Identification of Adolescents With Major Depressive Disorder Using Random Forest Based on Nocturnal Heart Rate...

Area of Science:

• Cardiology
• Psychiatry
• Biomedical Engineering

Background:

• Major Depressive Disorder (MDD) in adolescents is frequently underdiagnosed, relying heavily on subjective clinical assessments.
• Adolescent MDD is often associated with sleep disturbances and reduced heart rate variability (HRV).

Purpose of the Study:

• To develop and validate an automated classification model for identifying adolescent MDD using nocturnal HRV features.
• To assess the impact of data duration and feature importance on diagnostic accuracy.

Main Methods:

• Sixty-three adolescents (depressed and healthy controls) underwent three-night electrocardiogram (ECG) monitoring.
• Nineteen time, frequency, and nonlinear HRV features were extracted from overnight and short-term (5-min) data segments.
• A Bayesian-optimized random forest (BO-RF) classifier was employed and evaluated using ten-fold cross-validation.

Main Results:

• The BO-RF model achieved 80.6% predictive accuracy using long-term nocturnal HRV features.
• Short-term data analysis showed improved accuracy (75.0%) with increased data accumulation.
• Poincaré plot-derived features, particularly heart rate asymmetry (HRA), were crucial for differentiating depressed adolescents.

Conclusions:

• Nocturnal HRV analysis provides an effective, objective method for differentiating adolescents with MDD from healthy controls.
• This approach offers a potential advancement in the early and accurate diagnosis of adolescent depression.
• The findings support the integration of HRV monitoring into wearable devices for widespread application in adolescent mental health screening.