Mapping Symptom Complexity in Dementia With Lewy Bodies: A Network Analysis Approach

Dementia with Lewy bodies (DLB) is the second most common neurodegenerative dementia after Alzheimer's disease but remains underdiagnosed because of its heterogeneous symptomatology. Network analysis may capture interrelationships among symptoms, potentially improving diagnostic precision. The authors applied network analysis to a cohort of patients with DLB to map symptom connections, identify central features, and explore clustering patterns.

Clinical records of 107 patients with a confirmed DLB diagnosis from Brugmann University Hospital, Brussels, were retrospectively analyzed. Sixteen core and supportive symptoms were coded as binary variables. Pairwise Markov random fields with least absolute shrinkage and selection operator regularization were used to construct a symptom network. Centrality indices, predictability, and community detection were computed to assess the structural importance and clustering of symptoms. Stability was evaluated via nonparametric bootstrap procedures, and redundancy was addressed with unique variable analysis, prompting reestimation of a reduced network.

The network was characterized by positive correlations among symptoms, with visual hallucinations and urinary incontinence showing the highest centrality. Cognitive fluctuation and tactile hallucinations were disconnected despite high predictability from low variance. Three clusters emerged: cognitive-perceptual, motor-perceptual, and autonomic-somatic. Removal of agitation, prompted by redundancy between agitation and delirious thoughts, reduced centrality for delirious thoughts and depression. Stability analyses indicated moderate robustness for the full network and slightly improved stability in the revised model.

Central symptoms such as visual hallucinations and urinary incontinence may serve as high-yield diagnostic targets in DLB, and clustering patterns could represent meaningful clinical dimensions. Network analysis may help refine diagnoses and guide targeted interventions in DLB.