Clinical Data for Diabetic Microvascular Complications Risk Prediction

A study from 2026 by Min Xiao analyzes clinical and laboratory data from 1,498 patients to identify risk factors for diabetic microvascular complications. The research compares nine machine learning models, finding a Gradient Boosting Decision Tree (GBDT) model performed best. The dataset, shared on figshare, includes identified independent risk factors such as urea, fibrinogen, and D-dimer.

Use Cases

• Building a binary classification model for diabetic microvascular complications based on identified clinical risk factors.
• Comparing the performance of different machine learning algorithms (like GBDT) on clinical prediction tasks.
• Validating clinical utility of predictive models using metrics like AUC, sensitivity, and decision curve analysis mentioned in the description.

Strengths

• Data from 1,498 patients provides a substantive sample for analysis.
• Model performance was evaluated across training, validation, and test sets, suggesting rigorous validation.
• The study identifies nine specific independent risk factors (e.g., urea, fibrinogen, D-dimer) for the target condition.

Limitations

• Row count and column-level documentation are unknown, limiting suitability assessment.
• The primary data file is a 11.9 KB DOCX document, which is a very small size for a typical clinical dataset and may contain only summary results.
• Data may reflect bias inherent to the specific patient cohort studied.

Provenance

Source

figshare, author Min Xiao.

Collection Method

Clinical and laboratory data analysis from patient records, as described in the study methods.

Freshness

Last updated 2026-04-21 04:22:14.