Per-Seed Metrics for HCLmNet Lung Cancer Survival Model

Per-seed metrics evaluate the reproducibility of HCLmNet, a hybrid continual learning multimodal network for lung cancer survival prediction. The model integrates CT/PET imaging, genomic sequences, and clinical records, achieving a C-index of 0.84 and reducing mean absolute error to 140 days. The dataset is provided by MD Ilias Bappi and was last updated in March 2026.

Use Cases

• Compare C-index and Mean Absolute Error (MAE) metrics across different random seeds to assess model stability.
• Analyze the correlation between catastrophic forgetting scores and other per-seed performance metrics.
• Validate the reported 7.7% C-index improvement of HCLmNet over baseline models using the provided per-seed results.

Strengths

• Model performance is quantified with specific metrics: a C-index of 0.84 and an MAE of 140 days.
• Includes a measure of catastrophic forgetting, reported as 0.08, which is critical for continual learning evaluation.
• Released under a permissive CC BY 4.0 license, facilitating reuse and sharing.

Limitations

• The dataset is very small at 5.5 KB, indicating limited scope, likely containing only summary statistics rather than raw training data.
• No column names, sample data, or row count are provided, making the exact data structure and content unknown.

Provenance

Source

MD Ilias Bappi via figshare.

Collection Method

Derived from experimental results of the HCLmNet model on multimodal lung cancer datasets.

Freshness

Last updated on 2026-03-24.