Fuzzy Expert System to Diagnose Neonatal Peripherally Inserted Central Catheters Infection

Abstract

Introduction: Peripherally inserted central catheters (PICC) are utilized in neonatal intensive care units (NICUs) as an instrument to vascular access. The PICCs significantly reduce side effects compared to central and peripheral venous catheters and can be the cause of catheter-related bloodstream infections (CRBSIs). The purpose of this study was to create a fuzzy expert system for the early diagnosis of PICC-related infections in newborns. Methods: This descriptive-applied study was conducted in 2016. The statistical population of this research consisted of the medical files of newborns in Children's Medical Center in Tehran, Iran, and sampling was carried out using convenient sampling method. The research tools were a checklist and questionnaires. Factors affecting infection diagnosis were determined based on pediatric specialists’ comments. The system was designed bilingually (Persian and English) using C# software and SQL Server database. The output of the system is the percentage of infection risk. The system evaluations were carried out using data from the medical files of newborns in a hospital in Tehran, Iran. Data was analyzed using Excel software. Results: Based on system assessment and comparison of the system output with the diagnosis of the specialists, the sensitivity, specificity, and accuracy of the system were 95%, 88%, and 91%, respectively. Conclusion: The non-specificity of clinical signs and laboratory findings of blood infection in newborns have made its diagnosis difficult and uncertain. Using a designed expert system can be effective in the diagnosis of CRBSIs.

Keywords

• Clinical Decision Support Systems
• Fuzzy Logic
• Peripherally Inserted Central Catheters
• Catheter-related Infections