Background <p>Multimorbidity is highly prevalent among individuals with diabetes and exerts a substantial impact on healthcare systems. This study aims to investigate the prevalence and healthcare costs of multimorbidity in patients with type 2 diabetes and to assess the influence of multimorbidity on healthcare expenditures using machine learning approaches.</p> Methods <p>We conducted a retrospective cohort study utilizing chronic disease management database and health insurance claim database from a city in eastern China. Twenty-nine multimorbidities with a prevalence exceeding 1% among diabetic patients were identified using ICD codes. We analyzed the trends in prevalence and healthcare costs from 2014 to 2019. Machine learning models were developed to predict healthcare expenditures, and SHAP analysis was applied to the optimal model to evaluate the contribution of specific multimorbidity to healthcare costs.</p> Results <p>Among 79,910 patients, the prevalence of multimorbidity increased from 87.9% in 2014 to 99.3% in 2019, while the proportion of healthcare costs attributed to multimorbidity rose from 31.8% to 34.2%. In 2019, the most prevalent conditions were hypertension (88.3%), arthritis (74.7%), and chronic ischemic heart disease (54.6%), whereas the highest-cost conditions included sequelae of cerebrovascular disease ($3,860.8), cerebral infarction ($2,768.8), and renal failure ($1,543.9). SHAP analysis revealed that cerebrovascular disease sequelae, heart failure, chronic ischemic heart disease, and chronic obstructive pulmonary disease had the most significant impact on future healthcare costs for diabetic patients.</p> Conclusions <p>Multimorbidity is nearly universal among individuals with diabetes in China, with cardiovascular, cerebrovascular, and chronic respiratory diseases contributing disproportionately to healthcare expenditures.</p>