.ipynb)\
    - [eTable 3. MELD Graph model with harmonized vs non-harmonized MRI features](/notebooks/compare_results_MLP.ipynb)

**eFigures:**\
    - eFigure 1. MELD Graph model architecture\
    - eFigure 2. Pipeline for running a new patient’s MRI scan through MELD Graph\
    - [eFigure 3. Stability of NeuroCombat](/notebooks/combat_subsampling.ipynb)\
    - [eFigure 4. Characterisation of predictions](/notebooks/analysis_predictions.ipynb)\