Abstract
Most research on cognitive aging relies on either cross-sectional comparisons — snapshots of cognition at a single time point — or pre-defined clinical categories such as mild cognitive impairment (MCI) and Alzheimer’s disease. Both approaches have limitations. Cross-sectional designs cannot capture within-person change, and clinical categories impose hard boundaries on what is fundamentally a continuous process. An alternative is to let the data reveal the natural trajectories of cognitive change over time, without presupposing how many distinct trajectories exist or where the boundaries between them fall.
This project takes that data-driven approach. Using longitudinal cognitive assessments from cognitively normal older adults, we developed a methodology to identify natural cognitive trajectories — groupings that emerge from the temporal patterns of change themselves rather than from diagnostic labels applied after the fact. The analysis distinguishes Cognitive Decliners, Positive-Agers, and Normal-Agers as distinct trajectory classes, each with characteristic rates and patterns of change.
A particularly notable outcome of this work is a proposed procedure that, despite the study’s longitudinal design, can identify individuals likely to experience severe future cognitive decline based on baseline assessments alone. This is clinically significant because it means the trajectory model could function as a practical screening tool: a single assessment session could flag individuals who are on a declining path years before their cognition crosses a clinical threshold. If validated in independent cohorts, this approach could be integrated into routine health evaluations for older adults, enabling targeted monitoring and early intervention for those at highest risk.
The manuscript describing this methodology and its findings is currently in review at IISE Transactions on Healthcare Systems Engineering.