The subjects were eight members of the Chinese National Men's Curling Training Team preparing for the Winter Olympic Games, with an average age of 26.8 years (22–31 years), an average height of 181.1 ± 10.7 years, an average body weight of 77.2 ± 5.5 years, and an average number of years of training of 8.6 ± 2.4 years (Table 1). All athletes completed an informed consent form. The collection period was 211 consecutive days.

The study implemented three load-monitoring modalities: external load metrics, physiological/biochemical markers, and Omegawave state indices. Omegawave and external load data were collected daily, while physiological/biochemical parameters were assessed at 15-day intervals, followed by time-synchronized analyses. This 15-day period constitutes a mesocycle within the preparation phase, comprising three 5-day microcycles. Each microcycle featured four training days followed by a rest day, as designed by Head Coach Lindholm Peja.

The study quantified training load using three methods: external load measurements, physiological/biochemical markers, and Omegawave state indices. Athletes provided RPE via the Borg CR-10 scale 15–30 min post-training during field and physical sessions. Injured athletes without training were assigned 0 A.U. We also calculated training monotony, training strain, short-term (5-day) and long-term (20-day) loading, and the short-term:long-term load ratio (using a sliding-window average). Training duration (recorded to the nearest minute) was defined as the period from the start to end of formal training. The “start” denoted when athletes began coach-prescribed training after standardized warm-ups on the field. The “end” occurred when athletes completed prescribed tasks and exited the main training area, excluding post-session stretching/relaxation.

This study categorizes draws (curling stone throws) as either training draws or competition draws. Training draws encompass all stones thrown during practice, including coach-prescribed throws and athlete-initiated additional throws. Competition draws include those made during intra-squad scrimmages, simulated matches (where coaches directly set scenarios to mimic international opponents), and official matches. Drawing on training and match duration, we derived a secondary metric: curling density (draws per unit time). Higher density (more throws in less time) indicates reduced decision-making time and lower cognitive effort per throw, while lower density (fewer throws over longer duration) reflects greater time for tactical deliberation and higher cognitive effort. Thus, draw density serves as a proxy for the ratio of cognitive to physical effort. Specific draw types (e.g., guard, takeout) were not statistically analyzed, as their occurrence is heavily influenced by dynamic game tactics and strategy, limiting meaningful interpretation.

The Omegawave Athletic State Evaluation System, widely used in training practice for assessing athletes' immediate readiness (52), was employed to evaluate subjects 15–30 min after their final daily training session. This system qualitatively assessed central nervous system (CNS) status and cardiac function through simultaneous electroencephalogram (EEG) and electrocardiogram (ECG) analysis, including cardiac bioelectrical current activation levels. Primary evaluation indices comprised: Cardiopulmonary Regulation Functional State (1–7 points) and CNS Readiness State (1–7 points).

Physiological and biochemical markers were collected every 15 days under standardized protocols from the National Winter Sports Center of China to evaluate athlete fatigue. Testing occurred at 06:30 on the final rest day of each cycle, with athletes in a fasted state. Four indicators assessed physiological response to training load: blood urea, creatine kinase, testosterone, and cortisol. Testosterone was measured using Chemiluminescent Microparticle Immunoassay (CMIA) on an ARCHITECT i1000sr automated immunoassay analyzer (ARCHITECT i1000sr, Abbott Laboratories Co., Ltd, USA). Cortisol was determined by Enzyme-Linked Immunosorbent Assay (ELISA). Creatine Kinase (CK) activity was analyzed via the continuous monitoring (enzymatic kinetic) method using an OLYMPUS AU2700 analyzer (OLYMPUS AU2700, Olympus Corporation Co., Ltd, Japan). Hematocrit was assessed using the impedance method on an HT-ESR24 dynamic hematocrit analyzer (HT-ESR24, Zibo Hengtuo Analytical Instruments Co., Ltd, China). All analyses were performed by experienced technicians according to standardized protocols.

During each training session, the number and type of stone deliveries were recorded for each athlete. Within 15–30 min post-training, session-RPE (sRPE) was collected. Athletes then performed Omegawave state assessments in isolated, quiet environments. On Day 15 of each cycle at 06:30 AM, fasting athletes provided samples for Testosterone, Cortisol, Creatine Kinase (CK), and Hematocrit assessment (Figure 1).

Data were processed using SPSS (version 26; SPSS, IBM Corporation, Armonk, New York, USA), WPS (version 2023; Kingsoft Office Software Co., Ltd., Beijing, China), and GraphPad Prism (version 9.5.1; GraphPad Software, Inc., San Diego, USA). Results are expressed as mean ± SD. All datasets met normality assumptions. Pearson correlation analyzed relationships between sRPE and: a. external load metrics, b. physiological/biochemical markers, and c. Omegawave Athletic Status indices. Spearman correlation was used for non-normally distributed data. Normality was assessed via Shapiro–Wilk test with Q–Q plots (n < 2,000) or Kolmogorov–Smirnov with Q–Q plots (n ≥ 2,000). Inter-metric consistency was evaluated using intraclass correlation coefficients (ICC) with Bland-Altman plots on standardized data (95% CI). ICC interpretation followed established thresholds: <0.20: Very poor; 0.21–0.40: Weak; 0.41–0.60: Moderate; 0.61–0.80: Substantial; 0.80: Excellent. Missing values were coded as 0. While this approach may introduce bias, the large sample size mitigates its impact. The computational formulas for sRPE, ACWR, Monotony, and Training Pressure addressed in this study (53) were:Workload=RPE×trainingduration(min),A.U.(ArbitraryUnits)ACWR=Workload1weekWorkloadAverage4weeksMonotony=WorkloadlastweekWeeklyWorkloadSDTrainingPressure=Workloadweek×Monotony

This is the descriptive statistics of all data in this study (Table 2).

sRPE load metrics demonstrate strong correlation and consistency with most external load indicators, as evidenced by the high covariance with standardized training loads and total draws in Figure 10, affirming their utility for tracking external load variations. In curling, where physical exertion patterns remain relatively consistent across techniques and intensity primarily derives from ice sweeping and tactical cognition, the extended recovery periods during prolonged training/competition dilute acute physiological strain. Competition loads exhibit particular complexity due to: 1. strategic demands creating variable physical expenditure, 2. opponent strength disparities (intra-squad to international matches) causing mental exertion fluctuations—where superior opponents elevate sRPE through psychological stress while inferior opponents depress it through reduced engagement, and 3. the consistent phenomenon of lower draw volumes but higher sRPE values in matches vs. training, attributable to both heightened cognitive load and increased sweeping intensity from competitive mentality. Consequently, neither draw counts nor session duration—even in this cognition-dominated sport—adequately capture athletes' psychophysiological exertion, evidenced by significant intensity differences between equally timed training and competition. Furthermore, comparing sRPE against external loads reveals load sensitivity: divergent sRPE responses to statistically similar external loads may indicate high sensitivity (suggesting fatigue onset) or low sensitivity (indicating training adaptation), providing actionable biomarkers for athletic status that warrant further validation.

Cortisol and blood urea exhibited weak correlations with sRPE-quantified load, representing the only significant biochemical relationships. Consistency analyses revealed moderate agreement between both short- and long-term loads with blood urea, while ACWR showed moderate consistency with cortisol. Critically, cortisol demonstrated a negative correlation with ACWR, indicating that increased training volatility reduces cortisol concentration within physiological ranges. This suggests optimal load fluctuation mitigates chronic fatigue accumulation. Conversely, elevated long-term load correlated with increased blood urea, signifying physiological fatigue from excessive loading. These patterns collectively establish sRPE as a viable proxy for biochemical markers in load monitoring.

The Omegawave Athletic State evaluation and sRPE both assess athlete load states yet differ fundamentally (Figure 11). Omegawave precisely measures current physiological status but cannot isolate daily training load impact, as residual fatigue from prior sessions may elevate readings even during rest days. This temporal insensitivity limits its accuracy for single-session evaluation. While Omegawave testing surpasses biochemical markers in convenience, it remains more time-intensive than sRPE collection. Coaches requiring rapid daily load assessment should prioritize sRPE, whereas Omegawave better characterizes underlying physiological mechanisms. Crucially, sRPE reflects the organism's response to applied external load and effectively quantifies training impact when athletes reliably report subjective fatigue, though its subjectivity raises reliability concerns. External load metrics offer objective quantification with similar operational efficiency but fail to capture internal physiological strain or psychological exertion. Physiological biomarkers provide superior quantification precision yet require invasive procedures, strict collection protocols, and retrospective analysis—offering high-validity moment-state evaluation without predictive capacity. Omegawave's distinctive advantage lies in identifying directional load effects to guide training adaptations, though equipment dependency constrains practical implementation.

In curling, front-end positions (S: first/second basemen) primarily execute ice sweeping while back-end positions (V: third/fourth basemen) direct tactical decision-making. Analysis of pre-competition cycle loads (Figure 12) revealed minimal differentiation between positions across external (draw counts) and internal (sRPE-derived) load metrics. Notably, significant training monotony divergence emerged, attributable to distinct load sources: S positions experienced predominantly physiological stress from sweeping, whereas V positions incurred cognitive demands influenced by shot difficulty and opponent tactics, thereby generating position-specific monotonicity profiles.

sRPE provides superior assessment of the organism's response to external load when athletes accurately self-report, offering reasonable evaluation of training impact despite reliability concerns stemming from its inherent subjectivity. External load metrics deliver objective quantification with comparable operational efficiency but fail to capture internal physiological strain or psychological exertion. Physiological biomarkers afford greater quantification precision yet incur substantial time costs, require stringent collection protocols that limit utility to retrospective analysis, involve invasive procedures, and lack predictive capacity despite high validity for momentary state assessment. The Omegawave system's principal advantage lies in identifying directional load effects to inform training adaptations, though practical implementation faces portability constraints.

Different training load quantification methods exhibit distinct characteristics, ranging from exceptionally precise quantitative tools (accurate to 2–3 decimal places) to qualitatively analytical approaches. When selecting appropriate quantification methods, coaches and researchers must consider not only sport-specific requirements, athlete proficiency levels, and training phase demands, but also required precision thresholds. In practical training contexts, single-decimal accuracy typically suffices for fatigue assessment relevance, making precision needs a critical selection criterion. Given varying theoretical and applied values across quantification tools, deliberate evaluation of methodological alignment with both research objectives and practical utility remains essential.

sRPE constitutes a reliable indicator for evaluating training load in curling programs. The sRPE-workload assessment process demonstrates operational convenience, facilitates phased evaluation of training loads, and serves as an effective tool for coaches and multidisciplinary support teams to implement load management strategies. The sRPE evaluation methodology exhibits inherent limitations, including the absence of categorical differentiation of load magnitudes, susceptibility to subjective influences, and potential cumulative overestimation of training loads when applied across extended temporal frameworks. Prudent selection of heterogeneous load assessment methodologies should be predicated on the practical demands of training programs.