Enhancing Dementia Risk Screening with CTGAN-Synthesized Periodontal Examination and General Blood Test Data

This study explores dementia risk prediction by synthesizing data using Conditional Tabular Generative Adversarial Networks (CTGAN) from periodontal examinations and blood tests. Developing models that handle various conditions faces challenges such as high costs, limited sample sizes, and missing data. Despite these limitations, periodontal examination data are cost-effective for screening. This study investigates the effectiveness of CTGAN-synthesized data from periodontal examinations and blood tests. We analyzed data from 108 participants, validating with an external set of 41. Results indicate improvements in prediction accuracy and robustness against anomalies and missing values for practical use of synthesized data in medical research.