Abstract: This paper addresses the challenges posed by conventional monitoring techniques in achieving continuous and comprehensive deformation monitoring of dump slopes under high-altitude permafrost conditions, as well as the difficulty in accurately predicting deformation trends. Focusing on Pit No. 2 in Jiangcang mining area, Muli Coalfield in Qinghai Province, this paper integrates SBAS-InSAR and Long Short-Term Memory (LSTM) neural networks using Sentinel-1 A imagery to obtain deformation results for the northern and southern dump slopes of Pit No. 2 in Jiangcang mining area, Muli Coalfield from 2019 to 2022. The paper investigates the deformation characteristics of these dump slopes, the impact of precipitation on deformation trends, and develops an LSTM-based prediction model using time-series deformation data. The results show that the annual average deformation rates of the northern and southern dump slopes of Pit No. 2 in the Jiangcang mining area of the Muli Coalfield in Qinghai Province range from-62 mm·a^-1 to 21 mm·a^-1, with maximum cumulative settlements of 255 mm and 214.5 mm, respectively. A comparative analysis with daily rainfall data reveals that precipitation accelerates the deformation process, with the dump slopes exhibiting an accelerated settlement trend following rainfall events. The LSTM model predictions demonstrate a maximum error of 2.42 mm, with coefficients of determination (R²) exceeding 0.9. Additionally, the model achieves a maximum root mean square error (RMSE) of 1.14 mm and a mean absolute error (MAE) of 0.97 mm, indicating the reliability and feasibility of using SBAS-InSAR technology combined with the LSTM model for deformation monitoring and prediction in high-altitude mining areas. This integrated approach provides valuable insights for the stability assessment and disaster prevention of dump slopes in permafrost mining regions and similar engineering projects.