Bayoud disease, caused by Fusarium oxysporum f. sp. albedinis (Foa), threatens date palm cultivation, especially in North Africa, where date production is crucial for food security and economic stability. Current management strategies, including resistant cultivars and chemical treatments, have limited success because of the genetic adaptability and survival of the pathogen in soil. Early detection remains challenging because current methods rely on visible symptoms that appear after significant damage. This study explored the potential of machine learning (ML) to predict the soil suppressiveness of Bayoud disease by analyzing microbial metabolic activity using Biolog SF-P2 assays. Sixty Fusarium isolates from suppressive and conducive soils in Morocco were assessed using 95 different carbon sources. Six ML models Logistic Regression, Naive Bayes, K-Nearest Neighbors (KNN), XGBoost, Gradient Boosting, and Support Vector Machine (SVM) were applied to the data. The results showed that the XGBoost and Gradient Boosting models achieved the highest predictive accuracy, with AUC values exceeding 90%, indicating a strong classification ability. SHAP analysis identified key metabolic markers linked to disease suppression, thereby highlighting the role of microbial communities in pathogen resistance. This study established a data-driven framework for predicting soil suppressiveness and facilitating proactive disease management.