A novel and efficient personalized stress detection technique using a deep learning model.

Area of Science:

• Computational linguistics
• Artificial intelligence
• Psychology

Background:

• Stress significantly impacts adult and elder health, leading to chronic conditions.
• Detecting stress from social media text presents challenges due to complexity and computational demands.
• Existing machine learning and deep learning models face limitations like long training times and feature constraints.

Purpose of the Study:

• To develop an efficient and accurate technique for stress detection from social media text.
• To overcome the limitations of existing methods in terms of training time and feature utilization.
• To improve the accuracy of stress detection using novel model integration and optimization.

Main Methods:

• Integration of advanced text representation techniques: FastText, Global Vectors for Word Representation (Glove), DeepMoji, and XLNet.
• Utilizing a Depth-wise Separable Convolution with Residual Network (DSC-ResNet) for accurate stress detection.
• Employing the Chaotic Fennec Fox Optimization Algorithm (CFFO) for hyperparameter tuning.

Main Results:

• The proposed technique achieved a high accuracy of 98.42%.
• Precision, recall, specificity, and F1-score were reported at 97.58%, 98.12%, 98.28%, and 98.38%, respectively.
• The model demonstrated superior performance compared to existing techniques.

Conclusions:

• The novel integration of text representations and DSC-ResNet offers an efficient solution for stress detection.
• The proposed method effectively addresses limitations of previous approaches, providing high accuracy and performance.
• This technique shows promise for real-world applications in mental health monitoring via social media analysis.