Silicon-based piezoresistive pressure sensors suffer from insufficient reliability and reduced service life due to issues such as output drift and sensitivity degradation in harsh environments. This study aims to systematically elucidate the physical mechanisms behind their stability degradation and to develop a high-precision life prediction model. Utilizing the physics of failure analysis theory, the research employed variable-amplitude cyclic loading and accelerated fatigue testing. Accelerated tests were conducted by applying alternating pressure with different amplitudes. A dataset of sensor failure degradation was established through microscopic examination and performance monitoring. This approach overcame the challenge of analyzing the coupled effects of multiple mechanisms, including diaphragm cracking, piezoresistor creep, and packaging stress failure, ultimately enabling the construction of a life prediction model under uniaxial pressure loading conditions. Accelerated life testing demonstrated that under a pressure load of 140% of the full-scale range, the sensor's linearity increased by over 50% after approximately 2.2 million cycles, which was defined as failure. The developed model achieved an error of less than 15% between the predicted and measured lifespan, enabling effective prediction of the sensor's failure cycle. The fatigue experiments conducted and the life prediction model developed in this study effectively meet the engineering requirements for reliability assessment and life extension of pressure sensors. This work holds significant theoretical and practical application value, providing crucial support for advancing the design optimization and lifetime prediction of highly reliable silicon-based pressure sensors.