Self-Prognostic Electronic Skin for Reliability-Aware Wearable Sensing.

Flexible wearable sensors are increasingly used for long-duration monitoring of body-state information, but they can degrade silently during operation while continuing to output physiologically plausible signals, making it difficult to determine whether an incoming measurement remains trustworthy. Here, we report a self-prognostic electronic skin platform that couples multimodal body-state sensing with autonomous sensor-health assessment. The device monolithically co-integrates piezoelectric and triboelectric transduction by origami folding into an ultrathin, six-channel wrist-mounted array (∼0.1 mm, <1 g), so that a single mechanoelectrical stream simultaneously encodes cardiovascular pulsation, wrist activity, and the device's own degradation state. A two-stage self-diagnostic framework uses an autoencoder-CUSUM pipeline to establish device-specific failure labels from degradation trajectories, followed by an LSTM-Transformer model for remaining useful lifetime prediction. Across a twenty-device cohort with a nearly sevenfold variation in lifetime, the model predicts remaining useful lifetime across the cohort's degradation trajectories with R2 = 0.95 and captures uncertainty that increases as degradation becomes more severe. In single-participant daily-use proof-of-concept testing, the system estimates physiological parameters and sensor remaining lifetime from the same stream. The predicted lifetime of a newly fabricated device remained near 100% during a representative 1.5 h segment and declined modestly to 97.2% over approximately 12 h of wear distributed across multiple days; this single-device demonstration illustrates the operation of the diagnostic channel during daily use rather than validated cross-device predictive accuracy. These results establish reliability-aware wearable sensing, in which each body-state measurement is interpreted together with the operational health of the sensor itself, with potential value for long-duration applications such as health monitoring and embodied intelligence.