Multimodal data from 108 patients, including CT imaging, dose distribution, and clinical features, used to predict radiation-induced oral mucositis. The dataset was created by Ling Li and published on figshare in April 2026. It contains radiomic features extracted for a comparative study of machine learning and deep learning model performance.
Use Cases
- Compare traditional ML vs. deep learning performance for multi-class prediction based on radiomic and dosimetric features.
- Train ensemble models like Extra Trees for toxicity prediction based on the multimodal clinical data.
- Investigate overfitting and mode collapse in high-dimensional architectures like 3D-CNNs on limited medical datasets.
- Build lightweight 1D-CNN models using fused low-dimensional features from CT and dose data.
Strengths
- Dataset size is 132.7 KB, indicating a compact and focused feature set.
- Based on data from 108 patients, providing a defined cohort for small-sample studies.
- Model performance was evaluated using a stratified 5-fold cross-validation, suggesting methodological rigor.
- Published under a CC-BY-4.0 license, allowing for open reuse and modification.
Limitations
- Row count is unknown, which may limit suitability assessment for specific modeling tasks.
- Column-level documentation is absent; field semantics must be inferred after download.
- The dataset is very small (132.7 KB), indicating a limited scope of extracted features or a highly summarized dataset.
Provenance
- Source
- figshare, author Ling Li.
- Collection Method
- Multimodal data collected from 108 patients, likely from a clinical study.
- Time Range
- null
- Freshness
- Last updated 2026-04-09 17:28:56; freshness should be verified.
- Geography
- null