Model Performance Metrics for a Melanoma Recognition Ensemble Model

Nine convolutional neural network models achieved AUCs between 0.921 and 0.967 for distinguishing malignant melanoma from other malignant skin lesions. An XGBoost ensemble model built on their outputs achieved an AUC of 0.988 on a test dataset, as documented in an Excel file by Jinyan Jiang last updated in May 2026. The model was trained and validated using the ISIC-2024 and HAM10000 dermatoscopic datasets.

Use Cases

• Benchmarking deep learning model performance for melanoma classification based on reported accuracy, precision, recall, F1 Score, and AUC metrics.
• Evaluating the effectiveness of ensemble learning strategies in medical image analysis based on the described XGBoost integration.
• Comparing the diagnostic performance of nine specific CNN architectures (e.g., DenseNet201, EfficientNetB7) for dermatoscopic lesion recognition.
• Developing clinical decision support systems based on the described model combining deep learning and ensemble learning.

Strengths

• Model performance is rigorously validated with multiple metrics, including AUCs up to 0.988 on the test set.
• The methodology is clearly described, involving nine distinct CNN models and an XGBoost ensemble.
• The dataset is openly licensed under CC-BY-4.0, facilitating reuse and verification.

Limitations

• The dataset is very small at 9.3 KB, suggesting it likely contains only summary performance metrics, not the underlying image data or model weights.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for certain analyses.

Provenance

Source

Jinyan Jiang

Collection Method

Model performance results derived from training on the ISIC-2024 and HAM10000 dermatoscopic datasets.

Freshness

Last updated 2026-05-15 04:28:16; freshness should be verified.