Comprehensive Summary
This paper by Seo et al. examines how chronic diseases (CD) affect mental health later on in life by measuring depression and suicide ideation (SI). This information was analyzed via the Korea National Health and Nutrition Examination Surveys through the use of five different machine learning models (MLMs). These models predicted depression and SI based on the number of CDs as well as the type of CDs present. Additionally, the paper assessed the independent predictive ability of the number and specific types of CDs by ranking feature importance. Shapley Additive Explanation (SHAP) values were used to predict depression and suicidal ideation during the study. The MLMs had a strong predictive power, showing the area under curves (AUC) for depression to range from 0.729 to 0.772, and the AUC for SI ranged from 0.754 to 0.793. The top predictors for depression, according to the models, were osteoarthritis, myocardial infarction, diabetes, asthma, and stroke. For SI, the top CD predictors were stroke, hypertension, asthma, and myocardial infarction.
Outcomes and Implications
This paper uses machine learning to highlight the prevalence of depression and suicidal ideation in those with chronic diseases. Using data from multiple surveys, the study was able to determine that older adults with chronic diseases had an increased risk of depression. Clinically, the findings suggest that routine mental health checkups should be utilized in aging demographics, specifically among those with chronic diseases. These routine mental health checkups should be utilized in those with stroke, arthritis, myocardial infarctions, and diabetes. As the study used survey data from people over the age of 65, future research should investigate the effects of CDs on depression and SI among younger populations.