Conference on Information and Control Systems 2025
Classification of Cognitive Task Performance using HRV Features: A Convolutional Neural Network Approach
- Venue and date
- Conference on Information and Control Systems 2025 · October 22–25, 2025
- Location
- Delpino Resort, Goseong, Gangwon-do, Republic of Korea
- Publication type
- Conference Proceedings Paper
- Pages
- 291–292
- My role
- First author · Poster preparation · Presentation participation
Abstract
This study classifies cognitive task performance from PPG-derived HRV features and evaluates a convolutional neural network approach.
Research Problem
The work examines whether physiological HRV features can distinguish cognitive task performance states.
System Architecture
The pipeline proceeds from PPG acquisition to IBI extraction, HRV feature construction, CNN classification, and metric reporting.
Methodology
PPG timing and IBI/HRV preprocessing produce features used as inputs to the CNN.
Results
The paper abstract reports an average AUC of 0.85 and an F1-score of 0.82.
- Average AUC
- 0.85
- F1-Score
- 0.82
Next step
The paper abstract and conclusion report average AUC 0.85 and F1-score 0.82, while the DBpia abstract reports AUC 0.99 and F1 0.992 and the paper body lists Fold 5 values. These use different evaluation scopes; next, I plan to recalculate them with the same dataset split and metric definitions.
The work examines whether physiological HRV features can distinguish cognitive task performance states.
The pipeline proceeds from PPG acquisition to IBI extraction, HRV feature construction, CNN classification, and metric rep...
PPG timing and IBI/HRV preprocessing produce features used as inputs to the CNN.
The paper abstract reports an average AUC of 0.85 and an F1-score of 0.82.
Paper and related project
BibTeX
@inproceedings{ryu2025cics25cnnhrv,
title={ Classification of Cognitive Task Performance using HRV Features: A Convolutional Neural Network Approach },
author={ Hyeong-Rok Ryu and Woo-Seok Kang and Kyung-Ho Kim },
booktitle={ Conference on Information and Control Systems 2025 },
year={ 2025 },
pages={ 291–292 }
}The public PDF and DBpia record agree on the title, authors, and pp. 291–292, so the matching record URL is linked here.