Machine Learning-Driven Identification of Diagnostic Biomarkers in Ischemic Stroke: Focus on PI3K Pathway

Summary :High mortality and disability rates in ischemic stroke patients continue to pose substantial societal challenges, with the PI3K signaling pathway emerging as a critical mediator of post-stroke pathological processes. While this pathway's involvement in stroke pathophysiology is established, the complex interplay between PI3K-associated genes, stroke outcomes, and the immune microenvironment remains poorly understood, limiting the development of targeted immunotherapies. Here, we conducted a comprehensive analysis of PI3K pathway-related gene expression patterns in ischemic stroke samples, employing consensus clustering and immune infiltration analysis, coupled with machine learning algorithms and molecular docking experiments. Our analysis revealed two distinct patient subgroups with significant differences in immune infiltration profiles and identified five key diagnostic genes (PIN1, CDK2, VAV3, YWHAB, and CFL1). The developed predictive nomogram demonstrated high accuracy in disease onset prediction, validated through ROC analysis, while molecular docking experiments confirmed strong binding affinities between these genes and potential therapeutic compounds. These findings establish the PI3K signaling pathway as a crucial regulator of cerebrovascular and neural tissue repair following ischemic stroke, with the identified gene signature offering promising applications for early detection and prognostic assessment. Importantly, this classification system may enable the development of personalized immunotherapy strategies, potentially transforming the landscape of individualized stroke management.