Fig. 3: Confusion matrices for MADRS-BERT model.
From: Using a fine-tuned large language model for symptom-based depression evaluation
The confusion matrices illustrate the model performances and errors across the nine items using the fine-tuned MADRS-BERT model by comparing the predicted (x-axis) versus the actual (y-axis) MADRS scores. The intensity of the color represents the count of predictions, with darker shades indicating higher values. Diagonale entries represent correctly classified instances, while off-diagonal entries indicate errors.
