arXiv:2607. 21817v1 Announce Type: cross Abstract: Longitudinal studies often collect data at sparse, irregular, and unequally spaced time points.
Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.
Random Hazard Forests (RHF) is a survival tree ensemble that models how a patient's hazard changes over continuous time as new measurements arrive. RHF directly estimates a nonparametric hazard likelihood for predictable covariate processes, using an efficient working model to guide tree construction and then estimating flexible time‑varying hazards at each terminal node. By routing each tree based on the covariate state immediately before each time point, RHF can handle irregular and asynchronous covariate updates, and averaging across trees yields a pathwise hazard estimate that accurately captures changing risk in simulations and an intensive‑care application.
DeMMO is an interpretable framework that models longitudinal digital mobility outcomes (DMOs) across multiple diseases and outcomes using multi-task learning. It introduces a cross-disease, cross-outcome relation-learning mechanism that learns signed relationships from longitudinal DMO coefficient matrices, allowing selective information sharing even when disease cohorts lack shared participants. Evaluated on the Mobilise‑D dataset, DeMMO outperforms nine strong baselines and identifies reliable longitudinal DMO patterns for clinical validation.
The paper introduces a new algorithm that uses decision trees and random forests to estimate individual treatment effects while providing interpretability. It modifies the standard random forest splitting criterion by combining a heterogeneity-focused criterion with a bias-correction criterion, enabling the model to handle observational studies with varying treatment propensities without separately estimating propensity scores. The resulting tree structure directly reveals which features drive treatment effect differences, and simulation studies show the method matches or surpasses existing approaches in prediction accuracy while improving interpretability.
arXiv:2606. 09907v1 Announce Type: cross Abstract: Multimodal clinical learning is increasingly important for integrating diverse patient data, including imaging, text, and personalised health records.
The paper introduces Neural ODE-LMM, a hybrid model that integrates a Neural Ordinary Differential Equation into a linear mixed‑effects framework to flexibly learn time‑varying associations between exposures and outcomes. By encoding covariate trajectories into a continuous‑time latent state, the method preserves standard likelihood inference while capturing complex, potentially cumulative effects without pre‑specifying functional forms. Simulations demonstrate accurate recovery of instantaneous and cumulative effects, and application to the 3C cohort uncovers BMI and fasting glucose trajectories linked to cognitive decline.
arXiv:2606. 05488v1 Announce Type: cross Abstract: Identifying subtypes of complex conditions, such as Inflammatory Bowel Disease (IBD), often requires capturing latent patterns in longitudinal omics data.