Artificial Intelligence to Evaluate Postoperative Pain Based on Facial Expression

Artificial Intelligence to Evaluate Postoperative Pain Based on Facial Expression Recognition and Analgesia Nociception Index

Patients' subjective complaints about pain intensity are difficult to objectively evaluate, and may lead to inadequate pain management, especially in patients with communication difficulties.

Analgesia nociception index (ANI 0-100) and patient-reported numeric rating scale (NRS 0-10) were trained on a convolutional neural network (CNN) model by linking the patients' facial expression with the score. By applying the predicted pain score by the AI model to evaluate pain, it is intended to measure the intensity of pain in an automatic, fast, and objective way for appropriate pain management.

Inclusion Criteria:

• Patients aged 19-75 years who were scheduled for elective laparoscopic abdominal surgery under general anesthesia
• American Society of Anesthesiology (ASA) class I-II

Exclusion Criteria:

• Patients who have difficulty in communicating and reporting pain
• Underlying diseases: liver, kidney, brain
• Patients with BMI greater than 30 and less than 18.5
• Alcohol or drug dependent patients
• Patients with severe or acute respiratory failure
• Opioid, NSAID allergy
• Patients who are scheduled to be admitted to the intensive care unit after surgery
• Patients who undergo cooperative surgery