Hospital Readmission — Federated Random Forest Report

Each clinic trained a local Random Forest; feature importances were shared (not raw data).

Metrics

Metric	Value
N Samples Train	626
N Samples Val	157
N Features	132
Federated Rounds	5
Total Trees	100
Train Accuracy	0.9904
Val Accuracy	0.8854
Train Auc	1.0000
Val Auc	0.4639
Val Precision	0.0000
Val Recall	0.0000
Val F1	0.0000

Confusion Matrix

	Predicted: Not Readmitted	Predicted: Readmitted (<30d)
Actual: Not Readmitted	TN=139	FP=1
Actual: Readmitted	FN=17	TP=0

Classification Report

                 precision    recall  f1-score   support

 Not Readmitted       0.89      0.99      0.94       140
Readmitted <30d       0.00      0.00      0.00        17

       accuracy                           0.89       157
      macro avg       0.45      0.50      0.47       157
   weighted avg       0.79      0.89      0.84       157

Hospital Readmission — Federated Random Forest Report

Metrics

Confusion Matrix

Classification Report

Training History (per Round)

ROC Curve (Validation Set)

Feature Importance — Top 20