Model Performance Metrics for a Melanoma Recognition Ensemble Model

Nine convolutional neural network models achieved AUC scores between 0.921 and 0.967 for distinguishing malignant melanoma from other skin lesions. The final XGBoost ensemble model, built by Jinyan Jiang, achieved an AUC of 0.988 on the test dataset. This 455.8 KB Excel file contains performance data for a model trained on the ISIC-2024 and HAM10000 datasets.

Use Cases

• Benchmarking deep learning models for skin lesion classification based on reported AUC, accuracy, precision, recall, and F1 scores.
• Evaluating the performance gain from ensemble methods like XGBoost based on the described model combination strategy.
• Reproducing or validating the study's methodology for clinical translation of AI-assisted melanoma diagnosis.

Strengths

• Model performance is quantitatively validated with high AUC scores, including a test AUC of 0.988 for the ensemble model.
• The dataset is derived from two established public dermatoscopic datasets, ISIC-2024 and HAM10000.
• The file is openly licensed under CC-BY-4.0 for reuse.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The 455.8 KB file size suggests a limited scope, likely containing summary performance metrics rather than raw image data.

Provenance

Source

figshare

Collection Method

Performance data from a study employing nine CNN models and an XGBoost ensemble.

Freshness

Last updated 2026-05-15 04:28:18; freshness should be verified.