Cell-cell co-occurrence patterns are critical in shaping biological and clinical phenotypes, but existing tools for analyzing single-cell RNA sequencing (scRNA-seq) data often lack phenotype annotations and face scalability limitations. To address this, we developed iPhenoChat, a cluster-free computational method that integrates bulk and single-cell RNA-seq data to identify phenotype-associated cell-cell interactions. Leveraging machine learning and optimized for speed using PyTorch and GPU acceleration, iPhenoChat bridges single-cell resolution with bulk-level phenotyping, enabling deeper insights into disease progression and clinical outcomes.