Landslides are among the most serious and frequent environmental disasters, involving the fall of large masses of rock and soil that can significantly impact human structures and inhabited areas. Anticipating these events is crucial to reduce risks through real-time monitoring of areas at risk during extreme weather events, such as heavy rains, allowing for early warnings. This study aims to develop a methodology to enhance the prediction of landslide susceptibility, creating a more reliable system for early identification of risk areas. Our project involves creating a model capable of quickly predicting the susceptibility index of specific areas in response to extreme weather events. We represent the terrain using cellular automata and implement a random forest model to analyze and learn from weather patterns. Providing data with high spatial accuracy is vital to identify vulnerable areas and implement preventive measures. The proposed method offers an early warning mechanism by comparing the predicted susceptibility index with the current one, allowing for the issuance of alarms for the entire observed area. This early warning mechanism can be integrated into existing emergency protocols to improve the response to natural disasters. We applied this method to the area of Prunella, a small village in the municipality of Melito di Porto Salvo, known for numerous historical landslides. This approach provides an early warning mechanism, allowing for alarms to be issued for the entire observed area, and it can be integrated into existing emergency protocols to enhance disaster response.