Using clinical and thrombus characteristics to predict the etiology of ischemic stroke: An analysis of the INSIGHT registry
Simple blood counts help predict stroke type, offering inexpensive clues to guide treatment decisions in acute stroke.
Analyzing 388 stroke patients in the INSIGHT registry with 123 features including CBC parameters, LASSO machine learning models achieved AUC 0.71-0.73 for predicting cardioembolic and AFIB-specific stroke etiologies. CBC-derived parameters (platelet count, RBC count) showed significant independent associations with stroke subtype, suggesting blood count features contribute useful signal to stroke etiology classification.
What the study was
- Study design
- Retrospective registry analysis (INSIGHT; NCT04693767) with LASSO ML modeling
- Population
- 388 ischemic stroke patients (208 cardioembolic, 180 non-CE) from INSIGHT registry
- Sample size
- 388
- Category
- Diagnostics
- Maturity
- Exploratory
- Journal
- J Stroke Cerebrovasc Dis
Why it surfaced
Registry-based ML study with moderate AUC suggesting CBC parameters add predictive value for stroke etiology. AUC 0.71-0.73 is modest; retrospective design. Relevant to CBC-based ML diagnostics topic.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.