Machine learning-based fatigue classification using heart rate variability and cortisol: A multimodal approach to wearable health monitoring

Abstract

Objective: Fatigue is a critical indicator in modern health management, and efficient, accurate methods for predicting fatigue levels using wearable devices have garnered increasing attention. Although recent advancements have enabled non-invasive cortisol measurement via wearable sensors, it remains unclear how effectively cortisol, in combination with other physiological biomarkers, predicts fatigue. Therefore, this study aimed to evaluate the effectiveness of a multimodal machine learning model that integrates cortisol levels and heart rate variability (HRV) for fatigue prediction.

Methods: Data from 336 participants who completed the Fatigue Severity Scale (FSS) were analyzed. Missing data mechanisms for cortisol were examined, and multivariate imputation by chained equations (MICEs) were applied. A TabNet deep-learning model was used to predict low and high fatigue levels based on HRV and cortisol data.

Results: The model using only HRV variables achieved a test AUC of 0.774, whereas the model incorporating both HRV and cortisol levels achieved 0.741, indicating a minimal overall performance difference. Feature importance analysis revealed that, in the cortisol-included model, predictions relied on a limited set of features. When feature selection was applied to this model, a reduced set of variables-age, cortisol, and logarithmic very low frequency-achieved comparable predictive performance (AUC = 0.759) without performance degradation.

Conclusion: This study demonstrated that a fatigue prediction model based on cortisol and HRV can maintain significant predictive power with a reduced number of variables. These findings suggest the potential for practical implementation in wearable devices, enabling accurate fatigue monitoring while minimizing sensor count and computational burden.

Keywords: Heart rate variability; cortisol biosensing; fatigue severity prediction; multimodal machine learning; wearable health monitoring.

Figure 1.

Receiver operating characteristic (ROC) curves for each model condition. (A) The model, including the imputed cortisol variable (Model with Cortisol), yielded an AUC of 0.74. (B) The model excluding the cortisol variable (Model without Cortisol) achieved an AUC of 0.77. (C) The Feature Selection Model, which used only age, cortisol, and VLF (ln), demonstrated an AUC of 0.76. All models showed moderate discriminative performance, with minimal differences across experimental conditions.

Figure 2.

SHAP summary plots for feature importance in different model settings. (A) In the model, including the imputed cortisol variable, a small number of top features—including cortisol and age—exhibited wide SHAP value dispersion, indicating a strong influence on model predictions. Lower-ranked features showed values tightly clustered around zero. (B) In the Model without Cortisol, the overall distribution of SHAP values was more uniform, and no single feature showed dominant predictive power. These findings support the role of cortisol as a key contributor and justify its inclusion in the final feature set.