Decision Support System Based on Artificial Neural Network for Prediction of Antibiotic Sensitivity of Causative Agents of Urinary Tract Infection in Certain Geographical Regions

Urinary tract infection (UTI) is very common and caused by various bacteria. The treatment of UTIs should be handled carefully. With the rise of neural networks, a possibility occurred to predict the outcome of consuming antibioticis for treating UTIs. This paper presents the development of expert system based on neural network for prediction of antibiotic sensitivity of two bacteria: Escherichia coli and Klebsiella pneumoniae. For the development of expert system based on neural network, total of 3226 samples were used: 486 samples of Klebsiella pneumoniae and 2740 samples for Escherichia coli. All samples were collected in one geographical area from hospitals and primary healthcare units. Feedforward neural network based on Bayesian regularization backpropagation training algorithm resulted in accuracy of 72.16% for prediction of antibiotic sensitivity of K. pneumoniae bacteria and 99.81% for prediction of antibiotic sensitivity of E. coli bacteria. The results of this study are promising since the usage of such expert systems in healthcare environment contributes to rational usage of antibiotics for treatment of infections and therefore contribute in fighting the antimicrobial resistance which is one of the rising challenges of healthcare nowadays.