A fuzzy logic model for differential diagnosis of lower urinary tract dysfunctions. Issue 3 (9th December 2015)

• Record Type: Journal Article
• Authors: Lopes, Maria HBM; D'Ancona, Carlos AL; Ortega, Neli RS; Silveira, Paulo SP; Faleiros‐Martins, Anna C; Marin, Heimar F
• Is Part Of: International journal of urological nursing. Volume 10:Issue 3(2016)
• Page Start: 146
• Page End: 153
• Publication Date: 2015-12-09
• Subjects: Diagnosis -- Differential -- Expert systems -- Fuzzy logic -- Lower urinary tract symptoms -- Nursing -- Urinary incontinence; Urological nursing -- Periodicals; 616.60231
• DOI: 10.1111/ijun.12108
• Languages: English
• ISSNs: 1749-7701

Abstract

ABSTRACT: Lower urinary tract dysfunctions (LUTD) restrict quality of life, resulting in decreased work productivity and emotional well‐being. However, most people are not diagnosed because they do not seek medical treatment. In addition, some facilities do not adequately train health professionals in the evaluation, diagnosis and treatment of these conditions. The study's objective was to develop a decision support system modelled on fuzzy logic that defines LUTD using the terminology of the International Continence Society. This methodological study aimed to develop a model that uses the maximum–minimum composition (max–min) of fuzzy relations that can perform differential diagnoses of LUTD. The model was tested in 100 cases (50 men and 50 women), and the data were obtained from medical records containing the clinical data and results of urodynamic studies. All medical records were reviewed by a specialist in urology. The model was capable of determining a diagnosis in full (62%) or partial (36%) agreement with the medical report. Agreement between the model and the medical report was excellent (kappa = 0·98, p ≪ 0·001, CI = 0·88–1) or substantial (kappa = 0·53, p ≪ 0·001, CI = 0·45–0·60), considering overestimative accordance (where accordance is assumed when at least one diagnosis is equal) and underestimative accordance (where accordance is assumed when all diagnoses are equal), respectively. The proposed model based on the max–min composition of fuzzy relationships is very simple and performed well. However, more tests are recommended before the model is used as a decision support system.