Portal de Periódicos da CAPES

Urinary Tract Infections (UTIs): Laboratory diagnosis - The role of artificial intelligence and smart diagnosis

2024; Volume: 8; Issue: S1 Linguagem: Inglês

10.53730/ijhs.v8ns1.15294

2550-6978

Maryam Abdullah AlThowaimer, Waseem Ali Alquwayi, Abdulaziz Ali Almuarik, Yasser Abdrab Alameer Alkuwaiti, Ahmed Mohammed Almehainy, ‏Bakr Mansour Alqahtani, Mazen Ibrahim Mohammed Otaif, Tahani Abbas Alkattan, Mohammed Abdullah Alharbi, Mohammed Hasan Albather, Abdulaziz Alanazi, Ahmed Salem Almohammadi,

Pediatric Urology and Nephrology Studies

Background: Urinary tract infections (UTIs) are prevalent outpatient conditions affecting up to 50% of individuals, with diagnostic errors common in clinical settings. The traditional reliance on clinical criteria alone yields a diagnostic error rate of about 33%, necessitating improved diagnostic methods. Aim: This mini-review evaluates the role of artificial intelligence (AI) and smart diagnostic tools in enhancing UTI diagnosis, particularly within vulnerable populations. Methods: A comprehensive literature review was conducted, assessing 782 articles, of which 14 met the inclusion criteria for AI applications in UTI diagnosis. These studies were categorized based on their focus: uncomplicated UTIs, complicated UTIs, and specific demographic groups. Results: The review revealed that 12 studies employed machine learning techniques while 2 utilized deep learning. The most frequently used models included artificial neural networks (ANNs) and extreme gradient boosting (XGBoost). Key variables influencing predictive models encompassed demographic data, anamnesis, and comorbidities. Notably, models for diagnosing uncomplicated UTIs achieved accuracy rates of up to 98.3%, while approaches for complicated UTIs demonstrated area under the curve (AUC) values ranging from 0.71 to 0.904. AI models were particularly effective in stratifying high-risk subgroups, including pregnant women and children, with models achieving AUCs of 0.82 and 0.83 for specific populations.