Acute Ischemic Stroke Prognostic Model with CAR and CTI Indices

A Korean single-center stroke registry from 2010-2016 contains data for 1,484 patients with acute ischemic stroke. It analyzes associations between inflammation-related indices (CAR and CTI) and 90-day functional outcomes, with a derived six-variable prognostic model achieving an AUC of 0.837.

Use Cases

• Validate the six-variable prognostic model (sex, age, BMI, NIHSS, CAR, CTI) for 90-day functional outcomes in new stroke cohorts.
• Analyze the non-linear risk gradient between the C-reactive protein-to-albumin ratio (CAR) and unfavorable outcomes using multivariable logistic regression.
• Compare the prognostic performance of the CRP–triglyceride–glucose index (CTI) against individual biomarkers like albumin and fasting plasma glucose.
• Assess model calibration using the reported Brier score, calibration intercept, and calibration slope from the internal validation.

Strengths

• Registry includes 1,484 patient records from a defined 2010-2016 period.
• Study reports specific performance metrics for its prognostic model, including a cross-validated AUC of 0.831 and a Brier score of 0.142.
• Analysis includes internal validation via repeated 10-fold cross-validation, providing evidence of stable model performance.

Limitations

• Dataset scope is limited to a single center in Korea, which may limit generalizability to other populations.
• Primary data is presented in a summary DOCX file (618.2 KB), not as a raw, analyzable table, restricting direct computational use.
• The derived prognostic model is exploratory and requires external validation, as noted by the authors.

Provenance

Source

Korean single-center stroke registry.

Collection Method

Retrospective analysis of registry data; CAR and CTI were calculated from admission laboratory tests.

Time Range

2010 to 2016.

Freshness

Data covers patients from 2010 to 2016; the analysis was last updated in March 2026.

Geography

Single center in Korea.