Cluster Analysis Identifies Ischemic Stroke Subgroups Linked to Long-Term Recurrence

A data-driven cluster analysis has identified four distinct clinical phenotypes among ischemic stroke patients, revealing significant differences in long-term recurrence across the groups. Patients in the subgroup characterized by substantial vascular comorbidities and inflammatory profiles experienced the highest frequency of recurrent stroke during follow-up.

The study evaluated risk stratification and secondary prevention needs in ischemic stroke by analyzing patients hospitalized in the stroke unit of Istanbul Faculty of Medicine between 2011 and 2018. To assess outcomes, researchers included 799 patients who attended a three-month follow-up outpatient clinic. The cohort was 60 percent male with an overall mean age of 63.53 years (± 15.17).

To classify the patient population, investigators applied Gower distance to a mixed dataset comprising clinical, demographic, and imaging features, followed by partitioning around medoids for clustering. Feature selection was conducted through a systematic approach that combined univariate methods, internal validity indices, and visualization techniques.

The analysis classified the cohort into four distinct phenotypes based on clinical and demographic characteristics. Cluster 1 consisted of younger male patients with a favorable in-hospital course. Cluster 2 comprised high-risk male patients with substantial vascular comorbidity and inflammation, who went on to exhibit the highest recurrence. Cluster 3 included older female patients with cancer and inflammatory profiles, while Cluster 4 was composed of the oldest female patients, who presented with atrial fibrillation, prior stroke, and more severe stroke presentation.

Over a median follow-up period of 26 months, recurrent stroke differed significantly among the four clusters. The investigators noted that cluster analysis may serve as a valuable tool for clinical risk stratification in ischemic stroke, pointing out that the high recurrence observed in Cluster 2 highlights a group that may require closer monitoring and more intensive secondary prevention strategies.

The analytical strategy specifically incorporated survival analysis to compare long-term stroke recurrence across the four identified patient phenotypes. By tracking outcomes over the median follow-up duration of 26 months, the survival analysis evaluated how time to recurrent stroke differed among the distinct clinical groups. This analytical framework aimed to address the critical clinical requirement for accurate risk stratification, supporting the broader objective of developing individualized treatment strategies for secondary prevention in ischemic stroke.

Subject of Research: Data-driven cluster analysis of ischemic stroke phenotypes and long-term recurrence

Article Title: Data-driven phenotyping of ischemic stroke using cluster analysis: association with long-term stroke recurrence

Article References: Sezgin, M., Alp, O., Begik, E., Bingöl, M. S., Cansever, M. Y., Karaduman, C. K., Ekizoglu, E., Yesilot, N., & Günver, M. G. (2026). Data-driven phenotyping of ischemic stroke using cluster analysis: association with long-term stroke recurrence. BMC Neurology. https://doi.org/10.1186/s12883-026-05563-4

Image Credits: AI Generated

DOI: 10.1186/s12883-026-05563-4

Keywords: ischemic stroke, cluster analysis, stroke recurrence, risk stratification, secondary prevention, partitioning around medoids, vascular comorbidity, neurology, Data-driven, phenotyping, ischemic, stroke