AI-assisted smartphone-based colorimetric biosensor for visualized, rapid and sensitive detection of pathogenic bacteria

Abstract

Accurate and effective detection is essential to against bacterial infection and contamination. Novel biosensors, which detect bacterial bioproducts and convert them into measurable signals, are attracting attention. We developed an artificial intelligence (AI)-assisted smartphone-based colorimetric biosensor for the visualized, rapid, sensitive detection of pathogenic bacteria by measuring the bacteria secreted hyaluronidase (HAase). The biosensor consists of the chlorophenol red-β-D-galactopyranoside (CPRG)-loaded hyaluronic acid (HA) hydrogel as the bioreactor and the β-galactosidase (β-gal)-loaded agar hydrogel as the signal generator. The HAase degrades the bioreactor and subsequently determines the release of CPRG, which could further react with β-gal to generate signal colors. The self-developed YOLOv5 algorithm was utilized to analyze the signal colors acquired by smartphone. The biosensor can provide a report within 60 min with an ultra-low limit of detection (LoD) of 10 CFU/mL and differentiate between gram-positive (G+) and gram-negative (G-) bacteria. The proposed biosensor was successfully applied in various areas, especially the evaluation of infections in clinical samples with 100% sensitivity. We believe the designed biosensor has the potential to represent a new paradigm of “ASSURED” bacterial detection, applicable for broad biomedical uses.