TY - JOUR
T1 - External Validation of an Algorithm to Guide Opioid Administration at the End of Surgery-Protocol for an Observational Cohort Study of the OPIAID Algorithm
AU - Karlsen, Anders Peder Højer
AU - Sunde, Pernille Bjersand
AU - Rasmussen, Ida Houtved
AU - Folkersen, Caroline
AU - Tran, Trang Xuan Minh
AU - Nguyen, Markus Kien Trung
AU - Elkjær, Line Maria
AU - Lunn, Troels Haxholdt
AU - Meyhoff, Christian S
AU - Andersen, Jonas Valbjørn
AU - Mathiesen, Ole
AU - Olsen, Markus Harboe
N1 - © 2025 The Author(s). Acta Anaesthesiologica Scandinavica published by John Wiley & Sons Ltd on behalf of Acta Anaesthesiologica Scandinavica Foundation.
PY - 2025/7
Y1 - 2025/7
N2 - BACKGROUND: Despite advances in pain management, inadequate pain relief and opioid-related adverse events remain common challenges in perioperative care, often contributing to prolonged recovery and reduced quality of life. The perioperative opioid algorithms for individualized dosing (OPIAID) project aims to develop machine-learning algorithms tailored to provide patient-specific opioid dosing across the different phases of perioperative care. For each phase, eight models are trained on granular data from 1.1 million surgical procedures, including demographic and surgical details, vital signs, administered analgesics, pain, and opioid-related adverse events. The two most accurate models will proceed to external validation. The best-performing model will subsequently be tested as a decision support against current standard of care.OBJECTIVES: This protocol describes the design and external validation of the intraoperative OPIAID algorithm, which suggests the end-of-surgery opioid dose intended for postoperative analgesia by approximating clinical performance and evaluating reliability, agreement, and calibration.METHODS: In this multicenter, TRIPOD+AI-adherent, prospective observational cohort study, we will collect data from a diverse surgical population of 656 adult patients undergoing elective or acute surgery under general anesthesia. All patients will require intraoperative opioid administration at the end of surgery for postoperative pain management and a subsequent stay in the post-anesthesia care unit. The cohort will be used to externally validate two machine-learning models through standardized measures of reliability, agreement, and calibration, and thereby designate the intraoperative OPIAID algorithm. Subsequently, the cohort will be used to approximate the clinical efficacy, safety and overall performance of the intraoperative OPIAID algorithm's recommended doses versus the clinician-administered doses. These comparisons will be based on each approach's proximity to a golden standard "optimal dose," which is calculated based on a predefined generic ruleset incorporating intraoperative opioid dosing, postoperative pain, opioid-related adverse events, and need for rescue opioid administrations.CONCLUSION: The intraoperative OPIAID algorithm is intended as a clinical decision aid for anesthesiologists and nurse anesthetists in providing adequate postoperative pain management.
AB - BACKGROUND: Despite advances in pain management, inadequate pain relief and opioid-related adverse events remain common challenges in perioperative care, often contributing to prolonged recovery and reduced quality of life. The perioperative opioid algorithms for individualized dosing (OPIAID) project aims to develop machine-learning algorithms tailored to provide patient-specific opioid dosing across the different phases of perioperative care. For each phase, eight models are trained on granular data from 1.1 million surgical procedures, including demographic and surgical details, vital signs, administered analgesics, pain, and opioid-related adverse events. The two most accurate models will proceed to external validation. The best-performing model will subsequently be tested as a decision support against current standard of care.OBJECTIVES: This protocol describes the design and external validation of the intraoperative OPIAID algorithm, which suggests the end-of-surgery opioid dose intended for postoperative analgesia by approximating clinical performance and evaluating reliability, agreement, and calibration.METHODS: In this multicenter, TRIPOD+AI-adherent, prospective observational cohort study, we will collect data from a diverse surgical population of 656 adult patients undergoing elective or acute surgery under general anesthesia. All patients will require intraoperative opioid administration at the end of surgery for postoperative pain management and a subsequent stay in the post-anesthesia care unit. The cohort will be used to externally validate two machine-learning models through standardized measures of reliability, agreement, and calibration, and thereby designate the intraoperative OPIAID algorithm. Subsequently, the cohort will be used to approximate the clinical efficacy, safety and overall performance of the intraoperative OPIAID algorithm's recommended doses versus the clinician-administered doses. These comparisons will be based on each approach's proximity to a golden standard "optimal dose," which is calculated based on a predefined generic ruleset incorporating intraoperative opioid dosing, postoperative pain, opioid-related adverse events, and need for rescue opioid administrations.CONCLUSION: The intraoperative OPIAID algorithm is intended as a clinical decision aid for anesthesiologists and nurse anesthetists in providing adequate postoperative pain management.
KW - Humans
KW - Analgesics, Opioid/administration & dosage
KW - Algorithms
KW - Pain, Postoperative/drug therapy
KW - Cohort Studies
KW - Prospective Studies
KW - Machine Learning
KW - Pain Management/methods
KW - Adult
KW - Reproducibility of Results
KW - Observational Studies as Topic
KW - Perioperative Care/methods
UR - http://www.scopus.com/inward/record.url?scp=105008896957&partnerID=8YFLogxK
U2 - 10.1111/aas.70071
DO - 10.1111/aas.70071
M3 - Journal article
C2 - 40550516
SN - 0001-5172
VL - 69
SP - e70071
JO - Acta Anaesthesiologica Scandinavica
JF - Acta Anaesthesiologica Scandinavica
IS - 6
M1 - e70071
ER -