Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Predicting Risk Factors of Pressure Injury for Perioperative Patients Through Machine Learning With SHapley Additive

Guirong Shi1,2, Xin Xu1, Peipei Zhang1

  • 1Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine, Shanghai, China.

Advances in Skin & Wound Care
|June 24, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Histone lactylation-driven intratumoral IGFBP2<sup>+</sup> NK cells promote tumor immune evasion.

Cell reports·2026
Same author

Dipeptidylpeptidase 4 inhibition attenuates gestational pathologies via immune homeostasis restoration in the pulmonary-uterine axis.

Nature communications·2026
Same author

Global, Regional, and National Incidence Trends of Pressure Injury From 1990 to 2021.

Journal of nursing scholarship : an official publication of Sigma Theta Tau International Honor Society of Nursing·2025
Same author

Characterization of Periplaneta americana glycoproteins' structures and their potential to treat immunological liver fibrosis.

International journal of biological macromolecules·2025
Same author

Investigating perioperative pressure injuries and factors influencing them with imbalanced samples using a Synthetic Minority Over-sampling Technique.

Bioscience trends·2025
Same author

Operative Positioning and Intraoperative-Acquired Pressure Injury: A Retrospective Cohort Study.

Advances in skin & wound care·2024

This study developed an interpretable machine learning model to predict perioperative pressure injuries (PIs). Key predictors include operation duration, age, and serum albumin, enabling early risk assessment.

Area of Science:

  • Medical Informatics
  • Machine Learning in Healthcare
  • Surgical Patient Safety

Background:

  • Perioperative pressure injuries (PIs) are a significant complication in surgical patients, leading to adverse health outcomes.
  • Early identification and prevention of PIs are crucial for improving patient recovery and reducing healthcare burdens.

Purpose of the Study:

  • To develop an interpretable machine learning (ML) model for early prediction of perioperative pressure injuries (PIs).
  • To identify key clinical factors contributing to the risk of developing PIs in surgical inpatients.

Main Methods:

  • Retrospective analysis of 14,416 surgical inpatients' clinical data from 2017-2023.
  • Feature selection using XGBoost and development of seven ML models for PI prediction.
  • Model performance evaluation using accuracy, recall, AUC, and 10-fold cross-validation; interpretability via SHapley Additive exPlanations (SHAP).
Keywords:
SHapley Additive exPlanationbig datafeature selectionhospital-acquired pressure injurymachine learningpressure injuries

Related Experiment Videos

Main Results:

  • The Random Forest and XGBoost models demonstrated excellent performance in predicting perioperative PIs.
  • SHAP analysis identified operation duration, age, serum albumin, body mass index, body temperature, and anesthesia grade as significant risk factors.
  • The SHAP framework provided clear insights into the importance of each predictive variable.

Conclusions:

  • The developed ML model effectively predicts perioperative PIs, offering a valuable tool for early risk assessment.
  • The integration of SHAP values enhances model interpretability, facilitating clinical understanding and application.