The study identified three symptom clusters in older adults with diabetes mellitus (DM) and coronary heart disease (CHD) using exploratory factor analysis: cardiopulmonary fatigue, emotional perceptual, and metabolic, explaining 62.3% of the variance. Latent profile analysis divided patients into three subtypes: class 1 "Low burden pattern" (45.4%), class 2 "Psycho-somatic co-dominant pattern" (31.8%) and class 3 "Metabolic-physical dominant pattern" (22.8%). Social isolation scores (LSNS-6) differed significantly between classes (F = 35.67, p < 0.001). Multiple linear regression after adjustment for confounders showed that Class 2 (β = -4.82, p < 0.001) and Class 3 (β = -3.25, p < 0.001) had lower LSNS-6 scores than Class 1, indicating a higher degree of social isolation. The psycho-somatic pattern subtype showed the strongest association with social isolation. Early identification of high burden subgroups can facilitate integrated interventions targeting physical, psychological and social aspects.