Comparison of Logistic Regression and Machine Learning Approaches in Predicting Depressive Symptoms: A National-Based...

Area of Science:

• Computational psychiatry
• Epidemiology
• Statistical modeling

Background:

• Machine learning (ML) shows promise for enhanced predictive capabilities over traditional statistical methods.
• The COVID-19 pandemic heightened concerns about mental health, particularly depressive symptoms.
• Accurate prediction of depressive symptoms is crucial for timely intervention.

Purpose of the Study:

• To evaluate the predictive performance of various ML algorithms against logistic regression (LR) for depressive symptoms.
• To compare the efficacy of ML models (random forest, support vector machine, neural network, gradient boosting machine) and LR.
• To identify potential risk factors for depressive symptoms during the COVID-19 pandemic.

Main Methods:

• A national cross-sectional study of 21,916 participants was analyzed.
• ML algorithms including random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) were employed.
• Performance was assessed using sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC).

Main Results:

• Logistic regression (LR) and neural network (NN) exhibited strong performance based on AUC.
• Gradient boosting machine (GBM) achieved the highest sensitivity, specificity, accuracy, precision, and F1-score.
• Most ML models showed a negligible risk of overfitting, with LR, NN, and GBM identified as top-performing models.

Conclusions:

• Logistic regression (LR) performed comparably to machine learning (ML) models in predicting depressive symptoms.
• LR models were effective in identifying potential risk factors for depressive symptoms.
• LR offers a valuable alternative to ML models due to its comparable performance and lower risk of overfitting.