<p>Performance Comparison of Clinical Model, Clinical-Rad Model, and ViT.</p>

A 5.5 KB Excel file uploaded by Huiting Chen on April 3, 2026, compares the predictive performance of clinical, combined clinical-radiology, and Vision Transformer (ViT) models for thyroid cancer. The dataset includes metrics such as Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI) from internal and external validation cohorts. It focuses on papillary thyroid carcinoma.

Use Cases

• Benchmarking deep learning models against traditional clinical models based on performance metrics like NRI and IDI.
• Evaluating the added value of radiomics features in clinical prediction models for thyroid cancer.
• Comparing model generalizability using results from internal and external validation cohorts mentioned in the description.

Strengths

• Dataset is small (5.5 KB), facilitating quick download and inspection.
• Includes specific performance metrics like Net Reclassification Improvement and Integrated Discrimination Improvement.
• Explicitly compares multiple model types: clinical, clinical-radiology, and Vision Transformer.

Limitations

• Row count and column-level documentation are absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Huiting Chen via figshare

Collection Method

Likely results from a medical AI research study comparing model performance.

Time Range

null

Freshness

Last updated 2026-04-03 17:43:08; freshness should be verified.

Geography

null