Abstract
In this work, we report a novel proof-of-concept biosensing diagnostic tool for the multiplexed electrochemical quantitation of a unique combination of three UTI-relevant biomarkers, Prostaglandin E2 (PGE2), Interleukin-8 (IL-8), and Lipopolysaccharide (LPS), in unfiltered human urine. The proposed device, called USENSE, integrates lateral flow microfluidic channels, a gold-based sensor array for quantifying PGE2, IL-8, and LPS levels, and a random forest machine learning model for reliable diagnosis of UTI. The device is unique as it not only acts as a diagnostic device but also provides information on UTI by providing a risk score for UTI recurrence. USENSE is culture-free and label-free, requires no sample preparation at the user end, and can be adapted for use in home-based self-screening. In less than 5 minutes, USENSE directly measures the urinary concentration of PGE2, IL-8, and LPS and provides a UTI severity state classification: 0 = Healthy, 1 = Asymptomatic Bacteriuria, 2 = Symptomatic; low risk of relapse, 3 = Symptomatic; high risk of UTI relapse. In postmenopausal women, the PGE2, IL8, and LPS concentrations measured via the device correlated well with the levels measured using traditional enzyme-linked immunosorbent assay (ELISA). Our machine learning diagnostic model allowed for UTI diagnosis with 93% test accuracy and UTI prognosis state classification with >84% accuracy for the human urine samples tested. Further development of USENSE for clinical and home-based use could create a paradigm shift in point-of-care UTI diagnostics by allowing timely intervention and minimizing unwarranted empirical administration of antibiotics.