Contribution of Chronic Disease in Predicting Depression and Suicidal Ideation Among the Older Adult Population.

Abstract

Objective: This study aimed to clarify how chronic diseases (CDs) contribute to depression and suicidal ideation (SI) prediction using machine learning (ML) techniques among the older adult population.

Methods: National representative data of 5,419 older adults from the Korea National Health and Nutrition Examination Survey conducted in 2013, 2015, 2017, and 2019 were used in this study. The number and type of CDs were incorporated into Models 1 and 2, respectively, using five ML methods.

Results: The average age of the participants was 72.7 years, with 43.2% males, 15.2% reporting depression, and 7.3% reporting SI. The number of CDs was correlated with increased depression and SI. The ML models showed moderate-to-good performance in the prediction of depression and SI. The area under the curve (AUC) values for Model 1 ranged from 0.729 to 0.772 for depression, and from 0.754 to 0.793 for SI. In Model 2, the AUC ranged from 0.704 to 0.768 for depression and from 0.750 to 0.785 for SI. More depression and SI were expected when the number of CDs was one or more and two or more, respectively. The top predictors of depression were osteoarthritis, myocardial infarction, diabetes, asthma, and stroke, whereas those predicting SI were stroke, hypertension, asthma, myocardial infarction, and rheumatoid arthritis.

Conclusion: The number and specific types of CDs predicted depression and SI among Korean older adults. These results may help enhance cooperation with clinicians treating CDs and promote the early detection and prevention of further SI and behaviors.