Table 1_Predicting long-term clinical mortality of elderly patients with vertebral compres

440 patients aged 65 and older diagnosed with vertebral compression fractures between 2017 and 2020 were used to develop a mortality prediction model. The dataset was split into a training set of 296 patients and a validation set of 144 patients. An XGBoost model achieved a C-index of 0.753, identifying age, sex, previous fracture, cancer history, and co-morbidity as top predictors.

Use Cases

• Training survival analysis models based on demographic and clinical variables mentioned in the description.
• Benchmarking machine learning algorithms like XGBoost and Random Survival Forests for clinical prediction tasks.
• Investigating feature importance for mortality risk using SHAP analysis on variables like age and co-morbidity.
• Validating clinical risk scores through Kaplan-Meier survival analysis and calibration curves.

Strengths

• Includes 440 patient records from a defined study period (2017-2020).
• Model validation was performed using a dedicated set of 144 patients.
• The predictive model achieved a reported C-index of 0.753.
• Key predictive variables, including age and cancer history, were identified via SHAP analysis.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data is from a single-center retrospective study, which may limit generalizability.

Provenance

Source

Shuofan Wang via figshare.

Collection Method

Single-center retrospective study.

Time Range

2017-2020

Freshness

Last updated 2026-04-21 04:21:48; freshness should be verified.