Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Multi-Representation Diagrams for Pain Recognition: Integrating Various Electrodermal Activity Signals into a Single Image

Created by
  • Haebom

Author

Stefanos Gkikas, Ioannis Kyprakis, Manolis Tsiknakis

Outline

This paper aims to develop an AI-based pain assessment system, specifically proposing a pain assessment pipeline utilizing electrodermal activity (EDA) signals. We present a method for generating various EDA signal representations and integrating and visualizing them for analysis. Experiments with various preprocessing and filtering techniques and representation combinations demonstrate that our proposed approach outperforms or exceeds existing fusion methods. This approach presents a novel approach for objective and accurate pain assessment and could contribute to the development of automated pain assessment systems utilizing various physiological signals. This research was submitted to the AI4PAIN (Second Multimodal Sensing Grand Challenge for Next-Gen Pain Assessment).

Takeaways, Limitations

•
Takeaways:
◦
The utility of pain assessment using electrodermal activity (EDA) signals was demonstrated.
◦
We present a novel pain assessment approach using multi-representational integrated visualization.
◦
We developed a new algorithm that demonstrates superior or equivalent performance compared to existing fusion methods.
◦
It can contribute to the development of an automated system for objective and accurate pain assessment.
•
Limitations:
◦
Since the evaluation was performed using only EDA signals, multi-modality fusion studies with other physiological signals are needed.
◦
Further research is needed on broader datasets and across different pain types to increase the generalizability of the study.
◦
Further clinical validation of the proposed method is needed.
👍
Made with Slashpage