[1L5-OS-18b] 生体信号を活用した医療・ヘルスケアAI

[1L5-OS-18b-02] AI model for predicting postoperative pain exacerbation using a wearable electrocardiogram sensor and an intravenous patient-controlled analgesia device

〇Toshiyuki Nakanishi^1,2, Koichi Fujiwara², Yoshiki Sento¹, Kazuya Sobue¹ (1. Nagoya City University, 2. Nagoya University)

Keywords:Postoperative pain, Wearable electrocardiogram sensor, Anomaly detection

There is a need to develop objective and real-time postoperative pain assessment methods in perioperative medicine. Few studies have evaluated the relationship between pain severity and temporal changes of physiological signals in actual postoperative patients. The aim of the study was to evaluate postoperative pain continuously and to predict pain exacerbation in real-time. We focused on intravenous patient-controlled analgesia (IV-PCA), a common analgesic modality utilized in post-surgical patients. We chose an electrocardiogram (ECG) as a feature to detect pain exacerbation. We developed a machine learning model which was trained from IV-PCA records and ECG of postoperative patients to predict pain exacerbation. A self-attentive autoencoder (SA-AE) model achieved 54% of sensitivity and a 1.76 times/h of false positive rate. In summary, we propose a novel pain evaluation method using an IV-PCA device. According to the current findings, ECG features may be used to predict postoperative pain exacerbation in real-time.