This study utilizes surface-enhanced Raman spectroscopy (SERS) in combination with a light gradient boosting machine (LGB) algorithm to establish a quantitative analysis model for atropine sulfate injection, enabling rapid and precise detection of atropine sulfate concentrations in eye drops. Initially, uniformly sized silver nanoparticles (Ag NPs) were synthesized, with their morphology and extinction spectra characterized experimentally, and their electric field distribution in silver colloidal dimers was calculated using Comsol simulation software. Using atropine sulfate injection as the research focus, we examined the impact of coagulants on Ag NPs' sensitivity in detecting atropine sulfate and optimized the type and concentration of coagulants, thereby establishing a foundation for its quantitative analysis. Using the LGB algorithm with spectral data from varying concentrations of atropine sulfate injections, we developed a quantitative analysis model. Finally, this quantitative model was applied to SERS spectral analysis of atropine sulfate in eye drops, yielding a correlation coefficient (R²) of 0.998, indicating high model accuracy. These findings may serve as a reference for the production and quality control of atropine sulfate eye drops in the marketplace.