Systematic Review of Thrombosis Risk Prediction Models for Leukemia Patients

A 2025 systematic review by Minlan Ye evaluating 16 risk prediction models for thrombosis in leukemia patients, sourced from 14 studies. The review, published on figshare under CC-BY-4.0, analyzed models developed from cohorts ranging from 102 to 1252 participants. It assessed model discrimination, validation status, and methodological quality using the CHARMS checklist and PROBAST tool.

Use Cases

• Benchmarking new thrombosis risk models against existing ones based on reported AUC/C-index values (0.641-0.917).
• Identifying common clinical predictors for feature engineering, such as central venous catheter placement or D-dimer levels.
• Assessing methodological quality and bias in clinical prediction models using the PROBAST tool framework.
• Designing future validation studies based on identified gaps, such as the lack of external validation.

Strengths

• Systematic methodology covering 1825 initially identified records from nine databases.
• Explicit evaluation of 16 models across 14 studies, with development cohort sizes provided (102-1252 participants).
• Clear reporting of model performance metrics, with discrimination (AUC/C-index) ranging from 0.641 to 0.917.

Limitations

• The dataset is a 687.3 KB DOCX document containing review text; the underlying model data or code is not included.
• Row count and column-level documentation for any underlying data are unknown.
• All included studies were rated high risk of bias, and five raised major concerns regarding applicability.

Provenance

Source

Author Minlan Ye, published on figshare.

Collection Method

Systematic literature review with independent study selection, data extraction, and quality assessment by two reviewers.

Time Range

Literature searched from database inception to August 4, 2025.

Freshness

Last updated 2026-03-18 07:37:43.