Dementia risk analysis using temporal event modeling on a large real-world dataset

Improvement of dementia diagnosis and risk stratification is of critical importance as the population continues to age. Unraveling temporally antecedent risk factors and health event trajectories has the potential to improve our understanding of disease processes, paving the way for tailored interventions and preventive strategies. Historically, research concerning the precursors to dementia has primarily focused on the identification of individual factors without considering temporal dependencies. Although some work has used tools such as temporal streaming clustering²⁶, and longitudinal latent class mixture models²⁷ to construct predictive models of dementia, they are limited in scope by both the size of the population analyzed and the breath of diseases or healthcare processes considered as predictive factors. These methodologies employ either fully, or semi-supervised approaches,…