Synthetic BIRADS Mammography Data For Rare Categories

Synthetic data generated by Claude Opus 4.6 specifically for rare BIRADS categories 5 and 6. The dataset is designed to augment training for machine learning models focusing on high-risk breast cancer classifications. Its creation method and specific volume are not detailed.

Use Cases

• Train classification models to distinguish between BIRADS 5 and 6 categories using synthetic feature vectors.
• Augment imbalanced training sets for breast cancer risk prediction by adding synthetic samples for rare high-risk classes.
• Benchmark generative models for medical data by comparing synthetic BIRADS feature distributions against real clinical data.
• Develop anomaly detection systems for rare, high-risk mammography findings represented by BIRADS 5 and 6 labels.

Strengths

• Focuses specifically on rare BIRADS categories 5 and 6, which are often underrepresented in real datasets.
• Created using a state-of-the-art AI model (Claude Opus 4.6), suggesting a structured generation approach.

Limitations

• The dataset is entirely synthetic, so its utility depends on how well it captures the statistical properties of real patient data.
• Key metrics like row count, column count, and feature names are unknown, making assessment of scale and applicability difficult.
• Lacks provenance details such as the real data source used for inspiration or validation of the synthetic generation process.

Provenance

Source

Synthetic data generated by Claude Opus 4.6.

Collection Method

AI-generated synthetic data created to model rare BIRADS classes.