Affiliation:
1. Department of Orthopaedic Surgery, Faculty of Medicine National University of Malaysia Bani Malaysia
2. Department of Immunology, School of Medicine Tehran University of Medical Sciences Tehran Iran
3. Burn and Regenerative Medicine Research Center Guilan University of Medical Sciences Rasht Iran
4. Student Research Committee, Department of Medical‐Surgical Nursing, School of Nursing and Midwifery Guilan University of Medical Sciences Rasht Iran
5. Center for Research and Training in Skin Diseases and Leprosy Tehran University of Medical Sciences Tehran Iran
6. Department of Medical Informatics, Faculty of Medicine Mashhad University of Medical Sciences Mashhad Iran
7. Student Research Committee Shahid Sadoughi University of Medical Sciences Yazd Iran
8. Department of Physiology, School of Medicine, Cellular and the Molecular Research Center Guilan University of Medical Science Rasht Iran
9. Department of Plastic & Reconstructive Surgery, School of Medicine Guilan University of Medical Sciences Rasht Iran
Abstract
AbstractPressure injury (PI), or local damage to soft tissues and skin caused by prolonged pressure, remains controversial in the medical world. Patients in intensive care units (ICUs) were frequently reported to suffer PIs, with a heavy burden on their life and expenditures. Machine learning (ML) is a Section of artificial intelligence (AI) that has emerged in nursing practice and is increasingly used for diagnosis, complications, prognosis, and recurrence prediction. This study aims to investigate hospital‐acquired PI (HAPI) risk predictions in ICU based on a ML algorithm by R programming language analysis. The former evidence was gathered through PRISMA guidelines. The logical analysis was applied via an R programming language. ML algorithms based on usage rate included logistic regression (LR), Random Forest (RF), Distributed tree (DT), Artificial neural networks (ANN), SVM (Support Vector Machine), Batch normalisation (BN), GB (Gradient Boosting), expectation–maximisation (EM), Adaptive Boosting (AdaBoost), and Extreme Gradient Boosting (XGBoost). Six cases were related to risk predictions of HAPI in the ICU based on an ML algorithm from seven obtained studies, and one study was associated with the Detection of PI risk. Also, the most estimated risksSerum Albumin, Lack of Activity, mechanical ventilation (MV), partial pressure of oxygen (PaO2), Surgery, Cardiovascular adequacy, ICU stay, Vasopressor, Consciousness, Skin integrity, Recovery Unit, insulin and oral antidiabetic (INS&OAD), Complete blood count (CBC), acute physiology and chronic health evaluation (APACHE) II score, Spontaneous bacterial peritonitis (SBP), Steroid, Demineralized Bone Matrix (DBM), Braden score, Faecal incontinence, Serum Creatinine (SCr) and age. In sum, HAPI prediction and PI risk detection are two significant areas for using ML in PI analysis. Also, the current data showed that the ML algorithm, including LR and RF, could be regarded as the practical platform for developing AI tools for diagnosing, prognosis, and treating PI in hospital units, especially ICU.