The paper introduces StratoLAMP, a label-free, multiplex digital loop-mediated isothermal amplification (LAMP) method based on visual stratification of precipitate byproducts. This method allows for the simultaneous quantification of multiple nucleic acid targets using deep learning image analysis to stratify the precipitate within each droplet. The design involves using two sets of primers with different concentrations to generate varying amounts of precipitate when amplifying different targets. The deep learning algorithm accurately detects and classifies droplets based on the precipitate level, achieving an overall accuracy of 94.3%. The method was optimized for various parameters, including primer concentrations, amplification mix concentrations, Mg²⁺ concentrations, and amplification times. The StratoLAMP was successfully applied to quantify two nucleic acid targets with high accuracy, demonstrating its potential for resource-limited settings and multiplex nucleic acid assays.The paper introduces StratoLAMP, a label-free, multiplex digital loop-mediated isothermal amplification (LAMP) method based on visual stratification of precipitate byproducts. This method allows for the simultaneous quantification of multiple nucleic acid targets using deep learning image analysis to stratify the precipitate within each droplet. The design involves using two sets of primers with different concentrations to generate varying amounts of precipitate when amplifying different targets. The deep learning algorithm accurately detects and classifies droplets based on the precipitate level, achieving an overall accuracy of 94.3%. The method was optimized for various parameters, including primer concentrations, amplification mix concentrations, Mg²⁺ concentrations, and amplification times. The StratoLAMP was successfully applied to quantify two nucleic acid targets with high accuracy, demonstrating its potential for resource-limited settings and multiplex nucleic acid assays.