This observational study focused on the development of a new analgesic index via photoplethysmogram (PPG) spectrograms and CNN for estimating pain in conscious patients.
For the postsurgery
pain assessment of conscious patients, the newly developed spectrogram–convolutional neural network (CNN) index was found
to outperform the commercialized surgical pleth index (SPI).
This
observational study focused on the development of a new analgesic index via
photoplethysmogram (PPG) spectrograms and CNN for estimating pain in conscious patients.
The PPGs were procured from a total of 100 surgical patients for 6 minutes both in the no pain (pre-surgically) and in the presence (post-surgically) of pain. PPG data of the later 5 minutes were used for examination.
A spectrogram–CNN
index as per the PPGs and a CNN was made to assess pain. Utilizing the ‘area
under the curve of the receiver-operating characteristic curve’ (AUC-ROC),
performance of the two indices was evaluated.
Mean spectrogram–CNN index value increased significantly in case of pain. The AUC was 0.76, and the balanced accuracy was 71.4%. With a sensitivity and specificity of 68.3% and 73.8%, the spectrogram–CNN index cutoff value for detection of pain was 48. The spectrogram–CNN index illustrated improved performance measures in terms of balanced accuracy, sensitivity, and particularly specificity, as shown in Figure 1:
CNN, a new analgesic index,
can proficiently detect pain after surgery in conscious patients. Future
studies are needed to assess the practicability of this index and avoid
over-fitting to several populations, together with patients under general
anesthesia.
Journal of Medical Internet Research
Novel Analgesic Index for Postoperative Pain Assessment Based on a Photoplethysmographic Spectrogram and Convolutional Neural Network: Observational Study
Byung-Moon Choi et al.
Comments (1)