Examining Factors Contributing to Age of Late-Onset Autoimmune Disease in U.S. Adults Using Hierarchical Clustering Mechanisms
Author(s)
Jordyn R. Homoki, BS, Vicky W. Li, MPH, Kyla Finlayson, MS, Lulu K. Lee, PhD.
Oracle, Austin, TX, USA.
Oracle, Austin, TX, USA.
Presentation Documents
OBJECTIVES: This study examined characteristics of individuals with autoimmune disease(s) (AD) in a general U.S. adult population using unsupervised clustering.
METHODS: Data on 17,295 adults with AD were analyzed from the 2022 and 2023 National Health and Wellness Survey. A total of 23 variables were included in analysis, including disease-specific factors, demographics, and comorbidities. Gower’s Distance was used to evaluate similarity between observations. Clusters were aggregated by within-cluster sum of squares. Average silhouette was examined using single, complete, average, and Ward’s linkage.
RESULTS: Clustering analysis identified four clusters to be optimal using Ward’s linkage: (1) oldest average age of late-onset AD (44.0 years old), fewest number of different ADs, predominantly white, least employed, presented fewest symptoms of depression and anxiety, had the least amount of social support, and highest prevalence of hypertension yet the fewest with migraines; (2) youngest average age of late-onset AD (35.3 years old), had the most organ-specific ADs, predominantly male, contained the largest proportion of Hispanics, the most educated and actively employed, contained the most smokers by 3-fold, lowest BMI, presented with the highest number of symptoms of depression and anxiety, and had the fewest number of family history risk factors; (3) highest prevalence of early-onset AD yet not the earliest age of onset, least diagnosed with multiple autoimmune syndrome, predominantly female, largest percentage of Asians, majority never smokers, had the most social support, and generally fewest comorbidities; (4) most diagnosed with systemic AD, had greatest number of different ADs, predominantly female, least educated and lowest income, highest BMI, nearly all individuals were diagnosed with migraines, and had the most family history risk factors.
CONCLUSIONS: Hierarchical clustering identified potential distinct profiles of individuals with varying ages of late-onset AD to be considered in future clinical research for AD development.
METHODS: Data on 17,295 adults with AD were analyzed from the 2022 and 2023 National Health and Wellness Survey. A total of 23 variables were included in analysis, including disease-specific factors, demographics, and comorbidities. Gower’s Distance was used to evaluate similarity between observations. Clusters were aggregated by within-cluster sum of squares. Average silhouette was examined using single, complete, average, and Ward’s linkage.
RESULTS: Clustering analysis identified four clusters to be optimal using Ward’s linkage: (1) oldest average age of late-onset AD (44.0 years old), fewest number of different ADs, predominantly white, least employed, presented fewest symptoms of depression and anxiety, had the least amount of social support, and highest prevalence of hypertension yet the fewest with migraines; (2) youngest average age of late-onset AD (35.3 years old), had the most organ-specific ADs, predominantly male, contained the largest proportion of Hispanics, the most educated and actively employed, contained the most smokers by 3-fold, lowest BMI, presented with the highest number of symptoms of depression and anxiety, and had the fewest number of family history risk factors; (3) highest prevalence of early-onset AD yet not the earliest age of onset, least diagnosed with multiple autoimmune syndrome, predominantly female, largest percentage of Asians, majority never smokers, had the most social support, and generally fewest comorbidities; (4) most diagnosed with systemic AD, had greatest number of different ADs, predominantly female, least educated and lowest income, highest BMI, nearly all individuals were diagnosed with migraines, and had the most family history risk factors.
CONCLUSIONS: Hierarchical clustering identified potential distinct profiles of individuals with varying ages of late-onset AD to be considered in future clinical research for AD development.
Conference/Value in Health Info
2025-05, ISPOR 2025, Montréal, Quebec, CA
Value in Health, Volume 28, Issue S1
Code
MSR156
Topic
Methodological & Statistical Research
Topic Subcategory
Artificial Intelligence, Machine Learning, Predictive Analytics
Disease
No Additional Disease & Conditions/Specialized Treatment Areas, SDC: Systemic Disorders/Conditions (Anesthesia, Auto-Immune Disorders (n.e.c.), Hematological Disorders (non-oncologic), Pain)