This research demonstrates how Large Language Models can transform genetic data analysis by selecting and engineering the most relevant features from complex genotype data.

• Uses LLMs to identify biologically significant features from high-dimensional genetic data
• Achieves more accurate phenotype predictions with smaller, interpretable feature sets
• Outperforms traditional data-driven feature selection methods in hereditary hearing loss prediction
• Successfully captures genetic ancestry information without explicit training

This approach represents a significant advance for precision medicine, enabling more targeted genetic testing and personalized treatment planning with improved interpretability of genetic markers.

Knowledge-Driven Feature Selection and Engineering for Genotype Data with Large Language Models