Systematic Review of Aspiration Risk Prediction Models for Stroke Patients

A systematic review document evaluating 18 model development studies for predicting aspiration risk in adult stroke patients. The review, authored by Xiang-Ru Li and published on figshare in April 2026, assessed model performance, applicability, and risk of bias. It includes a comprehensive search across nine databases up to January 2025 and is registered with PROSPERO (CRD420251007112).

Use Cases

• Benchmarking new aspiration prediction models based on reported AUC/C-index performance metrics (0.756–0.955).
• Identifying common predictor variables for model feature engineering based on cited predictors like age, NIHSS score, and Glasgow Coma Scale.
• Assessing methodological quality in clinical ML studies based on the review's application of the PROBAST and CHARMS tools.
• Informing the design of future validation studies based on identified limitations like small sample sizes and retrospective designs.

Strengths

• Includes 18 model development studies, providing a consolidated view of the field.
• Methodologically structured using the CHARMS checklist and PROBAST tool for bias assessment.
• Search was conducted across nine electronic databases, suggesting a thorough literature collection.
• Registered prospectively with PROSPERO (CRD420251007112), indicating a predefined protocol.

Limitations

• The dataset is a 17.0 KB DOCX file, indicating a limited scope document rather than raw model data or code.
• Column-level documentation is absent; field semantics must be inferred from the review text after download.
• Row count is unknown, which may limit suitability assessment for direct data analysis.

Provenance

Source

figshare, authored by Xiang-Ru Li.

Collection Method

Systematic review of literature from nine electronic databases.

Time Range

Literature search conducted until 20 January 2025.

Freshness

Last updated 2026-04-16 05:58:32; freshness should be verified.

Geography

null