대한신장학회

Skip Navigation: Skip to contents

KOR
ENG
전체메뉴보기

간행물 검색

현재 페이지 경로

HOME
간행물
간행물 검색

논문분류	춘계학술대회 초록집
제목	A Deep-Learning Algorithm Using Real-Time Collected Intraoperative Vital Sign Signals for Predicting Acute Kidney Injury After Major Non-Cardiac Surgeries: A Modelling Study
저자	Soomin Chung
출판정보	2025; 2025(1):
키워드	postoperative acute kidney injury, intraoperative vital sign, deep learning
초록	Postoperative acute kidney injury (PO-AKI) is a significant complication in non-cardiac major surgeries. Existing predictive models primarily rely on preoperative clinical data, overlooking intraoperative factors critical to PO-AKI development. This study aimed to improve PO-AKI prediction by integrating deep-learning analysis of intraoperative vital signs with preoperative clinical data. Using data from three hospitals (n=110,696), we developed a deep-learning model (DL-IVSS) using a convolutional neural network-based EfficientNet framework to analyze minute-scale intraoperative vital sign data (blood pressure, heart rate). We tested the model alone and in combination with preoperative variables, including an ensemble model incorporating 103 baseline variables. The development cohort included 51,345 patients with 59,351 in external validation cohorts. Model performance was compared with the conventional SPARK model. The DL-IVSS model using only intraoperative vital signs demonstrated comparable predictive power (AUROC: discovery cohort 0.707, validation cohorts 0.637 and 0.607) to preoperative-only models. Adding 11 essential clinical variables (age, sex, eGFR, albuminuria, hyponatremia, hypoalbuminemia, anemia, diabetes, RAAS inhibitors, emergency surgery, and estimated surgery time) significantly improved performance (AUROC: discovery cohort 0.765, validation cohorts 0.716 and 0.761). The ensemble model integrating both preoperative and intraoperative data achieved superior predictive accuracy (AUROC: discovery cohort 0.795, validation cohorts 0.762 and 0.786). Integrating intraoperative vital signs into PO-AKI prediction models enhances predictive performance, providing clinicians with more accurate risk assessments. Our findings highlight the importance of real-time intraoperative monitoring in mitigating PO-AKI risk through timely interventions, though further validation in prospective and multi-national settings is warranted.
원문(PDF)	PDF 원문보기

목록

위로가기