Patients with Parkinson's disease (PD) possess metabolic dysfunction and experience major adverse cardiovascular events (MACE), which further exacerbate disease progression, increasing mortality risk, subsequently complicate both clinical management and therapeutic outcomes. The risk of MACE increases with diabetes; however, the underlying mechanisms are yet to be determined. Hence, this study aims to evaluate adverse events associated with antidiabetic medications and their potential association with PD and CVD utilizing pharmacovigilance data, to identify demographic and clinical risk factors, and to develop predictive models using statistical and machine learning approaches for stratifying high-risk patients. This study analyzed the reported adverse events (AE) related to antidiabetic agents reported to the Korean Adverse Event Reporting System, constructed by the Korea Institute of Drug Safety and Risk Management, from January 2015 to December 2024 (KIDS KAERS DB (2505A0010)). A cross-sectional analysis and logistic regression models were employed to evaluate the association between antidiabetic drug exposure, Parkinson’s disease (PD), cardiovascular disease (CVD), and related risk factors. Additionally, machine learning models (random forest, XGBoost, etc.) were developed to estimate and predict risk profiles. This study received excemption by Kyung Hee University Institutional Review Board (KHSIRB-25-451). Older adults had a higher AE risk (OR: 1.383, 95% CI: 1.295–1.476). The presence of comorbidities such as metabolic and nutritional disorders (OR: 10.678, 95% CI: 7.381-15.449) and respiratory diseases (OR: 12.708, 95% CI: 6.685–24.157) had a higher risk of CVD and neurological disease AE. In patients with Parkinson's disease, the case count was limited, and the AEs associated with CVD were not statistically significant. In the predictive model, random forest showed high reproduction rate, XGBoost was excellent in accurate positive prediction, and logistic regression analysis showed an overall balanced performance.

The observed association between antidiabetic drug use and risk of PD and CVD appears to be largely influenced by patient-specific factors, including age and comorbidities. These findings emphasize the need for future investigations that comprehensively account for demographic and clinical characteristics when evaluating drug-related outcomes.