The indirect association of subjective work performance with sleep quality and wellbeing among high-stress workers in Japan: a cross-sectional study.

Resistance training plays a central role in both athletic performance development and health-oriented exercise prescription (). In practice, a large proportion of programming decisions—such as load prescription, volume management, and fatigue monitoring—depends on an accurate determination of an individual's one-repetition maximum (1RM) (). Despite its widespread acceptance as the criterion measure of dynamic strength, direct 1RM testing is time-consuming, requires extensive supervision, and may impose fatigue or injury risk, particularly in team-based or youth athletic settings (, ). This concern is especially relevant in adolescent populations, where ongoing neuromuscular development and variable maximal strength expression—combined with the practical realities of school or team-based training environments—create a particular need for efficient assessment methods. Although direct 1RM testing has been shown to be feasible and safe in youth populations (), practical constraints in team-based settings and the additional value of understanding load-velocity relationships make indirect estimation methods particularly valuable in adolescent training contexts. Accordingly, substantial effort has been directed toward the development of indirect estimation methods that can provide valid 1RM predictions while minimizing the burden associated with maximal testing. One prominent approach in this regard is load-velocity profiling, which rests on the assumption of a systematic relationship between external load and barbell movement velocity during resistance exercise (, ). Based on established force-velocity and power-load relationships of skeletal muscle, it is typically assumed that increasing external load is accompanied by decreasing movement velocity, enabling extrapolation of maximal strength from submaximal velocities. Linear regression models derived from a small number of lifted loads are therefore frequently used to estimate 1RM without requiring an actual maximal lift (, , ). Although this approach is conceptually appealing and pragmatically attractive, reported validity indices are inconsistent, and prediction errors of practically relevant magnitude have been documented in several studies (, ). This illustrates that this assumption may oversimplify the shape of the load-velocity relationship, which could vary across individuals and potentially across different loading regions of the curve. A key limitation in the current literature lies not only in the magnitude of estimation error but also in the way it is evaluated. Many studies rely primarily on correlation-based indices to assess the relationship between predicted and measured 1RM (, ), despite well-established concerns that correlation does not quantify agreement or practical prediction accuracy (, ). In addition, validation protocols are often conducted under conditions in which the actual 1RM is already known, and prediction models are tested within the same population from which they were derived. This limits ecological validity, as in practical settings the true 1RM is unknown and predictions must generalize to individuals not included in model development (). Beyond these methodological considerations, little attention has been given to how deviations from linearity are distributed across the load spectrum and whether these deviations follow systematic patterns.