Strategies for Reliable Stress Recognition: A Machine Learning Approach Using Heart Rate Variability Features.

Area of Science:

• Physiological computing
• Machine learning in healthcare
• Affective computing

Background:

• Stress recognition using physiological data, like heart rate variability (HRV), is crucial for mental health interventions.
• Limited datasets in affective computing and healthcare research pose challenges for accurate machine learning (ML) model performance evaluation.
• Robust ML methodologies are needed to ensure reliable stress detection in real-world scenarios.

Purpose of the Study:

• To employ supervised learning algorithms for classifying stress and relaxation states using HRV measures.
• To implement robust strategies for ML with limited datasets, including data segmentation, feature selection, and model evaluation.
• To assess the generalizability and interpretability of ML models for stress recognition.

Main Methods:

• Supervised learning algorithms were utilized to analyze heart rate variability (HRV) data.
• Methodological recommendations for ML with limited datasets were applied, including data segmentation and feature selection.
• Model performance was rigorously evaluated, focusing on generalizability across independent datasets.

Main Results:

• The random forest model demonstrated superior performance in distinguishing stress from non-stress states.
• The model achieved an F1-score of 86.3% for stress versus relaxation and 65.8% for stress versus neutral states.
• The model exhibited generalizability on independent datasets, confirming its ability to differentiate stress and relaxation states.

Conclusions:

• The random forest model is effective for stress recognition using HRV, particularly in differentiating stress from relaxation.
• Robust ML methodologies, despite potentially yielding lower metrics than some studies, are vital for generalizability and interpretability in limited-data applications.
• This study emphasizes the importance of reliable ML models for practical mental health interventions.