Abstract INTRODUCTION Neuropsychiatric symptoms (NPSs) are common in dementia, but their patterns in preclinical stages remain unclear. This study identified NPS clusters and associated health factors in a geriatric clinical population. METHODS We analyzed 1234 participants from the Italian GERIatric COgnitive evaluation memory clinic cohort with Neuropsychiatric Inventory data. Clusters were derived using machine learning (K‐means, Elbow method) separately for dementia and dementia‐free groups. Associations were assessed via multinomial logistic regression. RESULTS In the overall cohort, four NPS clusters emerged: minimal NPS, depression‐anxiety‐apathy, depression‐anxiety, and delusions‐agitation‐irritability. Cluster profiles differed between the dementia and dementia‐free groups. Specific clinical and metabolic factors – lipid abnormalities, glycemic control, thyroid dysfunction, and underweight status – were differentially associated with NPS clusters. DISCUSSION Distinct NPS patterns exist across the dementia continuum. These clusters differ in demographic, cognitive, functional, and metabolic profiles, suggesting NPS may precede cognitive decline and represent syndromic entities with diagnostic relevance. Multidimensional, personalized approaches are needed.
Guazzarini et al. (2026) studied this question.