This document describes the development of a diagnostic device to rapidly identify bacterial strains responsible for sepsis. Key points: - The device uses droplet digital PCR to amplify and quantify bacterial DNA signatures from patient blood samples. High resolution melt analysis is then used to generate strain-specific melt curves of the 16S rRNA gene. - A microfluidic chip is engineered to generate stable droplets and thermal gradients allowing acquisition of melt curves. Machine learning algorithms classify melt curves to identify bacterial strains. - The goal is to identify bacteria in patient blood samples within clinically relevant timeframes, shortening diagnostic time from 12 days to 3-4 hours for targeted antibiotic treatment. Results show a stable thermal gradient is achieved within the chip