New ARISE method enhances feature selection for small-sample biomedical omics data

Researchers have developed ARISE, a novel ensemble method for feature selection in small-sample biomedical omics data. This adaptive framework integrates multiple relevance signals, assesses class-balanced stability, and controls for residual-informed redundancy, particularly for multiclass outcomes. In evaluations across five molecular datasets and various feature-set sizes, ARISE consistently outperformed existing methods, achieving top rankings in balanced accuracy, macro-F1, and Cohen's kappa. AI

IMPACT This research offers a more robust approach to feature selection in omics data, potentially improving diagnostic and predictive models in biomedicine.

RANK_REASON The item describes a new method published in an arXiv preprint, detailing its methodology and performance results. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• ARISE
• Cohen's kappa
• k-nearest neighbours
• random forest
• support vector machine

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →