Policy aware fusion of educational and economic indicators for public health curriculum optimization

The optimization of public health curricula requires the effective integration of heterogeneous educational, socio economic, behavioral, and course context indicators under policy related constraints. Traditional curriculum design approaches often rely on static rules or isolated educational measurements, making it difficult to capture dynamic learner course interactions, policy consistency, and uncertainty in real world online learning environments. To address these challenges, this study proposes a policy aware framework, termed the Policy Driven Indicator Integrator, for supporting public health curriculum optimization through learning effectiveness prediction.

The proposed framework integrates multiple sources of learner course information into a shared latent representation while incorporating policy constrained regularization, event driven temporal modeling, and uncertainty aware refinement. The Manifold Constrained Optimizer learns policy consistent latent representations from educational, socio economic, behavioral, quiz, and assignment indicators; the Event Driven Curriculum Planner captures temporal learning dynamics across course weeks; and the Uncertainty Propagation Forecaster improves robustness under incomplete or noisy learning records. The Event Segmentation Alignment strategy refines curriculum related representations by aligning temporal, participation, and policy dimensions within a feasibility constrained optimization process. To quantitatively evaluate the framework, curriculum optimization is formulated as a supervised course learning effectiveness prediction task, where course passing status serves as an operational proxy for successful learning outcomes. Experiments on OULAD and the HarvardX MITx Person Course Academic Year 2013 De Identified Dataset demonstrate that the proposed method consistently outperforms representative statistical learning, machine learning, gradient boosted tree, sequence learning, deep tabular learning, and Transformer based tabular baselines in terms of Accuracy, Macro F1, AUROC, and AUPRC.

The results show that policy aware indicator fusion, event driven modeling, and uncertainty propagation provide complementary benefits for robust public health curriculum optimization. This work offers a systematic and data driven approach for improving curriculum decision support in public health education while maintaining consistency with policy aware educational objectives.