Evaluation of Perioperative Cardiovascular Event Risk in Gastrointestinal Surgery ― Predictive Modeling and Risk Stratification Using Machine Learning ―
Abstract
BACKGROUND: Preoperative risk assessment is very important to ensure surgical safety and predict postoperative complications. However, no large-scale studies have evaluated the risk of perioperative cardiovascular events in Japan. This study evaluated perioperative cardiovascular events using real-world data. In addition, the applicability of machine learning to risk stratification was examined to develop a predictive model for perioperative cardiovascular events. METHODS AND RESULTS: This was an observational cohort study using the Japan Medical Data Center database, which includes claim and health examination data in Japan, between January 2005 and April 2021. In all, 133,634 gastrointestinal surgeries were included in the analysis. The primary outcome was 30-day risk of major adverse cardiovascular events (MACE). The 30-day MACE incidence rate following surgery was 3.8%. Machine learning was used to perform a binary classification task to predict MACE occurrence within 30 days after surgery. A clustering algorithm was developed based on the Shapley additive explanation values obtained from training data, and generalizability was evaluated using test data. Of the variables, age, history of ischemic heart disease or heart failure, history of stroke, diabetes, hypertension, atrial fibrillation, cases of malignancy, and pancreatic biliary surgery were identified as factors associated with MACE occurrence. CONCLUSIONS: A machine learning model built from basic clinical information, comorbidities, and surgical information demonstrated the capacity to stratify MACE risk in patients undergoing gastrointestinal surgery.
Related
- Precision risk stratification for perioperative safety of minimally invasive versus open TLIF in elderly patients with lumbar degenerative disease
- Perioperative nutrition-related laboratory profiles predict outcomes after digestive procedures in patients with gastrointestinal cancer
- Prognostic value of pulmonary artery pulsatility index for perioperative adverse cardiovascular events in patients undergoing surgery for mitral regurgitation
- Perioperative diabetes related complications in patients undergoing major surgery are associated with the duration of gastrointestinal and orthopaedic surgery.
- Artificial intelligence and machine learning for prediction of postoperative complications: A systematic review focused on anesthesiology, perioperative risk stratification, and clinical applicability.
BibTeX
@article{ito2026evaluation,
title = {{Evaluation of Perioperative Cardiovascular Event Risk in Gastrointestinal Surgery ― Predictive Modeling and Risk Stratification Using Machine Learning ―}},
author = {Hiromasa Ito and Tomohisa Seki and Yoshimasa Kawazoe and Toru Takiguchi and Yu Akagi and Kazumi Kubota and Kana Miyake and Masafumi Okada and Kazuhiko Ohe},
journal = {Circulation Journal},
year = {2026},
month = aug,
doi = {10.1253/circj.cj-25-0032},
url = {https://www.jstage.jst.go.jp/article/circj/90/9/90_CJ-25-0032/_article},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - Evaluation of Perioperative Cardiovascular Event Risk in Gastrointestinal Surgery ― Predictive Modeling and Risk Stratification Using Machine Learning ―
AU - Hiromasa Ito
AU - Tomohisa Seki
AU - Yoshimasa Kawazoe
AU - Toru Takiguchi
AU - Yu Akagi
AU - Kazumi Kubota
AU - Kana Miyake
AU - Masafumi Okada
AU - Kazuhiko Ohe
PY - 2026
DA - 2026/08/25
JO - Circulation Journal
DO - 10.1253/circj.cj-25-0032
UR - https://www.jstage.jst.go.jp/article/circj/90/9/90_CJ-25-0032/_article
AB - BACKGROUND: Preoperative risk assessment is very important to ensure surgical safety and predict postoperative complications. However, no large-scale studies have evaluated the risk of perioperative cardiovascular events in Japan. This study evaluated perioperative cardiovascular events using real-world data. In addition, the applicability of machine learning to risk stratification was examined to develop a predictive model for perioperative cardiovascular events. METHODS AND RESULTS: This was an observational cohort study using the Japan Medical Data Center database, which includes claim and health examination data in Japan, between January 2005 and April 2021. In all, 133,634 gastrointestinal surgeries were included in the analysis. The primary outcome was 30-day risk of major adverse cardiovascular events (MACE). The 30-day MACE incidence rate following surgery was 3.8%. Machine learning was used to perform a binary classification task to predict MACE occurrence within 30 days after surgery. A clustering algorithm was developed based on the Shapley additive explanation values obtained from training data, and generalizability was evaluated using test data. Of the variables, age, history of ischemic heart disease or heart failure, history of stroke, diabetes, hypertension, atrial fibrillation, cases of malignancy, and pancreatic biliary surgery were identified as factors associated with MACE occurrence. CONCLUSIONS: A machine learning model built from basic clinical information, comorbidities, and surgical information demonstrated the capacity to stratify MACE risk in patients undergoing gastrointestinal surgery.
SN - 1347-4820
SN - 1346-9843
ER -