Machine Learning-Based Prediction of Postoperative Pneumonia Among Super-Aged Patients With Hip Fracture DOI Creative Commons

Miaotian Tang,

Meng Zhang, Yu Dang

и другие.

Clinical Interventions in Aging, Год журнала: 2025, Номер Volume 20, С. 217 - 230

Опубликована: Фев. 1, 2025

Hip fractures have become a significant health concern, particularly among super-aged patients, who were at high risk of postoperative pneumonia due to their frailty and the presence multiple comorbidities. This study aims establish validate model predict patients with hip fracture. Data derived from Chinese PLA General Hospital (PLAGH) Fracture Cohort Study, we included 555 (≧80 years old) fracture treated surgery. Patient's demographics, comorbidities, laboratory tests, surgery types collected for analysis. All randomly splitting into training group validation according ratio 7:3. The majority used train models, which was tuned using series algorithms, including decision tree (DT), random forest (RF), extreme gradient boosting machine (eXGBM), support vector (SVM), neural network (NN), logistic regression (LR). incidence 7.2% (40/555). Among six developed eXGBM demonstrated optimal model, area under curve (AUC) value 0.929 (95% CI: 0.900-0.959), followed by RF (AUC: 0.916, 95% 0.885-0.948). LR had an AUC 0.720 0.662-0.778). In addition, prediction performance in terms accuracy (0.858), precision (0.870), F1 score (0.855), Brier (0.104), log loss (0.349). It also showed favorable calibration ability clinical net benefits across various threshold risk. develops validates reliable learning-based specifically following can serve as useful tool identify guide strategies

Язык: Английский

Prediction Models for Postoperative Pneumonia in Elderly Hip Fracture Patients: A Systematic Review and Critical Appraisal DOI Open Access
Zhiqiang He, Geyu Zhong, Wenjin Han

и другие.

Journal of Clinical Nursing, Год журнала: 2025, Номер unknown

Опубликована: Янв. 14, 2025

ABSTRACT Background Although several models have been developed to predict postoperative pneumonia in elderly hip fracture patients, no systematic review of the model quality and clinical applicability has reported. Objective To systematically critically appraise existing for patients. Design Systematic meta‐analysis. Methods 10 databases were searched from inception April 15, 2024, updated on August 26. Two reviewers independently performed literature selection, information extraction assessment. A narrative synthesis was employed summarise characteristics models. Meta‐analysis using Stata 17.0. Results 13 studies containing 25 included. The prevalence 9.62% (95% CI: 7.62%–11.62%). Age (53.8%), hypoproteinemia (46.2%), chronic obstructive pulmonary disease (COPD, 30.8%), gender (30.8%), activity daily living score (ADL, 30.8%) American Society Anesthesiologists (ASA, top six predictors. All reported area under curve (AUC: 0.617–0.996). 9 (69.2%) used Hosmer‐Lemeshow (H‐L) test, calibration curves, or Brier scores evaluate calibration. 5 (38.5%) internal validation, 4 (30.8%) external validation. had a high risk bias due single sample source, inappropriate data processing, inadequate evaluation, negligence (76.9%) good applicability. Conclusions Prediction patients are still developing stage. validation evaluation poor. Future should focus robust updating. Additionally, Transparent Reporting Multivariable Model Individual Prognosis Diagnosis + artificial intelligence (TRIPOD+AI) statement be followed. Relevance Clinical Practice effective discriminating but further adjustment warranted.

Язык: Английский

Процитировано

1

Machine Learning-Based Prediction of Postoperative Pneumonia Among Super-Aged Patients With Hip Fracture DOI Creative Commons

Miaotian Tang,

Meng Zhang, Yu Dang

и другие.

Clinical Interventions in Aging, Год журнала: 2025, Номер Volume 20, С. 217 - 230

Опубликована: Фев. 1, 2025

Hip fractures have become a significant health concern, particularly among super-aged patients, who were at high risk of postoperative pneumonia due to their frailty and the presence multiple comorbidities. This study aims establish validate model predict patients with hip fracture. Data derived from Chinese PLA General Hospital (PLAGH) Fracture Cohort Study, we included 555 (≧80 years old) fracture treated surgery. Patient's demographics, comorbidities, laboratory tests, surgery types collected for analysis. All randomly splitting into training group validation according ratio 7:3. The majority used train models, which was tuned using series algorithms, including decision tree (DT), random forest (RF), extreme gradient boosting machine (eXGBM), support vector (SVM), neural network (NN), logistic regression (LR). incidence 7.2% (40/555). Among six developed eXGBM demonstrated optimal model, area under curve (AUC) value 0.929 (95% CI: 0.900-0.959), followed by RF (AUC: 0.916, 95% 0.885-0.948). LR had an AUC 0.720 0.662-0.778). In addition, prediction performance in terms accuracy (0.858), precision (0.870), F1 score (0.855), Brier (0.104), log loss (0.349). It also showed favorable calibration ability clinical net benefits across various threshold risk. develops validates reliable learning-based specifically following can serve as useful tool identify guide strategies

Язык: Английский

Процитировано

0