Related Experiment Video
Updated: Jul 20, 2025
09:16
Multi-Modal Signals for Analyzing Pain Responses to Thermal and Electrical Stimuli
Published on: April 5, 2019
10.8K
Parametric Modeling and Deep Learning for Enhancing Pain Assessment in Postanesthesia.
IEEE Transactions on Bio-Medical Engineering
|August 1, 2023
Summary
This study introduces a novel method for pain estimation using fractional-order impedance models and deep learning. This approach offers objective, personalized pain prediction to improve patient safety and avoid medication over-dosing.
Area of Science:
- Biomedical Engineering
- Artificial Intelligence in Medicine
- Pain Management
Background:
- Reliable and objective pain measurement remains a significant challenge in clinical practice.
- Current pain assessment methods, like the Numeric Rating Scale (NRS), are subjective and prone to variability.
- Post-surgical pain management requires accurate monitoring to prevent under- or over-dosing of analgesics.
Purpose of the Study:
- To develop and validate a more reliable method for post-surgical pain estimation.
- To enhance pain assessment objectivity by integrating fractional-order impedance models (FOIM) and deep learning (convolutional neural networks - CNNs).
- To explore the potential of AI in personalized nociception-antinociception prediction for closed-loop analgesia systems.
Main Methods:
- Utilized a recursive identification method to parameterize fractional-order impedance models (FOIM) from skin impedance measurements.
- Employed deep learning, specifically convolutional neural networks (CNNs), for pain classification using time-frequency data and spectrograms.
- Validated online parameter estimates and CNN predictions against patient-reported pain levels (NRS) in a proof-of-concept clinical trial.
Main Results:
- Changes in FOIM parameters and online-retrained CNN predictions correlated with nociception trends and NRS scores.
- The developed models demonstrated the ability to predict pain intensity and tissue heterogeneity.
- Online predictions from retrained CNNs showed more specific trends compared to offline population-trained models.
Conclusions:
- Tailored online identification and deep learning methods can provide objective pain estimations even in artifact-corrupted environments.
- AI-driven models offer personalized nociception-antinociception prediction, crucial for patient safety.
- These advancements support the design and evaluation of closed-loop analgesia controllers for improved pain management.
