Predictive Model for In-Hospital Mortality in AMI Patients with Diabetes

Seven predictors, including heart rate and neutrophil count, were used to build machine learning models for mortality risk. The XGBoost model achieved the highest AUC in validation using data from the Affiliated Hospital of North Sichuan Medical College and the MIMIC-IV database. This research dataset, published by Lang Zeng in 2026, includes the model's feature set and SHAP-based interpretations.

Use Cases

• Developing mortality risk prediction models based on the seven identified clinical predictors.
• Benchmarking machine learning algorithms like XGBoost for clinical outcome prediction tasks.
• Applying SHAP analysis for model interpretability based on the described feature contributions.
• Conducting external validation studies on clinical models using the MIMIC-IV database.

Strengths

• Model was validated on both internal and external sets from two distinct data sources.
• Feature selection was performed using two distinct methods: LASSO regression and the Boruta algorithm.
• The final model's interpretability was enhanced using the SHAP method for visual analysis.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The 987.0 KB file size suggests a limited scope, likely containing model details rather than a large patient dataset.

Provenance

Source

Affiliated Hospital of North Sichuan Medical College and MIMIC-IV database.

Collection Method

Retrospective study using data from the listed medical institutions.

Freshness

Last updated 2026-06-01 05:25:15; freshness should be verified.