The secondary stresses that result from nonlinear and transient thermal gradients during the start-up and shut down of the large gas turbine engines drive low-cycle fatigue at specific locations of the outer casing. Typical service inspection of the outer casing is primarily based on finite element analysis estimates, considering various safety factors. However, as finite element analysis includes the worst possible combination of loading scenarios and operating conditions any engine may encounter in actual operation, this results in a conservative estimation of the service interval. Therefore, a generic preventive maintenance plan for the whole fleet often underutilises the casing capability and added cost. Hence, this paper proposes a data-driven nonlinear dynamic reduced-order model developed using the temperature data from low-cycle fatigue critical casing locations, ramp rates, and the percentage load of operation to predict the stresses. As a result, a reduced-order model can assess the damage for low-cycle fatigue critical locations in real-time using the operational data and propose an appropriate service intervention plan for each casing in a fleet.