Prediction of Pressure Injury Risk in the Intensive Care Unit: A Comparative Analysis of Data Mining Algorithms in a Single-Center Retrospective Cohort
Trial Snapshot
- Phase
- Not Applicable
- Status
- Active, not recruiting
- Enrollment
- 500
- Locations
- 2
- Primary Endpoint
- Prediction accuracy of pressure injury development
Study Overview
Brief Summary
This study is a retrospective record review conducted among adult patients hospitalized in the intensive care unit of a tertiary hospital between October 10, 2020, and October 10, 2025. The aim of the study is to predict the risk of pressure injury development using demographic, clinical, laboratory, and nursing care-related variables by applying multiple data mining algorithms. No intervention, treatment, or patient contact will occur. All data will be extracted from existing electronic and paper-based medical records and will be fully anonymized prior to analysis. The study poses no risk to participants and will be conducted with approval from the institutional review board or ethics committee.
Detailed Description
This observational study uses a retrospective cohort design to analyze the clinical, demographic, laboratory, and nursing documentation records of adult intensive care unit (ICU) patients hospitalized between October 10, 2020, and October 10, 2025. The purpose of the study is to identify factors associated with the development of pressure injury and to compare the predictive performance of multiple data mining and machine learning algorithms, including logistic regression, decision trees, random forest, support vector machines, and gradient boosting models.
Data collection will involve reviewing archived ICU records, patient files, and nursing observation forms. No new data will be collected directly from patients, and no medical interventions or prospective follow-up will be performed. All extracted data will be fully anonymized prior to analysis. The study will be conducted in accordance with ethical principles and has been approved by the Bolu Abant Izzet Baysal University Non-Interventional Clinical Research Ethics Committee.
The expected outcome of this study is to identify the most accurate predictive model for pressure injury risk and to support clinical decision-making processes by contributing to early prevention strategies in the ICU.
Study Design
- Study Type
- Observational
- Observational Model
- Cohort
- Time Perspective
- Retrospective
Eligibility Criteria
- Ages
- 18 Years to — (Adult, Older Adult)
- Sex
- All
- Accepts Healthy Volunteers
- No
Inclusion Criteria
- •Patients hospitalized in the intensive care unit between October 10, 2020, and * October 10, 2025
- •Adult patients aged 18 years and older
- •Length of intensive care unit stay of at least 24 hours
- •Availability of complete and accessible electronic or paper-based medical records
Exclusion Criteria
- •Patients younger than 18 years
- •Intensive care unit stay shorter than 24 hours Incomplete, missing, or inconsistent electronic or paper-based medical records
- •Presence of a pressure injury diagnosed before or at the time of intensive care unit admission
- •Inability to extract pressure injury-related data from medical records
Arms & Interventions
ICU Patient Cohort
Adult patients who will be hospitalized in the intensive care unit between October 10, 2020 and October 10, 2025. No interventions will be applied, and all data will be obtained from existing medical records.
Intervention: No intervention (Other)
Outcomes
Primary Outcomes
Prediction accuracy of pressure injury development
Time Frame: 10 October 2020 to 10 October 2025
The primary outcome is the predictive accuracy of data mining algorithms in identifying the risk of developing pressure injury among patients hospitalized in the intensive care unit (ICU). Accuracy metrics such as the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, precision, recall, and F1-score will be calculated using retrospective medical record data.
Secondary Outcomes
No secondary outcomes reported
Investigators
Saadet Can Çiçek
Associate Professor
Abant Izzet Baysal University