Background/Objectives: Evidence on caffeine (CAF), Rhodiola rosea, and their combined use in DanceSport remains limited. This randomized, blinded, placebo-controlled trial, retrospectively registered at ClinicalTrials.gov (NCT07690085), evaluated Jive-specific performance after a four-week standardized intervention. Methods: Forty-eight male collegiate DanceSport athletes were randomly allocated by drawing lots to placebo control (CTR), CAF, RHO (Rhodiola rosea), or RHO + CAF (n = 12/group). RHO was administered using Tongrentang Brand Rhodiola Capsule (0.4 g/capsule; three capsules twice daily; 2.4 g·day⁻¹) for four weeks, and CAF was administered at 3 mg·kg⁻¹ body mass 30 min before Post testing. The registry lists four primary outcomes: kick-extension angle, post-kick leg-retraction angle, trunk anteroposterior sway, and 30 s valid-kick success rate. Post values were analyzed by ANCOVA using the corresponding Pre value as covariate, with Holm correction across the primary-outcome family. Results: Training attendance was 100%. All four primary-outcome tests remained significant after Holm correction (adjusted p = 0.000768–0.006182; partial η² = 0.248–0.386). Adjusted success rates were 84.66% (CTR), 90.69% (CAF), 84.43% (RHO), and 92.31% (RHO + CAF); CAF and RHO + CAF were higher than CTR and RHO. The CAF × RHO interaction was not significant. Conclusions: Group differences were outcome-specific. The combined condition showed the highest adjusted success rate, but the data did not establish a synergistic interaction. The findings apply to the Rhodiola rosea intervention used in this study.

Background: Acute exercise rapidly reorganizes skeletal-muscle metabolism. The metabolite pools that remain responsive with aging, and their relationship to baseline oxidative state, remain incompletely defined. Methods: We analyzed 135 metabolites measured in 93 skeletal-muscle biopsies from 47 participants and matched baseline transcriptomic measurements. We tested paired post–pre changes, biochemical-class enrichment, baseline associations with a four-pathway oxidative transcript-state score, and participant-held-out prediction of acetylcarnitine response. False-discovery-rate control was applied within each analysis family. Bootstrap confidence intervals and a baseline-adjusted sensitivity analysis quantified predictive performance. Results: In exercise-trained older adults, 17 metabolites changed after acute exercise after FDR correction. Acetylcarnitine increased, whereas free carnitine, malate, ATP, and choline decreased. No exercise-trained-versus-normally active response contrast survived correction. The CoA–glutathione-annotated feature and NAD⁺ remained associated with the oxidative transcript-state score after adjustment for group and sex (partial $\rho =0.512$ and 0.514; both $q=0.042$). A four-metabolite TabPFN model predicted acetylcarnitine change with an out-of-fold $R^2$ of 0.412 (95% CI, 0.079–0.641), similar to the reference model ($R^2=0.399$, 95% CI, 0.054–0.625). Baseline acetylcarnitine alone achieved an AUC of 0.817 (95% CI, 0.680–0.941) for a twofold-response endpoint. Additional metabolites provided only modest incremental information after baseline adjustment. Conclusions: Acute exercise induced heterogeneous, distributed remodeling of skeletal-muscle metabolite pools. Baseline redox-related molecular state showed cross-omic coherence and contained information about subsequent acetylcarnitine response. Transcriptomic augmentation and more complex predictive architectures provided limited incremental benefit.