Age-Specific Multimorbidity Networks and Key Diseases: A Disease-Disease Network Analysis Using Taiwan Biobank Data

<title>Abstract</title> <p>Background Multimorbidity is usually described by disease counts, but this approach may obscure how chronic conditions are organized within populations. We aimed to characterize age-specific multimorbidity organization among adults in Taiwan using disease–disease network analysis of Taiwan Biobank data. Methods This population-based retrospective cross-sectional study used baseline Taiwan Biobank data from adults with complete demographic and self-reported disease information. Participants were stratified into ≤ 39, 40–60, and ≥ 61 years. Thirty-six chronic diseases with sufficient case numbers were included. Pairwise disease associations were estimated using phi correlation coefficients, and statistically significant associations (p &lt; 0.05 by chi-square testing) were used to construct disease–disease networks. Network density, modularity, average weighted degree (strength), and hub diseases, defined as the top 10% of nodes by strength, were compared across age groups. Results Among 84,501 participants, 13,692 were aged ≤ 39 years, 45,688 were aged 40–60 years, and 25,121 were aged ≥ 61 years. The median number of chronic diseases increased from 2 to 3 and 4 across the three age strata. In the overall population, the strongest positive associations were hyperlipidemia–hypertension (phi = 0.213), hyperlipidemia–type 2 diabetes (phi = 0.194), peptic ulcer–gastroesophageal reflux disease (phi = 0.175), and hypertension–type 2 diabetes (phi = 0.173). Across age strata, the number of retained edges increased from 90 to 194 and 237, network density from 0.160 to 0.308 and 0.376, and average strength from 0.251 to 0.360 and 0.485, whereas modularity decreased from 0.504 to 0.479 and 0.304. Hyperlipidemia was the leading hub in all age-specific networks. Other hubs varied by age, including bipolar disorder and type 2 diabetes in younger adults, type 2 diabetes and depression in middle-aged adults, and gastroesophageal reflux disease, dry eye, and depression in older adults. Conclusions Multimorbidity among Taiwanese adults showed age-specific network organization, with greater connectivity and lower modularity in older adults. Disease–disease network analysis may complement disease counts by identifying structurally important conditions for age-tailored surveillance and future longitudinal research.</p>