This paper represents an integrated prognostics method dedicated to the wind turbine high-speed shaft bearing prognosis, which integrates physical degradation models and data driven approaches. In bearing failure prognostics, the excessive shaft vibration eventually leads to the system failure. In this case (crack growth prognostics), the measured data (crack size) is the same as a model prediction from Paris’s law. Indeed, we introduce an integrated prognostic approach based on usage model through Paris’s law and the use of a Kalman smoother to estimate the remaining useful life offering a solution to the inherent phase delay cancellation from Kalman filtering, providing a more accurate and smoother estimate with confidence bounds.The proposed method is validated on a real high-speed shaft bearing wind turbine generator. The used database contains one raw acquisition per day over 50 days of measurement at a high sample rate, 6 s each. The results show that the Kalman smoother is an effective way to improve trending and remaining useful life estimation.