Ultrasonic assisted high-speed machining (UAHSM) can be served as a thermomechanical surface sever plastic deformation (SSPD), because of the high-frequency impact load exerting to the sample together with thermomechanical loads due to shearing and plowing. Despite existing of few works which studied the impact of ultrasonic vibration on fatigue life assessment of difficult-to-cut material by experimental approach, they couldn’t provide an in-depth analysis to identify the underlying mechanisms of fatigue due time-consuming and costly fatigue life tests. Hence, elucidating the role of ultrasonic vibration in UAHSM on variation of fatigue life needs further studies. In order to do so, in the present work, a hybrid predictive approach based using ANFIS-based machine learning model and micromechanical Navaro-Rios (NR) fatigue crack propagation model has been introduced to directly correlates the UAHSM’s parameters to fatigue life. Here the former correlates feed rate, cutting velocity and vibration amplitude as process inputs, to surface integrity aspects (SIA) viz residual stress, roughness and grain size as output. Then, the modeled SIA are correlated to fatigue life using the former. The introduced hybrid model was then verified through series of UAHSM by examining the fatigue lives of milled Inconel 718 using four-point bending fatigue tests. Upon confirmation of the developed model, a comprehensive study was carried out to find how the process factors impact variation of SIA and subsequently fatigue. It was found from the results of developed models and confirmatory experiments that the role of ultrasonic vibration on improved fatigue life is mainly due to inducing compressive residual stress and more refined microstructure than the roughness.