Upload BC dataset splits and detailed data card

.gitattributes

@@ -61,3 +61,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 test.csv filter=lfs diff=lfs merge=lfs -text
 train.csv filter=lfs diff=lfs merge=lfs -text
 validation.csv filter=lfs diff=lfs merge=lfs -text
+dataset.csv filter=lfs diff=lfs merge=lfs -text

README.md

@@ -2,28 +2,132 @@
 license: unknown
 task_categories:
 - tabular-classification
-pretty_name: Breast Cancer DMFS Dataset
+task_ids:
+- binary-classification
+tags:
+- healthcare
+- breast-cancer
+- gene-expression
+- uncertainty-quantification
+- clinical-prediction
+pretty_name: Breast Cancer DMFS Gene Expression Dataset
+size_categories:
+- n<1K
 ---
 
-# Breast Cancer DMFS Dataset
+# Breast Cancer DMFS Gene Expression Dataset
 
-Derived from Bioconductor genefu.
+## Dataset Summary
 
-## Task
-Binary classification of distant metastasis-free survival (DMFS).
+This dataset is derived from the Bioconductor `genefu` breast cancer gene expression resources and is formatted for binary classification of distant metastasis-free survival (DMFS) event occurrence.
 
-## Splits
+The dataset contains gene expression features paired with a binary clinical outcome label. It represents a small-sample, high-dimensional prediction setting, making it useful for evaluating predictive uncertainty and model robustness under limited-data conditions.
 
-| Split | N | Positives | Positive Rate |
-|---|---:|---:|---:|
-| Train | 278 | 75 | 0.2698 |
-| Validation | 60 | 16 | 0.2667 |
-| Test | 60 | 17 | 0.2833 |
+## Source Data
 
-## Preprocessing
-- Merged using `sample_id`
-- Stratified 70/15/15 split
-- Fixed random seed
+Original source:
 
-## Source
-https://www.bioconductor.org/packages//2.10/bioc/vignettes/genefu/inst/doc/genefu.pdf
+- Bioconductor `genefu` package
+- Documentation: https://www.bioconductor.org/packages//2.10/bioc/vignettes/genefu/inst/doc/genefu.pdf
+
+## Prediction Task
+
+The prediction task is binary classification.
+
+Target variable:
+
+- `dmfs_label`
+
+Label definition:
+
+- `1`: DMFS event occurred
+- `0`: No DMFS event occurred
+
+## Dataset Files
+
+This repository contains the following files:
+
+- `train.csv`: training split
+- `validation.csv`: validation split
+- `test.csv`: test split
+- `dataset.csv`: full merged dataset containing features and labels
+
+Each split file contains:
+
+- Gene expression features
+- `dmfs_label`, the binary outcome label
+
+## Dataset Statistics
+
+| Split | N | Negatives | Positives | Positive Rate |
+|---|---:|---:|---:|---:|
+| Train | 278 | 203 | 75 | 0.2698 |
+| Validation | 60 | 44 | 16 | 0.2667 |
+| Test | 60 | 43 | 17 | 0.2833 |
+| Total | 398 | 290 | 108 | 0.2714 |
+
+Additional dataset information:
+
+- Number of samples: 398
+- Number of input features: 22283
+- Feature type: continuous gene expression values
+- Target type: binary label
+
+## Data Processing Pipeline
+
+The released dataset was generated using the following preprocessing procedure:
+
+1. Loaded the cohort file containing `sample_id` and `dmfs_label`
+2. Loaded the feature matrix from `dataset.csv`
+3. Merged features and labels using `sample_id`
+4. Randomly shuffled the merged dataset using `random_state=123`
+5. Removed `sample_id` from the modeling features
+6. Split the data into train, validation, and test sets using a 70/15/15 split
+7. Stratified both split steps by `dmfs_label`
+8. Used `random_state=132` for reproducible train/validation/test splitting
+
+No feature selection, normalization, dimensionality reduction, or imputation was applied in this released version.
+
+## Intended Use
+
+This dataset is intended for research on:
+
+- Binary clinical prediction
+- Uncertainty quantification
+- Decision-aware model evaluation
+- Risk-coverage analysis
+- Small-sample, high-dimensional biomedical machine learning
+
+## Out-of-Scope Use
+
+This dataset should not be used for direct clinical decision-making, diagnosis, prognosis, or treatment selection without external validation and appropriate clinical oversight.
+
+## Limitations
+
+Important limitations include:
+
+- Small sample size
+- High-dimensional feature space
+- Potential risk of overfitting
+- Retrospective data source
+- Possible label noise in clinical outcome annotation
+- Limited generalizability without external validation
+
+## Ethical Considerations
+
+This dataset is derived from publicly available research data. It does not contain directly identifiable patient information in this release. However, because it is biomedical data, users should handle it responsibly and avoid using it for clinical deployment without further validation.
+
+## Reproducibility
+
+The train/validation/test splits were generated using fixed random seeds:
+
+- Shuffle seed: `123`
+- Split seed: `132`
+- Split ratio: 70% train, 15% validation, 15% test
+- Stratification variable: `dmfs_label`
+
+## Citation
+
+If you use this dataset, please cite the original Bioconductor `genefu` resource and the associated source publications.
+
+You may also cite this dataset repository if used as part of an uncertainty quantification benchmark.

dataset.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b17e9a2fc5e4907fb330528e13e3d2ff6def40bac90c7c62545aafd88965bb2
+size 149964982