Machine Learning-driven Probability Calculators Can Accurately Predict 1-year Mortality After Proximal Humerus
Stijn R J Mennes1,2,3, Sebastian Engbers4, Bjarty L Garcia5,6
1Shoulder and Elbow Unit, Department of Orthopaedic and Trauma Surgery, OLVG Hospital, Amsterdam, the Netherlands.
Clinical Orthopaedics and Related Research
|January 21, 2026
Summary
Machine learning accurately predicts 1-year mortality risk for proximal humerus fractures (PHFs) in older adults. This tool aids shared decision-making for treatment options, improving patient understanding and informed consent.
Area of Science:
- Orthopedic Surgery
- Geriatric Trauma
- Machine Learning in Medicine
Background:
- Proximal humerus fractures (PHFs) in patients aged 65 and older carry a significant risk of mortality.
- Treatment decisions for PHFs in this demographic are complex due to high complication and reoperation rates with surgical intervention.
- Accurate mortality prediction is crucial for informed shared decision-making between surgeons and patients.
Purpose of the Study:
- To develop and externally validate machine learning (ML) algorithms for predicting 1-year mortality in patients aged 65+ with PHFs.
- To create a user-friendly online calculator for point-of-care risk assessment.
- To enhance informed consent and shared decision-making processes for PHF treatment.
Main Methods:
- Developed and validated four ML algorithms (logistic regression, XGBoost, random forest, LightGBM) using data from 2999 patients (≥65 years) with first-time PHFs.
- Trained and internally validated models on data from one hospital and externally validated on a geographically distinct patient cohort.
- Assessed model performance using discrimination (c-statistic) and calibration curves, and Brier scores.
Main Results:
- ML algorithms demonstrated strong performance with c-statistics ranging from 0.80-0.81 (internal) and 0.83-0.85 (external validation).
- Logistic regression was selected for its adequate calibration, interpretability, and strong negative predictive value (0.91).
- Key mortality predictors identified include hemiplegia, pre-fracture institutional residence, and heart failure.
Conclusions:
- An externally validated ML-driven prediction model accurately estimates 1-year mortality risk for individual patients with PHFs.
- This prognostic tool supports shared decision-making, improving patient and family expectations regarding treatment options.
- A freely accessible web application (https://bjarty.shinyapps.io/mortality_app/) is available for clinical use.
Related Concept Videos
01:49Probability Laws
44.0K
Overview
44.0K
02:27Predicting Products: SN1 vs. SN2
15.9K
Nucleophilic substitution reactions of alkyl halides can proceed via an SN1 or an SN2 mechanism. While in SN2 reactions, the nucleophile attacks the substrate simultaneously as the leaving group departs, in SN1 reactions, the substrate first dissociates to give the carbocation intermediate. Various factors such as the structure of the substrate, the strength of the nucleophile, and the nature of the solvent promote one mechanism over the other.
With increased substitution on the alkyl halide,...
With increased substitution on the alkyl halide,...
15.9K
01:27Pharmacodynamics in Geriatric Patients: Effects of Age
192
Age-related pharmacokinetic changes are extensively documented, but understanding age-related pharmacodynamic alterations is relatively limited. This knowledge gap can be partly attributed to the complexity of developing appropriate measures of drug responses compared to bioanalytical methods for determining drug concentrations.Most information regarding age-related differences in human pharmacodynamics originates from cross-sectional studies. However, these studies assume that observed mean...
192
01:22Pharmacokinetics in Geriatric Patients: Effect of Age on Drug Absorption
231
As individuals age, their body's physiology evolves, affecting drug pharmacokinetics. The most apparent changes occur in the gastrointestinal tract, where an increase in gastric pH, a delay in gastric emptying, and a reduction in gastrointestinal motility are observed. Remarkably, these changes do not substantially modify the absorption of orally administered drugs, particularly those absorbed via passive diffusion.Transdermal drug delivery emerges as a highly viable method for older adults due...
231
01:19Bones of the Upper Limb: Humerus
7.0K
The upper limb consists of the arm, forearm, wrist, and hand bones. The humerus is the single bone of the upper arm region. Proximally, it has a large, spherical, smooth head that articulates with the glenoid cavity of the scapula to form the glenohumeral or shoulder joint. The margin of the head is the anatomical neck, a residual epiphyseal plate. Laterally it extends to form bony projections called the greater tubercle and the lesser tubercle. Next to the tubercles is the surgical neck, a...
7.0K
01:00Pharmacokinetics in Geriatric Patients: Effect of Age on Drug Distribution
235
Drug distribution in the human body is influenced by several factors, including plasma protein concentration, body composition, blood flow, tissue-protein concentration, and tissue fluid pH. Among these, changes in plasma protein concentration and body composition due to aging significantly affect how drugs are distributed within the body. Specifically, aging is associated with a decrease in albumin levels by about 10% and an increase in α1-acid glycoprotein levels. These alterations are...
235