Development of a Bayesian nomogram for predicting depression in family caregivers for dementia patients at home in South Korea

This study identified multiple risk factors for depression using machine learning by using main caregivers for dementia patients in a home care setting who participated in a national epidemiologic survey in South Korea. Subjects were 25,634 main caregivers (without depression = 21,369, and with depression = 4265), who took care of dementia patients at home based on the 2019-to-2020 community health survey. This study developed a model for predicting the depression of dementia caregivers by using Bayesian nomogram and 6 variables with the highest feature importance identified in LightGBM to understand the relationship between predictive factors regarding the depression of caregivers in South Korea. The results of this study showed that subjective stress level, subjective health status, cognitive impairment counseling in the past year, economic activity, gender, and dementia screening in the past year were key risk factors for predicting the depression of dementia caregivers. The results of 10-fold cross-validation showed that the area under the curve, general accuracy, precision, recall, and F1-score of the developed nomogram were 0.82, 0.85, 0.83, 0.85, and 0.83, respectively. In conclusion, this study not only emphasizes the critical role of tailored screening and support for dementia caregivers to prevent depression but also sets the groundwork for future research on effective mental health interventions in this demographic. However, this study is limited by its cross-sectional design and reliance on self-reported measures, and future longitudinal research is warranted.