Stress management with HRV following AI, semantic ontology, genetic algorithm and tree explainer.

Area of Science:

• Utilizes artificial intelligence (AI) and machine learning (ML) for biomedical signal processing.
• Focuses on computational intelligence and knowledge representation in healthcare.
• Integrates data science with physiological monitoring for stress assessment.

Background:

• Heart Rate Variability (HRV) is a key physiological indicator of stress, with lower HRV signifying higher stress levels.
• Existing AI research seeks to leverage HRV data for precise stress classification and early well-being interventions.
• The SWELL-KW dataset provides labeled HRV data crucial for training and validating stress detection models.

Purpose of the Study:

• To construct a semantic model of HRV features within a knowledge graph.
• To develop an accurate, reliable, explainable, and ethical AI model for predictive HRV analysis.
• To enhance stress management systems through optimized HRV feature analysis and semantic frameworks.

Main Methods:

• Employed feature selection (genetic algorithms) and dimensionality reduction techniques to optimize HRV data.
• Applied various ML algorithms (traditional and ensemble, including Random Forest Classifier) to imbalanced and balanced datasets.
• Utilized oversampling techniques (SMOTE, ADASYN) and SHAP for model interpretability and bias mitigation.

Main Results:

• The combination of genetic algorithm-based feature selection and Random Forest Classifier demonstrated high accuracy in stress classification.
• Non-linear HRV features proved particularly effective when analyzed with optimized feature sets.
• A semantic framework with domain ontology improved data representation and knowledge acquisition, validated by Hermit reasoners.

Conclusions:

• The developed AI model provides accurate, explainable, and ethical stress level classification based on HRV.
• Optimized HRV features are crucial for effective stress management systems within a semantic context.
• HRV monitoring, when integrated with other assessments and semantic insights, offers a holistic approach to understanding and managing stress.