<p>Disease monitoring typically requires the acquisition of multiple physiological signals with different modalities, yet existing epidermal electronics rely on separate sensors for each modality, increasing device footprint, bandwidth and power consumption. Here we report a wearable electronic system that fuses multimodal physiological signals into a single cross-modal biosignal (X-Sig). Leveraging a cross-layered device architecture and in-sensor signal fusion strategy, the X-Sig sensor concurrently acquires biopotential signals (electrocardiography and electromyography) and biomechanical signals (force myography and radial pulse) through a single channel. This approach enables continuous monitoring of haemodynamic parameters with high accuracy, including heart rate, pulse arrival time, and diastolic and systolic blood pressure. In machine-learning-based gesture recognition, the sensor substantially reduced the decoding error rate compared with conventional electromyography. By fusing complementary modalities at the sensor level, the X-Sig sensor provides a versatile platform for designing bandwidth-efficient and low-power wearable electronics.</p>