Flexible devices assembled with low-surface-energy PDMS substrates often face challenges, such as poor interfacial adhesion among multilayer films and mismatched mechanical moduli, complicating the development of stable and repeatable pressure sensors. Herein, a PDMS with internal dynamic cross-linking ability is synthesized to alleviate these issues, which shows good tensile properties, flexibility, and self-healing ability at room temperature. Taking advantage of the material homogeneity, the electrodes and sensing layer of the sensor made of the composite ink and PDMS, serving as the additive, have strong peeling resistance and interfacial adhesion. Furthermore, the multilayer sensing layer of the microconvex structures formed by pressing with a microstructural template effectively improves the sensitivity and sensing range of the multilayer pressure sensor. This multilayer flexible pressure sensor can be applied in medical health monitoring, effectively classifying symptoms related to sleep apnea-hypopnea syndrome through machine learning. The constructed neural network accurately learns the sleep conditions, including normal sleep, tachycardia, sleep apnea, sleep talking, snoring, and rapid eye movement. Utilizing the homogeneity and dynamic interfacial cross-linking of multilayer sensor materials, this design aims to improve the interfacial stability of flexible devices, expanding their application potential for long-term and reliable use.