Cardiovascular disease remains a major global health challenge, and Myocardial Infarction (MI) is among its most critical manifestations. Post-MI complications significantly increase patient morbidity and mortality, underscoring the clinical importance of identifying individuals at high risk. This paper repats on work directed at analysing duel-modality flow cytometry data, specifically, tabular summaries and image-based plots of key markers, to enhance complication prediction. The proposed framework, FlowCytFuse, applies data preprocessing techniques optimized for flow cytometry, including the removal of count beads to reduce noise, normalization of fluorescence intensities, and visualization of crucial features as two-dimensional plots. This is combined with a dual neural network architecture: the first network handles tabular data, while the second processes image-based representations (scatter and density plots). A voting mechanism then fuses both outputs to produce a final prediction. In testing on a real-world dataset of 246 patients, the method demonstrates marked improvements in F1 scores for minority-class (complication) cases compared to earlier approaches. These findings highlight the potential of blending numerical and visual representations of flow cytometry data to deliver more accurate and clinically meaningful post-MI risk stratification.