Preprocedural Prediction Model for Failed Umbilical Venous Catheterization in Neonates

A retrospective cohort study of 123 neonates undergoing umbilical venous catheterization (UVC) between April 2023 and May 2025. The dataset was used to develop and internally validate an Elastic Net penalized logistic regression model predicting UVC failure based on preprocedural clinical and laboratory indicators. The model was created by Zhenzhen Ye and shared under a CC-BY-4.0 license.

Use Cases

• Developing risk stratification tools for neonatal catheterization based on preprocedural indicators like birth weight and blood gas parameters.
• Validating predictive models for medical device failure using internal cross-validation techniques.
• Analyzing associations between perinatal characteristics, such as Apgar scores and coagulation indices, and procedural outcomes.

Strengths

• Model performance is quantified with a high mean AUC of 0.9999 and a mean Brier score of 0.0062 from internal validation.
• The study includes 123 neonates, with a clear outcome definition (40 failures, 32.5% failure rate).
• Candidate predictors are restricted to variables available preprocedurally, enhancing clinical utility.

Limitations

• The dataset is from a single-center retrospective cohort, which may limit generalizability.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the underlying data is unknown, which may limit suitability assessment.

Provenance

Source

figshare

Collection Method

Single-center retrospective cohort study.

Time Range

April 2023 to May 2025

Freshness

Last updated 2026-06-03 02:02:58; freshness should be verified.