Accurate real-time prediction of aerodynamic forces is crucial for the navigation of unmanned aerial vehicles (UAVs). This paper presents a data-driven aerodynamic force prediction model based on a small number of pressure sensors located on the surface of a UAV. The model is built on a linear term that can make a reasonably accurate prediction and a nonlinear correction for accuracy improvement. The linear term is based on a reduced basis reconstruction of surface pressure, with the basis extracted from simulation data and the basis coefficients determined by solving linear pressure reconstruction equations at a set of optimal sensor locations, which are obtained by using the discrete empirical interpolation method (DEIM). The nonlinear term is an artificial neural network that is trained to bridge the gap between the DEIM prediction and the ground truth, especially when only low-fidelity simulation data are available. The model is tested on numerical and experimental dynamic stall data of a two-dimensional NACA0015 airfoil and numerical simulation data of the dynamic stall of a three-dimensional drone. Numerical results demonstrate that the machine-learning-enhanced model is accurate, efficient, and robust, even for the NACA0015 case, in which the simulations do not agree well with the wind tunnel experiments.