Objective To propose and validate a spatiotemporal risk prediction framework for porcine reproductive and respiratory syndrome virus (PRRSV) by integrating epidemiological data, environmental factors, and phylodynamic data, thereby addressing the limitations of existing transmission risk models in dynamically capturing transmission processes and quantifying risk gradients.

Method A multidimensional feature system was constructed, and a continuous risk score (risk_score) was used to consistently quantify transmission risk at the “province-month” scale. Comparative experiments were performed using a classical baseline model (Gradient boosting), traditional machine learning models (SVR, SVR-L, and XGBoost), and deep learning models (LSTM and Transformer). Model performance was further evaluated by province-wise cross-validation and rolling time-window validation.

Result Compared with the conventional binary-label model, the framework based on the continuous risk score reduced the mean absolute error (MAE) by 67.6%. Phylodynamic/transmission features and historical features were identified as the principal sources of information contributing to model performance, recovering 93.6% and 90.3% of the predictive performance of the full-feature model, respectively. As the proportion of training data increased and engineered features were introduced, the MAE of the Transformer model decreased from 0.39 to below 0.18. The MAE of the LSTM model also declined steadily, with a narrower range of error fluctuations. In province-level validation, XGBoost and Gradient boosting exhibited strong overall robustness, whereas LSTM and Transformer showed better regional adaptability in provinces such as Henan and Shandong.

Conclusion The proposed spatiotemporal risk prediction framework improves the accuracy of PRRSV transmission risk prediction and enhances the utilization efficiency of small-sample data, providing a generalizable methodological reference for the precise prevention and control of PRRSV and other animal infectious diseases.