Related Experiment Video
Updated: Jun 28, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.8K
Development and Internal Validation of a Multivariable Prediction Model for Mortality After Hip Fracture with Machine
Mathias Mosfeldt1,2, Henrik Løvendahl Jørgensen3,4, Jes Bruun Lauritzen4,5
1Department of Orthopaedics, Karolinska University Hospital, Stockholm, Sweden. mathias.mosfeldt@ki.se.
Calcified Tissue International
|April 16, 2024
Summary
Machine learning models accurately predict mortality risk in hip fracture patients using preoperative data. The extreme gradient boosting (XGB) model shows the most promise for improving patient care decisions.
Area of Science:
- Gerontology
- Medical Informatics
- Orthopedic Surgery
Background:
- Hip fractures are a significant cause of morbidity and mortality in older adults.
- Accurate prediction of mortality risk is crucial for perioperative decision-making and patient management.
Purpose of the Study:
- To estimate the likelihood of 1, 3, 6, and 12-month mortality in hip fracture patients.
- To evaluate the performance of machine learning models using readily available preoperative data.
Main Methods:
- Utilized prospectively collected preoperative biochemical and anamnestic data from 1186 hip fracture patients (aged 60+).
- Applied and compared Random Forest, extreme gradient boosting (XGB), and Generalized Linear Models after feature selection.
- Evaluated models using ROC curves, calibration metrics, and Decision Curve Analysis.
Main Results:
- Machine learning models achieved high accuracy in predicting mortality, with Area Under the Curve (AUC) ranging from 0.79 to 0.81.
- The XGB model demonstrated superior calibration and predictive performance for postoperative mortality.
- Combinations of 10-13 preoperative parameters effectively estimated mortality likelihood.
Conclusions:
- Machine learning, particularly XGB, can accurately estimate hip fracture patient mortality using preoperative data.
- These models can aid in perioperative decisions, research, and patient counseling.
- External validation is ongoing, with an online tool developed for educational purposes.
