Author(s): Sebastião, Raquel
Date: 2022
Persistent ID: http://hdl.handle.net/10773/37467
Origin: RIA - Repositório Institucional da Universidade de Aveiro
Subject(s): ECG monitoring; Pain; Postoperative; Feature correlation; Feature extraction; Statistical tests
Current standard practices to evaluate pain are mainly based on self-reporting instruments. However, pain perception is subjective and influenced by several factors, making objective evaluation difficult. In turn, the pain may not be correctly managed, and over or under dosage of analgesics are reported as leading to undesirable side-effects, which can be potentially harmful. Considering the relevance of a quantitative assessment of pain for patients in postoperative scenarios, recent studies stress out alterations of physiological signals when in the experience of pain. As the Autonomic Nervous System (ANS) functions without conscious control, it is difficult to deceive its reactions, this is a feasible way to assess pain. The goal of the proposed work is to characterize pain in postoperative scenarios through physiological features extracted from the electrocardiogram (ECG) signal, finding features with the potential to discriminate the experience of pain. Using ECG from ‘pain’ and ‘no-pain’ intervals reported from 19 patients during the postoperative period of neck and thorax surgeries, several features were computed and scaled regarding the baseline of each participant to vanish inter-participant variability. Upon, selected features, though pairwise correlation, were analyzed using pairwise statistical tests to infer differences between ‘pain’ and ‘no-pain’ intervals. Results showed that 6 features extracted from ECG are able to discriminate the experience of postoperative pain. These initial results open the possibility for researching physiological features for a more accurate assessment of pain, which is critical for better pain management and for providing personalized healthcare.
