Using machine-learning to decode postoperative hip mortality Trends: Actionable insights from an extensive clinical...

Area of Science:

• Orthopedic Surgery
• Geriatric Medicine
• Health Services Research

Background:

• Hip fractures are a common injury with high postoperative mortality.
• Effective preoperative risk stratification is needed to improve patient outcomes.
• Identifying mortality predictors is essential for tailored surgical interventions.

Purpose of the Study:

• To identify preoperative risk factors for 30-day mortality after hip fracture surgery.
• To develop predictive models for hip fracture-related mortality.
• To utilize a large patient cohort for robust model development.

Main Methods:

• Utilized data from the American College of Surgeons National Surgical Quality Improvement Program database.
• Included 107,660 patients undergoing surgical fixation for hip fractures.
• Employed least absolute shrinkage and selection operator (LASSO) regression to build predictive models.

Main Results:

• Identified 68 preoperative factors associated with 30-day mortality.
• Developed two models: one pre-operative, one combined pre- and post-operative.
• The combined model demonstrated strong predictive power (AUC = 0.83).

Conclusions:

• The developed models serve as risk assessment tools for clinicians.
• Facilitates identification of high-risk hip fracture patients.
• Aims to optimize patient-specific care and improve surgical outcomes.