AI-Based LOS Prediction in Hip Fracture Patients

Kocaeli University

Status: Completed

Conditions: Hip Fractures

Study type: Observational

Funder types: Other

Identifiers

NCT06392048

GOKAEK-2024/06.22

Details and patient eligibility

About

With increasing life expectancy, the elderly population is growing. Hip fractures significantly increase morbidity and mortality, particularly within the first year, among elderly patients. Managing anesthesia in these elderly patients, who often have multiple comorbidities, is challenging. Identifying perioperative factors that can reduce mortality will benefit the perioperative management of these patients.

The aim of this study is to develop and validate a machine learning based model to predict the length of hospital stay for hip fracture patients after PACU. Different machine learning algorithms such as R language Gradient Boosting, Random Forest, Artificial Neural Networks and Logistic Regression will be used in the study and the best performing model will be determined. In addition, the prediction mechanism of the model will be examined with SHAP analysis and its applicability in clinical decision processes will be evaluated. Thus, by predicting the length of hospital stay, clinicians will be enabled to manage patient care processes more effectively.

Enrollment: 366 patients

Sex: All

Ages: 65 to 100 years old

Volunteers: Accepts Healthy Volunteers

Inclusion criteria

• Patients who underwent hip fracture surgery at our institution between 2017 and 2024
• Patients aged 65 years or older
• Patients with hip fractures resulting from a low-energy trauma (simple fall from standing height)

Exclusion criteria

• Patients with pathological hip fractures due to malignancy
• Cancer patients with multiple organ metastases
• Patients who underwent revision hip fracture surgery

Trial design

366 participants in 2 patient groups

> 7 days LOS

Description:

This cohort includes patients whose postoperative hospital length of stay exceeded 7 days. The group was formed based on the median LOS determined in the overall study population. No intervention was administered. The group is used for training and evaluating a machine learning model aimed at predicting prolonged hospitalization (\>7 days) based on preoperative and intraoperative clinical features.

<= 7 days LOS

Description:

This cohort includes patients whose postoperative hospital length of stay was 7 days or less. The grouping was based on the median LOS observed in the total sample to ensure balanced classification for the machine learning model. No intervention was administered. Clinical data were used to train and test an AI algorithm for hospital LOS prediction.