Higher dietary carotenoid intake is associated with healthier aging profiles, but whether this association varies across participant characteristics remains unclear.

To explore heterogeneity in the adjusted association between dietary carotenoid intake and phenotypic age acceleration and to prioritize participant characteristics for independent validation.

Cross-sectional analysis.

The 2011–2014 National Health and Nutrition Examination Survey (NHANES).

A total of 5011 adults in the analytic NHANES sample with available dietary, biomarker, accelerometer-derived behavioral, and covariate data.

Not applicable.

Reported total dietary carotenoid intake was averaged across two 24-hour recalls, and phenotypic age acceleration (PhenoAgeAccel) was calculated from clinical biomarkers. Double machine learning (DML) estimated participant-specific model-adjusted association contrasts (not directly observed individual responses) using a covariate-dependent mapped dietary exposure scale and screened 26 candidate characteristics, including accelerometer-derived behavioral metrics.

In exploratory analyses based on the DML framework, interdaily stability (IS), an index of day-to-day regularity of the rest–activity rhythm, was the highest-ranked candidate characteristic. Residualized association estimates were more negative in the lowest than the highest IS quartile (−0.343 versus 0.026 years per SD of mapped dietary exposure). In five internal 50:50 split-sample analyses, IS ranked among the top five candidates in every discovery sample. The lower-versus-higher IS pattern was directionally consistent in the corresponding holdout samples, but estimates were imprecise and did not establish between-group differences.

This exploratory analysis prioritizes IS for independent validation as a candidate behavioral context for potential heterogeneity in the dietary carotenoid–PhenoAgeAccel association. The findings generate a hypothesis for studies integrating dietary exposure, behavioral regularity, and aging-related biomarkers.