ViTCFHNet: VisionTans Convolutional Forward Harmonic Net for stress detection using photoplethysmography signal

Recently, recognition of stress utilizing photoplethysmography (PPG) signals has attained extensive consideration from research investigators. Even though more interesting research has been reported in this domain, the major difficulty in the module for detecting stress is the utilization of an optimal number of PPG signals and real-world prediction with minimal complexity modules. To conquer these shortcomings, a novel stress detection framework using the PPG signal is introduced by enabling VisionTrans Convolution Forward Harmonic Net (ViTCFHNet). First, an input PPG signal taken from a database is forwarded to the pre-processing stage to remove noise. Then, feature extraction is performed in two phases, namely feature extraction phase-1 and feature extraction phase-2. The extracted features from both phases undergo a data‑augmentation process, which is carried out by enabling the Synthetic Minority Oversampling Technique (SMOTE). From the augmented data, stress is efficiently detected by employing ViTCFHNet. The effectiveness of the ViTCFHNet framework is demonstrated by analyzing performance metrics, such as accuracy, true positive rate (TPR), true negative rate (TNR), positive predictive value (PPV), negative predictive value (NPV), and F-measure, with optimal values of 91.242%, 91.935%, 91.060%, 90.563%, 91.965%, and 90.514%. Furthermore, the proposed ViTCFHNet shows notable performance gains in terms of accuracy over convolutional neural networks and extreme gradient boosting (Conv-XGBoost), multimodal deep learning-based stress detection (MMFD-SD), deep neural network (DNN), convolutional neural network (CNN), 2-dimensional (2D)-CNN, and 1D-CNN, with improvements of 9.268%, 6.278%, 5.488%, 4.364%, 3.283%, and 2.190%, respectively, highlighting its superior classification capability and robustness.

Authors

Institutions

Publication Details

Journal
Smart Science
Published
2026-11-18
DOI
https://doi.org/10.1080/23080477.2026.2749736
Primary Topic
Emotion and Mood Recognition
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

ViTCFHNet: VisionTans Convolutional Forward Harmonic Net for stress detection using photoplethysmography signal

J. Manimaran, C.S.L Vijaya Durga, M. Purushotham Reddy
Smart Science
Emotion and Mood Recognition
article

ViTCFHNet: VisionTans Convolutional Forward Harmonic Net for stress detection using photoplethysmography signal

J. Manimaran, C.S.L Vijaya Durga, M. Purushotham Reddy
article en

Abstract

Recently, recognition of stress utilizing photoplethysmography (PPG) signals has attained extensive consideration from research investigators. Even though more interesting research has been reported in this domain, the major difficulty in the module for detecting stress is the utilization of an optimal number of PPG signals and real-world prediction with minimal complexity modules. To conquer these shortcomings, a novel stress detection framework using the PPG signal is introduced by enabling VisionTrans Convolution Forward Harmonic Net (ViTCFHNet). First, an input PPG signal taken from a database is forwarded to the pre-processing stage to remove noise. Then, feature extraction is performed in two phases, namely feature extraction phase-1 and feature extraction phase-2. The extracted features from both phases undergo a data‑augmentation process, which is carried out by enabling the Synthetic Minority Oversampling Technique (SMOTE). From the augmented data, stress is efficiently detected by employing ViTCFHNet. The effectiveness of the ViTCFHNet framework is demonstrated by analyzing performance metrics, such as accuracy, true positive rate (TPR), true negative rate (TNR), positive predictive value (PPV), negative predictive value (NPV), and F-measure, with optimal values of 91.242%, 91.935%, 91.060%, 90.563%, 91.965%, and 90.514%. Furthermore, the proposed ViTCFHNet shows notable performance gains in terms of accuracy over convolutional neural networks and extreme gradient boosting (Conv-XGBoost), multimodal deep learning-based stress detection (MMFD-SD), deep neural network (DNN), convolutional neural network (CNN), 2-dimensional (2D)-CNN, and 1D-CNN, with improvements of 9.268%, 6.278%, 5.488%, 4.364%, 3.283%, and 2.190%, respectively, highlighting its superior classification capability and robustness.

Smart Science
Annamalai University (IN), Koneru Lakshmaiah Education Foundation (IN)
Good health and well-being
Emotion and Mood Recognition
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon