Table 2_Development and internal validation of an interpretable machine-learning model for

A retrospective two-center cohort study from January 2021 to October 2025, comprising 787 unique patient records with diabetic kidney disease. The dataset was used to develop and internally validate a machine-learning model for identifying comorbid atrial fibrillation. The model, created by Xiaoran Li, achieved an AUC of 0.927 on the test set and was last updated in May 2026.

Use Cases

• Training interpretable machine-learning models for disease comorbidity identification based on clinical and echocardiographic variables.
• Benchmarking model performance (e.g., AUC, sensitivity) for atrial fibrillation detection in a specific patient population.
• Conducting feature importance analysis (e.g., using SHAP) to understand clinical drivers like proteinuria or renal dysfunction.

Strengths

• Model performance is quantitatively reported, with a test set AUC of 0.927 and accuracy of 0.886.
• The dataset underpinning the model includes 787 unique patient records from a two-center study.
• Feature selection was performed using LASSO regression, resulting in a focused set of 14 clinical predictors.

Limitations

• Row count for the underlying clinical data is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The model has undergone internal validation only; external validation is noted as required.

Provenance

Source

figshare, author Xiaoran Li.

Collection Method

Retrospective two-center cohort study using routinely collected clinical data.

Time Range

January 2021 to October 2025.

Freshness

Last updated 2026-05-26 04:39:12.