A Hybrid Feature Selection Approach for Urinary Tract Infection Detection and Prediction in IoT-Fog Environment

Abstract

Urinary Tract Infections (UTIs) are a common health issue that millions of people experience globally and a significant impact on the general health. Depending on the location and intensity of the infection, UTIs appear differently clinically. Dysuria, frequency, urgency, suprapubic discomfort, and hematuria are typical symptoms. Patients with severe conditions could have a fever, flank discomfort, and other systemic symptoms that point to an upper UTI. Clinical assessment and lab tests are used to get the precise diagnosis of UTIs. The primary aim of this research is to utilize appropriate Machine Learning (ML)-based algorithms to predict Urinary Tract Infections (UTIs) in IoT-Fog environments. The ultimate goal is to develop a predictive model that can be effectively implemented in a smart toilet system. By achieving this objective, the study aims to offer an innovative and practical solution for UTI prediction, leveraging the potential of ML algorithms in IoT-Fog environments to enhance healthcare and improve public health. This paper presents hybrid approach for feature selection and using Guided Regularized Random Forest (GRRF) classification to assist with the diagnosis of UTI. Data from regular exams and definitive diagnostic results for UTI patients were used to generate a UTI dataset. Principle Component Analysis (PCA) is used for dimensionality reduction, while K-best and Lasso CV are used for feature selection. Using our suggested strategy, this research was able to identify UTI with a 98.8% accuracy and 98.90% precision rate. Future UTI prevention and treatment plans must be optimized via further research and ongoing efforts to overcome antibiotic resistance.