Exploring Depression and Nutritional Covariates Amongst US Adults using Shapely Additive Explanations

Area of Science:

• Nutritional Science
• Data Science
• Public Health

Background:

• Depression significantly impacts individual and public well-being.
• Identifying natural therapeutics, such as nutritional interventions, is crucial for addressing this public health issue.
• Dietary patterns and nutrient intake are increasingly recognized as potential modifiable factors in mental health.

Purpose of the Study:

• To identify feature importance in a machine learning model using only nutrition-related variables.
• To explore the relationship between specific nutrient intakes and depressive symptoms.
• To determine which nutritional factors are most predictive of depressive symptoms within a large population cohort.

Main Methods:

• Retrospective analysis of the National Health and Nutrition Examination Surveys (NHANES 2017-2020) data.
• Inclusion of 7929 adult patients who completed the 9-item Patient Health Questionnaire (PHQ-9) and a nutritional intake questionnaire.
• Application of univariable regression to select significant nutritional covariates and an XGBoost machine learning model to determine feature importance.

Main Results:

• The machine learning model identified 24 significant nutritional features out of 60.
• The XGBoost model achieved an Area Under the Receiver Operator Characteristic Curve (AUROC) of 0.603.
• Top features influencing the model included Potassium Intake (6.8%), Vitamin E Intake (5.7%), Number of Foods and Beverages Reported (5.7%), and Vitamin K Intake (5.6%).

Conclusions:

• Machine learning models can effectively identify nutritional covariates associated with depression.
• Feature importance analysis highlights specific nutrients for potential therapeutic intervention.
• Further research into these nutritional factors may lead to novel dietary strategies for managing depressive symptoms.