Data Sheet 5_Integrating CT radiomics and transcriptomics: a biologically-informed machine

161 advanced laryngeal cancer patient records with 1,321 extracted CT radiomics features and clinical predictors for chemotherapy response. Xin Xiong published this dataset on figshare in April 2026, detailing the development of a Random Forest model achieving an AUC of 0.810 on an external test set. The data integrates radiomics scores, transcriptomic analysis, and clinical variables to enable non-invasive prediction of treatment outcomes.

Use Cases

• Training predictive models for chemotherapy response based on radiomics scores and clinical features like gap invasion.
• Validating radiomics feature selection methods using the 1,321 extracted features and the final four robust features.
• Conducting SHAP analysis for model interpretability based on core predictors such as Rad-score and fat space invasion.
• Benchmarking machine learning algorithms (e.g., Random Forest, LASSO regression) on a clinical oncology dataset.
• Studying the biological basis of radiomics features through associated transcriptomic analysis.

Strengths

• Includes 1,321 radiomics features extracted from pre-treatment CT scans of 161 patients.
• Model performance is documented with specific metrics: Random Forest achieved an AUC of 0.914 (training), 0.856 (validation), and 0.810 (external test).
• Dataset supports interpretable AI with SHAP analysis identifying Rad-score and fat space invasion as core predictors.
• Published under a permissive CC-BY-4.0 license, facilitating reuse and sharing.

Limitations

• Row count and specific column names are unknown, limiting suitability assessment for direct analysis.
• Data is retrospective and from a single study of 161 patients; generalizability may be limited.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

figshare, author Xin Xiong.

Collection Method

Retrospective analysis of patient records and extraction of radiomics features from CT images.

Time Range

null

Freshness

Last updated 2026-04-15 04:36:55.

Geography

null