The field of predictive modeling is experiencing significant growth, with a focus on developing innovative solutions for healthcare and other applications. Recent research has emphasized the importance of leveraging complex data structures, such as graphs and temporal point processes, to improve predictive accuracy and transparency. Notably, graph neural networks and neural controlled differential equations have been employed to model patient criticalness and survival outcomes, demonstrating state-of-the-art performance and interpretability. Furthermore, research has highlighted the need to account for clinical presence and observation processes when developing predictive models, leading to improved transportability and performance. Overall, the field is moving towards more sophisticated and nuanced modeling approaches that can capture complex relationships and dynamics in data. Noteworthy papers include: Similarity-Based Self-Construct Graph Model for Predicting Patient Criticalness Using Graph Neural Networks and EHR Data, which proposes a novel graph-based approach for predicting patient mortality and criticalness. TrajSurv: Learning Continuous Latent Trajectories from Electronic Health Records for Trustworthy Survival Prediction, which develops a model that learns continuous latent trajectories from longitudinal EHR data for trustworthy survival prediction.