Neural network prediction model of risk for infection with carbapenem resistant Enterobacteriaceae among ICU patients in Colombia
Cargando...
Archivos
Fecha
2023-08
Autores
Título de la revista
Publicado en
Publicado por
URL de la fuente
Enlace a contenidos multimedia
ISSN de la revista
Título del volumen
Resumen
Antibiotic resistance, particularly the emergence of carbapenem-resistant Enterobacteriaceae (CRE), poses a significant threat to global public health. This study
aimed to develop a predictive model to estimate the clinical risk of CRE infection in patients admitted to an intensive care unit (ICU). A matched case-control study was conducted at a hospital in Bogotá, Colombia, involving 128 patients with CRE infections and 256 controls. The findings showed cardiovascular diseases and diabetes were the most common comorbidities among both groups. Univariate analysis revealed that patients in the case group were more likely to have undergone invasive procedures and medical devices such as central venous catheter insertion, urinary and foley catheter, also had a higher median number of hospitalization days. Moreover, patients with CRE infections had higher APACHE II scores. Previous infections caused by Enterobacteriaceae, hospital-acquired infections, and previous antibiotic treatments were significantly associated with CRE infections. The predictive model was developed using artificial neural networks (ANNs) and incorporated the identified risk factors. The model's performance was evaluated based on sensitivity, specificity, and accuracy, and different ANN configurations were compared. The model showed promise in accurately predicting the clinical risk of CRE infection in ICU patients. This study contributes to the understanding of risk factors associated with CRE infections in ICU settings and provides a practical tool for infection prevention and control strategies. The
use of predictive models based on neural networks in public health can revolutionize disease surveillance, resource allocation, and personalized healthcare interventions, ultimately enhancing population health outcomes.
Descripción
Abstract
Antibiotic resistance, particularly the emergence of carbapenem-resistant Enterobacteriaceae (CRE), poses a significant threat to global public health. This study
aimed to develop a predictive model to estimate the clinical risk of CRE infection in patients admitted to an intensive care unit (ICU). A matched case-control study was conducted at a hospital in Bogotá, Colombia, involving 128 patients with CRE infections and 256 controls. The findings showed cardiovascular diseases and diabetes were the most common comorbidities among both groups. Univariate analysis revealed that patients in the case group were more likely to have undergone invasive procedures and medical devices such as central venous catheter insertion, urinary and foley catheter, also had a higher median number of hospitalization days. Moreover, patients with CRE infections had higher APACHE II scores. Previous infections caused by Enterobacteriaceae, hospital-acquired infections, and previous antibiotic treatments were significantly associated with CRE infections. The predictive model was developed using artificial neural networks (ANNs) and incorporated the identified risk factors. The model's performance was evaluated based on sensitivity, specificity, and accuracy, and different ANN configurations were compared. The model showed promise in accurately predicting the clinical risk of CRE infection in ICU patients. This study contributes to the understanding of risk factors associated with CRE infections in ICU settings and provides a practical tool for infection prevention and control strategies. The use of predictive models based on neural networks in public health can revolutionize disease surveillance, resource allocation, and personalized healthcare interventions, ultimately enhancing population health outcomes.
Palabras clave
Carbapenem-resistant Enterobacteriaceae, Risk prediction model, Intensive care unit, Risk factors, Artificial neural networks
Keywords
Enterobacterias resistentes a carbapenemes, Modelo de predicción de riesgos, Unidad de cuidados intensivos, Factores de riesgo, Redes neuronales artificiales