This discussion paper proposes a conceptual state-space framework for analyzing human operator performance in manufacturing environments. The framework systematically integrates sensor selection, process performance metrics, and operator well-being considerations, drawing on techniques such as blind source separation, component analysis, and dynamic latent state estimation. By distinguishing between invasive sensing modalities (e.g., wearable devices) and non-invasive alternatives (e.g., human performance models), the paper highlights the importance of balancing data richness against potential drawbacks—including reduced operator trust and increased cognitive load. In doing so, it underscores that carefully calibrating sensor invasiveness can preserve both productivity and well-being. While this approach is grounded in existing literature, no empirical validation is presented here. Instead, we advocate for future experimental studies or simulations, particularly in learning factory settings, to confirm the framework’s feasibility and further refine sensor selection strategies. Such practical evaluations will help realize a human-centric paradigm in manufacturing system design.