SOFTWARE USING ARTIFICIAL INTELLIGENCE TO PREDICT URINARY TRACT INFECTIONS CAUSED BY MULTIDRUG-RESISTANT BACTERIA

Abstract

Antimicrobial resistance is a critical global issue that threatens advances in public health. In recent years, the rise of bacterial resistance to antibiotics has led to an alarming number of deaths and increased healthcare costs. Additionally, the development of new antibiotics is a slow and challenging process, further exacerbating the problem. In this context, artificial intelligence (AI) emerges as a powerful tool, enabling the analysis of multiple criteria to help combat multidrug-resistant bacterial infections. The aim of this project is to develop AI-based software to predict urinary tract infections (UTIs) caused by bacteria resistant to specific antibiotics. The methodology involved a retrospective study of 762 patients hospitalized with UTIs between October 2013 and April 2019. Patients were randomly assigned to training and validation groups. The results show that the artificial neural network (ANN) developed to predict infections caused by carbapenemase-producing enterobacteria (CRE) bacteria has an accuracy of 96.1%, with a sensitivity of 89.2% and specificity of 98.4%. For extended-spectrum beta-lactamase (ESBL) bacteria, the accuracy was 91.6%, with a sensitivity of 92.6% and specificity of 94.8%. These findings demonstrate the high effectiveness of the model in identifying patterns of urinary infections caused by these multidrug-resistant bacteria. Implementing this system could innovate medical diagnostics, reduce bacterial resistance through more accurate treatments, and lower costs related to patient readmissions and hospitalizations.