| Abstrakt: |
Foot-and-mouth disease (FMD) is a highly contagious viral disease that affects cloven-hoofed animals such as sheep, cattle, goats, and buffalo. It is recognized as one of the most destructive animal diseases, primarily attributed to its significant financial impact. We used a machine learning model to predict FMD occurrence at the county level in Iran from 2008 to 2018 using historical data on cases of FMD, so the monthly occurrence of FMD per county was predicted using spatial and temporal lags using the Random Forest (RF) classifier. The cost-sensitive balancing and Borderline Synthetic Minority Oversampling Technique (Borderline SMOTE) methods were also used to handle the imbalanced classification. The results revealed that after applying the cost-sensitive balancing method, the model’s performance, as measured by the Area Under the Receiver Operating Characteristic (ROC) Curve, was 81%. However, when we implemented the Borderline SMOTE Oversampling technique, the method performance increased from 81 to 88%. We also found that our proposed model’s performance is better than when we use climate-related features to predict disease occurrence. The methodologies and findings of this study provide valuable insights for decision-makers to make informed decisions, mitigating county-level FMD occurrences and preventing economic losses. |
Full text from SpringerLink