Existing methods for determining the humidity adjustment factor and dry-wet cycle reduction factor of the resilient modulus of subgrade soil are mostly based on laboratory tests that consume a large amount of manpower and time, and the prediction accuracy is limited due to the constraints of specification value ranges. In order to achieve fast and accurate prediction of these two factors, the effects of stress state, moisture content, and the number of dry-wet cycles on the resilient modulus of subgrade soil were investigated through laboratory dynamic triaxial tests. Based on existing literature, physical parameters, state parameters, and stress parameters of subgrade soil were selected, and an artificial neural network prediction model optimized by a genetic algorithm was developed to enable rapid prediction of the humidity adjustment factor and dry-wet cycle reduction factor of subgrade soil. The results show that moisture content and dry-wet cycles have a significant influence on the resilient modulus of subgrade soil, while the humidity adjustment factor and dry-wet cycle reduction factor show stress dependence. The intelligent prediction model demonstrates high prediction accuracy for both the humidity adjustment factor and the dry-wet cycle reduction factor.