Workers who simultaneously engage in dementia caregiving and paid employment represent a critically vulnerable group facing compounded mental health burdens. This study developed a machine learning-based nomogram to predict depression risk among 25,634 family caregivers of home-dwelling dementia patients in South Korea, using data from the 2019-2020 national community health survey. LightGBM identified six key predictors from 135 variables: subjective stress level, subjective health status, cognitive impairment counseling history, economic activity status, gender, and dementia screening history. A Bayesian nomogram derived from multiple logistic regression achieved an AUC of 0.82, accuracy of 0.85, precision of 0.83, recall of 0.85, and F1-score of 0.83 under 10-fold cross-validation. Economically active female caregivers with a history of cognitive impairment counseling and high subjective stress showed a 96% predicted probability of depression, underscoring the urgent need for targeted mental health screening among workers who balance caregiving and employment responsibilities.

• 변해원(한국기술교육대학교 미래융합학부) | Haewon Byeon (Professor, Department of Future Technology, Korea University of Technology and Education (KOREA TECH)) Corresponding author