The field of urban planning and time series analysis is witnessing significant developments, driven by the integration of machine learning and geographical information systems. Researchers are exploring novel frameworks and techniques to analyze complex nonlinear relationships between urban characteristics and health outcomes, as well as to forecast epidemic spread and player behavior in online games. Notable advancements include the development of explainable and interpretable models, such as graph neural networks and causal spatiotemporal graph neural networks, which are capable of capturing spatial heterogeneity and temporal trends. These models have shown promising results in predicting disease transmission, drug response, and player behavior, and are likely to have a significant impact on public health and urban planning decisions. Some noteworthy papers in this area include:
- MedGNN, which improved predictions of medical prescriptions by over 25% compared to baseline methods.
- Unifying Physics- and Data-Driven Modeling, which introduced a novel causal spatiotemporal graph neural network for interpretable epidemic forecasting.
- Explainable and Interpretable Forecasts, which proposed a novel deep Actionable Forecasting Network for forecasting player behavior in online games.