Chronic Nonhealing Wound Infection Prediction Model Data from Two Chinese Hospitals

A dataset supporting a predictive model for infection risk in chronic nonhealing wounds, developed using routine electronic medical record data. The study involved a primary cohort of 500 patients split into training and testing sets and an external validation cohort of 300 patients from two tertiary hospitals in China. The model was developed by Yang Jiang and published on figshare in May 2026.

Use Cases

• Training infection risk prediction models based on the six consensus predictors identified in the study.
• Benchmarking machine learning algorithms for clinical prediction tasks using the described cohorts.
• Studying the clinical utility of predictive models through decision curve analysis as performed in the study.
• Exploring model interpretability for healthcare applications using SHAP analysis as referenced.

Strengths

• Model achieved AUROC of 0.884 in testing and 0.855 in external validation, indicating strong predictive performance.
• The study used a structured methodology with a primary cohort of 500 patients and an external validation cohort of 300 patients.
• Model is based on six readily available clinical predictors, enhancing its potential for pragmatic application.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic bias inherent to the two Chinese hospital sources.

Provenance

Source

Electronic medical record data from two tertiary hospitals in China.

Collection Method

Two-center retrospective study with external validation.

Freshness

Last updated 2026-05-19 05:44:10; freshness should be verified.

Geography

China