Abstract: Landslide-debris flow disasters are characterized by sudden occurrence and complex movement processes. The inversion and prediction of their deformation processes represent key and challenging tasks in disaster prevention and mitigation efforts. To enhance the timeliness and accuracy of early disaster warning in the geologically complex Jishishan area, this study proposed a monitoring and prediction method that integrates the interferometric synthetic aperture radar (InSAR) with an improved genetic algorithm-optimized Elman network (IGA-Elman)-the InSAR-IGA-Elman method. First, the time-series InSAR technology was employed to process images from Sentinel-1. As a result, information on large-scale surface deformation across the study area was obtained. Subsequently, an IGA-Elman prediction model was built by using deformation monitoring data as core inputs while also incorporating multi-source influencing factors such as rainfall amount, slope, and elevation. Using the IGA, the model performed a global search for the optimal initial weights and thresholds of the Elman network, thereby overcoming conventional models' tendency toward local optimum. The experimental results indicate that the IGA-Elman model yielded high-accuracy predictions, with a root mean square error (RMSE) of 2.649 and a mean absolute error (MAE) of 1.098. All its performance metrics were significantly better than those of the standard Elman network and conventional models, such as the backpropagation neural network (BPNN) and support vector machine (SVM). The InSAR-IGA-Elman method effectively captured subtle deformation signals in the early stage of landslide-debris flow disasters. Moreover, its model performance increased with the number of training samples, demonstrating robust model stability. The results of this study provide strong technical support for the early warning, prevention, and control of geological hazards in the Jishishan and similar areas, holding significant potential and prospects of widespread application.