LLM-Lasso introduces a novel framework that uses large language models to guide feature selection in regression analysis, creating a bridge between domain expertise and data-driven modeling.

Key Innovations:

• Leverages LLMs to generate penalty factors for each feature in Lasso regression
• Incorporates domain-specific knowledge through retrieval-augmented generation (RAG)
• Creates a seamless integration of contextual insights with traditional statistical methods
• Demonstrates practical applications across multiple domains

Medical Impact: For healthcare and biomedical applications, LLM-Lasso enables more intelligent feature selection that aligns with medical domain knowledge, potentially improving diagnostic accuracy and treatment planning while maintaining statistical rigor.

LLM-Lasso: A Robust Framework for Domain-Informed Feature Selection and Regularization