AI-assisted microfluidic immunoassay chip enabling early multiplex viral antibody detection in epidemics

Rapid serological diagnostics are crucial in the early stages of viral outbreaks, when timely immune profiling guides containment and treatment strategies. Conventional assays are slow, equipment dependent, and unsuitable for field use. Here, we present a peptide-coated microfluidic immunoassay chip that integrates AI-guided epitope prediction with a pump-free PDMS design. The prediction model, ABEpre, achieved AUCs of 0.80-0.85 across benchmarks, and its large-scale outputs were consolidated into a searchable viral epitope database. Guided by these predictions, epitopes were immobilized in parallel microchannels patterned with micropillar arrays to enlarge surface area and enhance antigen-antibody interactions. Reagents are introduced with a simple pen-style manual dispenser, while antigen-antibody binding is visualized through an indirect ELISA reaction.Validation with peptides from SARS-CoV-2, hepatitis B virus, and dengue virus confirmed antibody binding in ELISA and reproducible multiplex detection on-chip within 30 min. This structurally simple, low-cost system represents a promising platform for outbreak response, point-of-care diagnostics, and scalable viral surveillance. Overall, the study delivers a rapid sequence-to-chip translation pipeline that links epitope prediction to a deployable diagnostic tool.