Clinical Predictors for Diabetic Foot Ulcer Screening in 2,532 Patients

1,854 hospitalized patients from 2012-2020 and 678 outpatients from 2021-2025 were analyzed to develop a pre-screening model for diabetic foot ulcer (DFU). The dataset, created by Weidi Wang and shared under CC-BY-4.0, includes demographic, lifestyle, and laboratory parameters used to identify four key predictors. The model achieved an area under the curve (AUC) of 0.807 in the training set and 0.817 in the validation set.

Use Cases

• Developing a simplified pre-screening tool for diabetic foot ulcers based on age, injury history, alcohol consumption, and the Alb/HbA1c ratio.
• Validating a logistic regression model for clinical risk prediction using a geographically distinct outpatient validation set.
• Analyzing the interaction between nutritional status (via Alb/HbA1c ratio) and age as synergistic risk factors for DFU.
• Creating risk heatmaps or nomograms to visualize patient risk stratification for clinical decision support.

Strengths

• Model performance is validated on a distinct outpatient cohort of 678 patients.
• Four independent predictors are identified with reported odds ratios, such as a history of injury (OR = 7.57).
• The dataset is derived from a substantial training set of 1,854 hospitalized patients.

Limitations

• Row count and column-level documentation are unknown, which may limit suitability assessment.
• Data may reflect geographic and temporal bias inherent to a single hospital system.
• The dataset is very small (12.9 KB), indicating limited scope, likely containing model results rather than raw patient records.

Provenance

Source

figshare, author Weidi Wang.

Collection Method

Retrospective analysis of hospitalized patients and outpatients from a hospital system.

Time Range

Training set: January 2012–December 2020; Validation set: January 2021–December 2025.

Freshness

Last updated 2026-05-29 04:47:39; freshness should be verified.

Geography

Patients from a geographically distinct branch of a hospital system; specific location not stated.