Acute bacterial conjunctivitis is a prevalent ocular disease that can lead to severe vision impairment if not promptly treated. The conventional diagnostic method for bacterial conjunctivitis still relies on microbial culture, which, although highly sensitive, is time-consuming, labor-intensive, and unable to meet the demand for rapid detection. In this study, we developed a SERS@Au microarray chip as an enhanced substrate for collecting the SERS spectra of conjunctivitis-associated bacteria, including Escherichia coli, Staphylococcus aureus, Pseudomonas aeruginosa, and Staphylococcus epidermidis. The results demonstrated that the SERS@Au microarray chip has an excellent enhancement effect, reproducibility, and stability. By selecting the Raman shifts from 400 to 1 800 cm^-1 of pathogenic bacteria to perform an SVM model and OPLS-DA model, the discrimination accuracies achieved 97% and 90%, respectively. Detecting spiked tears using the SERS@Au microarrays chip can quickly, accurately, and conveniently screen pathogenic bacteria with simple culture. Collecting tears for direct testing is simple, painless, and non-destructive for patients. Combining the SERS@Au microarray with a portable Raman spectrometer is suitable for on-site screening of ophthalmic bacterial infections. Moreover, it enables the detection of mixed infections, thereby greatly enhancing overall efficiency in diagnosis and providing a valuable assistive tool for ophthalmic disease screening.