Challenges and opportunities of wearable molecular sensors in endocrinology and metabolism

Wearable technologies that analyse non-conventional biological matrices, such as interstitial fluid, sweat, tears or breath, have the potential to provide longitudinal biomarker data with minimal invasiveness. These data could provide insights into physiological and behavioural patterns, in particular outside medical care facilities. Despite the success of continuous glucose monitoring, the adoption of wearable sensors for managing endocrine and metabolic diseases remains limited. This Perspective highlights five key challenges and proposes solutions. First, understanding the physiology of longitudinal biomarker profiles is crucial for uncovering rhythmic patterns and physiological interrelations in the prediction of health trajectories. Second, technical barriers currently hinder the continuous monitoring of most clinically relevant biomarkers. Third, machine learning models often struggle with the complexity of dense biomarker datasets, which increases the risk of spurious correlations. Fourth, the diagnostic value of wearable sensor data requires validation through clinical studies, and predicting treatment outcomes necessitates diverse and large patient cohorts over extended observation periods in real-world settings. Finally, most wearable devices function as isolated solutions. Thus, they lack interoperability and integration into clinical pathways, and often fail to incorporate context and user input. Addressing these challenges will be key for advancing the role of wearable sensors in endocrine and metabolic care in future health-care settings.