Screening and Correlation Analysis of "Material Units" in Houttuynia cordata Volatile Oil Based on Supramolecular "Imprinting Template" Theory Combined with Factor Rotation

• • The matching frequency method compressed 58 GC-MS fingerprints into 34 structural "material units" (A1–A34), reducing average peak count and information entropy at P < 0.01 while preserving total zero-, first-, and second-order moments and information content. This dimensional reduction is industrially critical because it enables batch-to-batch quality control without discarding the statistical moments that encode the overall chemical profile. • • Factor rotation extracted eight common factors, with the top four accounting for the highest variance and enriched in phytol, β-elemene, D-limonene, α-pinene, and caryophyllene. These compounds have documented anti-lung-cancer mechanisms, including PI3K/Akt pathway inhibition and mitochondrial apoptosis induction, providing a mechanistic rationale for prioritizing these factor groups in raw material sourcing. • • Comprehensive quality scores negatively correlated with A549 IC50 values (r = −0.739, P < 0.01) across a 73.6–269.9 nL/mL range. The correlation is strong enough for trend-level screening but not for single-batch potency prediction, as the highest-scoring sample S16 did not exhibit the absolute lowest IC50, exposing a gap between chemical scoring and biological endpoint. • • High-contribution units A34, A31, and A7 were flagged as a potential pharmacodynamic component group, yet they remain statistically derived candidates without isolation, enrichment, or independent in vivo validation. Industrial translation requires preparative-scale fractionation and multi-cell-line testing before these markers can be adopted as pharmacopoeial quality indicators.