A Longitudinal Hybrid Framework for Modeling Neurodegenerative Disease Progression: A Comparative Approach with Proportional Hazards Models

This preprint introduces a longitudinal hybrid framework for modeling neurodegenerative disease progression using structured observational data. The proposed approach focuses on capturing temporal dependencies and inter-individual variability while maintaining compatibility with established statistical methods, particularly proportional hazards models. The framework is designed to operate on longitudinal datasets with irregular sampling and mixed clinical features, enabling a more flexible representation of disease trajectories compared to traditional survival analysis approaches. By integrating temporal encoding, state aggregation, and nonlinear risk estimation, the model provides a conceptual structure for analyzing complex progression dynamics. This work is presented as an early-stage methodological contribution. The current version emphasizes conceptual clarity, reproducibility, and structural design, rather than finalized clinical validation. The framework is intended to support further independent evaluation, replication, and extension by the research community. No sensitive patient-level data are included in this publication. All concepts are presented at an abstracted methodological level. This work is part of an ongoing independent research effort focused on longitudinal systems modeling and predictive analytics in complex biological processes.