This research pioneers the application of artificial intelligence for predicting brucellosis in dairy cattle in Bangladesh by developing a highly accurate deep learning model (up to 93.94%). A significant contribution is the use of the SMOTE technique to effectively manage imbalanced veterinary data, a common challenge in disease diagnostics. Furthermore, the study identifies and ranks critical clinical risk factors, establishing that a retained placenta is the most significant predictor. By creating association rules to clarify the interplay between these factors, this work provides veterinarians and farmers with a powerful and practical tool for early diagnosis, paving the way to mitigate substantial economic losses in the dairy industry.