Academic Bibliography

 Search 200 years of publications by Ghent University researchers.
Advanced search
1 file | 1.78 MB
Add to list
  1. Search results
  2. Artificial intelligence for pain clas...

Artificial intelligence for pain classification with the non-invasive pain monitor Anspec-Pro

Tim De Grauwe, Mihaela Ghita (UGent) , Dana Copot (UGent) , Clara-Mihaela Ionescu (UGent) and Martine Neckebroek (UGent)
(2022) ACTA ANAESTHESIOLOGICA BELGICA. 73. p.45-52
Author
Organization
Project
Abstract
Background: Reliable measurement of perioperative pain is still an ongoing problem. Pain monitors are commercially available, but to date are not commonly used clinically. Anspec-Pro was developed as a new pain monitor device by Ghent University in 2018. The validation study compared this monitor to the commercially available and validated MedStorm pain monitor. Although the results were comparable with the validated monitor, the absolute results were debatable. Objectives: The data were reanalyzed by means of artificial intelligence (AI), examining the correlation and prediction between the measured data and clinical parameters, to explore if this delivers complementary information that assists pain assessment. Design and setting: A cohort study at Ghent University Hospital. Methods: During two monitoring periods, data were collected from patients while measuring pain with Anspec-Pro. Patients were monitored in the preoperative period and during their postoperative recovery. Measurements by Anspec-Pro were processed with AI, more specifically with a convolutional neural network ( CNN), and classified into pain classes. CNN's were trained both with offline (training prior to monitoring) and online (offline training followed by real-time retraining with incoming data) training methods. Performance was assessed with Receiver Operating Characteristic (ROC) curves. Main outcome measures: Pain values as quantified by Anspec-Pro and NRS values as reported by the patients. Results: Data from 11 patients were used for analysis. Good device performance was seen with offline training with all data and with online retraining every seven minutes with device output and an NRS from the last seven minutes. Conclusions: CNN online training with recent patient data led to good algorithm performance. Hence, our results indicate that there is a potential for AI to deliver useful information that can be used in complementary models of monitoring devices.
Keywords
STIMULATION, ANESTHESIA, Pain Measurement, Artificial Intelligence, Convolutional Neural Network, Analgesia, Pain Monitor

Downloads

  • Download
    (...).pdf
    • full text (Published version)
    • |
    • UGent only
    • |
    • PDF
    • |
    • 1.78 MB

Citation

Please use this url to cite or link to this publication:

MLA
De Grauwe, Tim, et al. “Artificial Intelligence for Pain Classification with the Non-Invasive Pain Monitor Anspec-Pro.” ACTA ANAESTHESIOLOGICA BELGICA, vol. 73, 2022, pp. 45–52, doi:10.56126/73.S1.29.
APA
De Grauwe, T., Ghita, M., Copot, D., Ionescu, C.-M., & Neckebroek, M. (2022). Artificial intelligence for pain classification with the non-invasive pain monitor Anspec-Pro. ACTA ANAESTHESIOLOGICA BELGICA, 73, 45–52. https://doi.org/10.56126/73.S1.29
Chicago author-date
De Grauwe, Tim, Mihaela Ghita, Dana Copot, Clara-Mihaela Ionescu, and Martine Neckebroek. 2022. “Artificial Intelligence for Pain Classification with the Non-Invasive Pain Monitor Anspec-Pro.” ACTA ANAESTHESIOLOGICA BELGICA 73: 45–52. https://doi.org/10.56126/73.S1.29.
Chicago author-date (all authors)
De Grauwe, Tim, Mihaela Ghita, Dana Copot, Clara-Mihaela Ionescu, and Martine Neckebroek. 2022. “Artificial Intelligence for Pain Classification with the Non-Invasive Pain Monitor Anspec-Pro.” ACTA ANAESTHESIOLOGICA BELGICA 73: 45–52. doi:10.56126/73.S1.29.
Vancouver
1.
De Grauwe T, Ghita M, Copot D, Ionescu C-M, Neckebroek M. Artificial intelligence for pain classification with the non-invasive pain monitor Anspec-Pro. ACTA ANAESTHESIOLOGICA BELGICA. 2022;73:45–52.
IEEE
[1]
T. De Grauwe, M. Ghita, D. Copot, C.-M. Ionescu, and M. Neckebroek, “Artificial intelligence for pain classification with the non-invasive pain monitor Anspec-Pro,” ACTA ANAESTHESIOLOGICA BELGICA, vol. 73, pp. 45–52, 2022.
@article{01H11FGZWTZ529NQAN585HBT8P,
  abstract     = {{Background: Reliable measurement of perioperative pain is still an ongoing problem. Pain monitors are commercially available, but to date are not commonly used clinically. Anspec-Pro was developed as a new pain monitor device by Ghent University in 2018. The validation study compared this monitor to the commercially available and validated MedStorm pain monitor. Although the results were comparable with the validated monitor, the absolute results were debatable.

Objectives: The data were reanalyzed by means of artificial intelligence (AI), examining the correlation and prediction between the measured data and clinical parameters, to explore if this delivers complementary information that assists pain assessment.

Design and setting: A cohort study at Ghent University Hospital.

Methods: During two monitoring periods, data were collected from patients while measuring pain with Anspec-Pro. Patients were monitored in the preoperative period and during their postoperative recovery. Measurements by Anspec-Pro were processed with AI, more specifically with a convolutional neural network ( CNN), and classified into pain classes. CNN's were trained both with offline (training prior to monitoring) and online (offline training followed by real-time retraining with incoming data) training methods. Performance was assessed with Receiver Operating Characteristic (ROC) curves.

Main outcome measures: Pain values as quantified by Anspec-Pro and NRS values as reported by the patients.

Results: Data from 11 patients were used for analysis. Good device performance was seen with offline training with all data and with online retraining every seven minutes with device output and an NRS from the last seven minutes.

Conclusions: CNN online training with recent patient data led to good algorithm performance. Hence, our results indicate that there is a potential for AI to deliver useful information that can be used in complementary models of monitoring devices.}},
  author       = {{De Grauwe, Tim and Ghita, Mihaela and Copot, Dana and Ionescu, Clara-Mihaela and Neckebroek, Martine}},
  issn         = {{0001-5164}},
  journal      = {{ACTA ANAESTHESIOLOGICA BELGICA}},
  keywords     = {{STIMULATION,ANESTHESIA,Pain Measurement,Artificial Intelligence,Convolutional Neural Network,Analgesia,Pain Monitor}},
  language     = {{eng}},
  pages        = {{45--52}},
  title        = {{Artificial intelligence for pain classification with the non-invasive pain monitor Anspec-Pro}},
  url          = {{http://doi.org/10.56126/73.S1.29}},
  volume       = {{73}},
  year         = {{2022}},
}
Altmetric
View in Altmetric