Xinwei Ji, Tianming Zhao, Wei Li, Albert Y. Zomaya
{"title":"Automatic Pain Assessment with Ultra-short Electrodermal Activity Signal","authors":"Xinwei Ji, Tianming Zhao, Wei Li, Albert Y. Zomaya","doi":"10.1145/3555776.3577721","DOIUrl":null,"url":null,"abstract":"Automatic pain assessment systems can help patients get timely and effective pain relief treatment whenever needed. Such a system aims to provide the service with pain identification and pain intensity rating functions. Among the physiological signals, the electrodermal activity (EDA) signal emerges as a promising feature to support both functions in pain assessment. In this work, we propose a machine learning framework to implement pain identification and pain intensity rating using only EDA and its derived features. Our solution also explores the feasibility of using ultra-short EDA segmentation of about 5 seconds to meet real-time requirements. We evaluate our system on two datasets: Biovid, a publicly available dataset, and Apon, the one we build. Experimental results demonstrate that using just the ultra-short EDA signal as input, our algorithm outperforms state-of-the-art baselines and achieves a low regression error of 0.90.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":null,"pages":null},"PeriodicalIF":0.4000,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Applied Computing Review","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/3555776.3577721","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}
引用次数: 0
Abstract
Automatic pain assessment systems can help patients get timely and effective pain relief treatment whenever needed. Such a system aims to provide the service with pain identification and pain intensity rating functions. Among the physiological signals, the electrodermal activity (EDA) signal emerges as a promising feature to support both functions in pain assessment. In this work, we propose a machine learning framework to implement pain identification and pain intensity rating using only EDA and its derived features. Our solution also explores the feasibility of using ultra-short EDA segmentation of about 5 seconds to meet real-time requirements. We evaluate our system on two datasets: Biovid, a publicly available dataset, and Apon, the one we build. Experimental results demonstrate that using just the ultra-short EDA signal as input, our algorithm outperforms state-of-the-art baselines and achieves a low regression error of 0.90.