Desertification is a major environmental challenge in Iran, particularly in arid and semi-arid regions such as Golestan Province, where climate change and human pressures intensify land degradation. Therefore, accurate assessment and prediction of desertification sensitivity are essential for effective management and mitigation planning. This study aimed to assess and predict desertification sensitivity in Golestan Province using the Desertification Sensitivity Severity Index (DSSI), based on five components: soil quality, vegetation cover, climate, land management, and atmospheric particulate matter concentration during 2019–2025. To construct the DSSI, multi-source remote sensing data and climatic variables were normalized using the Min–Max method to standardize data scales and then integrated through geometric combination to preserve sensitivity to the weakest component. Future projections were performed under the climate scenarios RCP2.6 and RCP8.5 until 2040, and the model was validated using Random Forest and XGBoost algorithms. Results showed that 32% of the province falls within the high-risk class, 41% within the moderate-risk class, and 27% within the low-risk class. Northern and northeastern areas, including Gonbad-e Kavus, Kalaleh, Aqqala, and Maraveh Tappeh, exhibited the highest sensitivity. Under RCP8.5, high-risk areas are projected to increase to 45% by 2040. The final model achieved 89% accuracy with a Kappa coefficient of 0.85, confirming the effectiveness of integrating remote sensing and machine learning for desertification monitoring and supporting targeted soil and land management strategies.