Predicting One-Year Mortality in Elderly Hip Fracture Patients Using Clinical Data

A retrospective cohort study of 2,604 elderly patients undergoing urgent hip fracture surgery at Sheba Medical Center, Israel, between January 2017 and November 2023. Adi Shuchami published the dataset on figshare in 2026. It compares manual and automated machine learning models for predicting one-year all-cause mortality.

Use Cases

• Benchmarking automated ML pipelines against manually optimized models based on the described clinical prediction task.
• Developing mortality risk stratification tools based on clinical, demographic, perioperative, and laboratory variables mentioned in the description.
• Studying feature importance for mortality prediction based on identified predictors like serum albumin, urea levels, age, intraoperative hypothermia, and chronic disease count.

Strengths

• Contains 2,604 patient records, providing a substantial cohort for analysis.
• Models were rigorously validated using stratified 5-fold cross-validation and techniques like SMOTE to address class imbalance.
• Includes detailed performance metrics (AUC, accuracy, F1-score) and interpretability analyses (permutation importance, SHAP) for model evaluation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic and temporal bias inherent to a single medical center in Israel between 2017 and 2023.

Provenance

Source

Sheba Medical Center, Israel

Collection Method

Retrospective cohort study

Time Range

January 2017 to November 2023

Freshness

Last updated 2026-06-01 04:17:40; freshness should be verified.

Geography

Israel