A systematic hybrid machine learning approach for stress prediction.

Area of Science:

• Computer Science
• Psychology
• Health Informatics

Background:

• Stress is a growing global health concern, impacting individuals' well-being and performance.
• Prevalence of stress symptoms like frustration, nervousness, and anxiety is high, particularly among younger populations (40%).
• Early and accurate stress prediction is crucial for mitigating its adverse effects and preventing related health issues.

Purpose of the Study:

• To develop and validate an automated system for accurate stress level prediction.
• To propose a novel hybrid machine learning model for efficient stress detection.
• To demonstrate the superiority of the proposed model over existing state-of-the-art methods.

Main Methods:

• A hybrid model (HB) was developed by integrating Gradient Boosting Machine (GBM) and Random Forest (RF) algorithms.
• Soft voting criteria were employed, utilizing individual model prediction probabilities for the final stress level determination.
• The model's performance was rigorously evaluated using 10-fold cross-validation and statistical T-tests.

Main Results:

• The proposed hybrid model achieved a remarkable 100% accuracy in stress prediction.
• 10-fold cross-validation demonstrated a mean accuracy of 1.00 with a standard deviation of +/-0.00, indicating high model stability.
• Statistical T-tests confirmed the proposed approach's significant outperformance compared to other existing methods.

Conclusions:

• The developed hybrid machine learning model offers a highly accurate and efficient solution for automated stress detection.
• This approach holds significant potential for early intervention and management of stress-related health issues.
• The model's robust performance and validation suggest its applicability in real-world stress monitoring systems.