Stress detection is a critical component in modern healthcare and wellness systems, especially for identifying mental health issues before they escalate. This study presents a comprehensive comparison of various machine and deep learning models for stress detection using multimodal physiological data. We investigate classical machine learning models including Logistic Regression (LR), Gaussian Naïve Bayes (GaussianNB), AdaBoost Classifier (AB), XGBoost Classifier (XGBoost), Decision Trees Classifier (Decision Trees), Extra Trees Classifier (Extra Trees), and Random Forest Classifier (Random Forest). In addition, we explore deep learning approaches such as Deep Neural Network (DNN), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN), along with advanced ensemble techniques like Stacking Classifier and Voting Classifier. The models were evaluated based on their accuracy, precision, recall, and F1-score using a diverse physiological dataset containing variables such as heart rate, skin conductance, and respiratory rate. Our findings provide insights into the performance of these models in detecting stress from physiological signals, demonstrating the potential of both classical and deep learning techniques in healthcare applications. The results suggest that while deep learning models like CNN, RNN, and LSTM offer higher accuracy, ensemble methods like Stacking and Voting Classifiers enhance the robustness of the predictions.

Keywords:
Stress detection, machine learning, deep learning, multimodal physiological data, Logistic Regression, Gaussian Naïve Bayes, AdaBoost, XGBoost, Decision Trees, Extra Trees, Random Forest, Deep Neural Network, Convolutional Neural Network, Recurrent Neural Network, Stacking Classifier, Voting Classifier, LSTM, ensemble methods, healthcare.