Key result
Machine learning predicts depression with ~94% accuracy, linking sufficient mineral and vitamin intake to lower risk.
Why the study?
Depression has rapidly increased in Korea and globally, with limited understanding of how nutritional factors affect depression risk using big data and machine learning.
Cross-Sectional (n=11,463)
Effect estimate: Accuracy 94.3, AUC 0.88
Machine learning models incorporating nutritional factors alongside general characteristics achieved high accuracy (94.3%) in predicting depression, highlighting the importance of diet in mental health.
Nutritional data may enhance ML depression prediction; hypothesis-generating and should not yet change practice.
Background: In recent years, depression has been rapidly increasing in Korea, with a 35.1% rise in cases over the past 5 years.Globally, about 280 million people suffer from depression.While many studies focus on social factors, some also point to the impact of diet and nutrients on mental health.This research aims to identify how nutrition affects depression using big data and machine learning (ML), providing evidence for precision nutrition medicine and developing a depression prediction system utilizing explainable artificial intelligence for clinical use.Methods: This study utilized data from the Korea National Health & Nutrition Examination Survey (2020, 2018, 2016, and 2014) to analyze depression in adults aged 20-59, focusing on the Patient Health Questionnaire-9 (PHQ-9) scale, which is measured every 2 years.After excluding missing values, 11,463 subjects were included, with 594 classified as depressed (PHQ-9 score ≥ 10) and 10,869 classified as normal (PHQ-9 score ≤ 9).Statistical analysis involved χ 2 tests for general characteristics and independent t-tests for body mass index, sedentary hours, and nutritional intake.ML classification was performed using XGBoost, with an 8:2 data split.Accuracy, precision, recall, F1 score, and area under the receiver operating characteristic curve were calculated to evaluate performance.The results were further interpreted using SHAP summary plots, broken down by sex and age groups.Results: Although there were slight differences in the effects of nutrients on depression depending on sex and age, overall, maintaining sufficient intake of minerals and vitamins was significantly associated with a reduction in depression.Conclusion: ML based depression prediction, despite including well-known factors such as stress, subjective physical health status, socioeconomic factors, health behaviors, and disease conditions, many types of nutrients ranked high in importance and proved effective in enhancing the accuracy of depression diagnosis.Including nutritional factors in depression prediction could contribute to realizing precision medicine as part of a digital healthcare system.
No takes yet. Share an insight, caveat, or question.
Kim et al. (2025) conducted a cross-sectional in Depression (n=11,463). Nutritional factors vs. Unexposed/reference group was evaluated on Depression prediction performance (Accuracy 94.3, AUC 0.88). A machine learning model using XGBoost predicted depression with 94.3% accuracy, demonstrating that sufficient intake of minerals and vitamins was significantly associated with a reduction in depression.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: