Landslides on Mount Marapi, West Sumatra, Indonesia, pose significant threats to human life, infrastructure, and the surrounding environment. This study examines patterns of surface deformation and predicts landslide susceptibility on Mount Marapi by integrating Interferometric Synthetic Aperture Radar (InSAR) interferograms with a multi-model machinelearning framework. Landslide susceptibility prediction was conducted using K-Nearest Neighbors (KNN), Random Forest (RF), Support Vector Machine (SVM), and Gradient Boosting (GB) algorithms. Integration of InSAR-derived deformation data with geospatial variables and machine learning facilitates a more comprehensive understanding of deformation dynamics and landslide susceptibility. The findings reveal a dominant trend of subsidence accompanied by persistent fluctuations along the mountain slopes, largely driven by recurrent landslides across multiple locations, thereby reshaping the surface morphology. InSAR data provide a spatially continuous representation of deformation within landslide-prone zones. Among the evaluated algorithms, RF demonstrated the highest predictive performance, yielding an accuracy of 94.33% and a precision of 91.30%. Key variables such as slope gradient, elevation, and rainfall emerged as the most influential determinants of landslide susceptibility. Additionally, seismic activity along the subduction zone and tectonic earthquakes act as further triggers, intensifying landslide occurrence in conjunction with extreme rainfall events. Although secondary factors such as curvature, soil type, and localized deformation exert relatively weaker influences, their interactions remain important in constructing a holistic landslide prediction framework. Overall, this study highlights the effectiveness of integrating InSAR with machine learning to advance multivariable approaches for forecasting and mitigating landslide hazards on Mount Marapi.