|Abstract:||The capability to detect bacteria at a low cell density is critical to prevent the delay in therapeutic intervention and to avoid the emergence of antibiotic-resistant species. Till date, great advancement has been made to develop sensing platform for rapid and reliable bacterial detection. However, critical requirements i.e. limit of detection (LOD), fast time of response (TOR), ultra-sensitivity with high reproducibility and ability to distinguish between bacterial strains have yet to be met within a single sensing platform. In this contribution, we present two approaches for bacterial detection. First, we proposed a novel label-free sensor based on pH-sensitive fluorescent yttrium doped carbon nanoparticles (YCNPs) embedded in agarose that can rapidly and accurately detect and discriminate pathogens in real-time. The developed sensor matrix presented pH-triggered aggregation-induced emission quenching of YCNPs in a wide pH range. When the pH increased from 10.0 to 4.0, the fluorescence of the matrix decreased linearly (R2 = 0.9229). The sensor ‘s high sensitivity in a physiologically relevant pH range enables to monitor the presence of live pathogens to single-cell resolution. In addition, the 3D matrix sensor showed low cytotoxicity and long stability (>30 days). Moreover, the YCNPs platform is stable over several hours (5h) in complex medium and do not alter the bacterial growth, allowing real-time monitoring of bacterial growth with a small volume (100 µL) and rapid response time (25 min). Furthermore, using machine learning assisted tools, different bacterial strains with various cell densities was discriminated with an accuracy of 100%. Moreover, blends of pathogens and pathogen in real-world sample can also be identified accurately. Second, we present a novel fluorescence sensing array. The sensing array is based on three different lanthanide-doped carbon nanoparticles (LnCNPs) embedded in agarose that can rapidly and accurately detect and discriminate pathogens in real-time. The developed sensor array presented selective fluorescence response and a distinctive molecular fingerprint for each type of bacteria. The differential fluorescence response pattern was generated from the unique interaction of bacterial strains with three differently doped carbon nanoparticles. The PrCNPs fluorescence response found to have a linear relationship with the bacterial CFUs with R2 >0.96 for B. subtilis, and S. mutans. The individualized fluorescence response pattern allowed us to employ machine learning assisted tools to demonstrate that the sensor array can rapidly and effectively discriminate three model species of bacteria predicting their properties (Gram-positive or Gram-negative) with >95% accuracy. Real-world samples are also classified with high accuracy, confirming the ability of the developed sensor array in rapid and reliable identification of pathogen information to allow continuous monitoring of infectious diseases.