Sistem Pakar Diagnosis Kanker Prostat Berbasis Web Menggunakan Logika Fuzzy Tsukamoto
TL;DR
Results indicate that the developed expert system can effectively support early screening of prostate cancer risk and serve as a decision-support tool for healthcare providers, particularly in facilities with limited access to urology specialists.
Abstract
Prostate cancer is one of the malignancies most commonly affecting men, particularly older adults, and remains a major contributor to cancer morbidity and mortality worldwide. Diagnosing prostate cancer requires interpreting multiple clinical parameters that are often uncertain and gradual in nature, which complicates medical decision-making. This study aims to design and implement a web-based expert system for prostate cancer risk diagnosis using the Tsukamoto fuzzy logic method. The system was developed with the Laravel 10 framework, PHP 8.1, and a MySQL 8.0 database, using four clinical input variables: Prostate-Specific Antigen (PSA) level, patient age, International Prostate Symptom Score (IPSS), and PSA density. A knowledge base of 25 fuzzy rules was formulated through iterative interviews with urology specialists and validated against 50 electronic medical record cases from a referral hospital covering the 2021–2024 period. The inference process follows the Tsukamoto method through fuzzification, rule evaluation using the minimum operator, and weighted-average defuzzification, producing a risk percentage classified into low, medium, or high categories. Black-box testing of eight functional scenarios achieved a 100% success rate. Accuracy testing on 30 of these 50 cases, validated by a urology specialist, yielded an overall accuracy of 90%, with 27 of 30 cases correctly classified. These results indicate that the developed expert system can effectively support early screening of prostate cancer risk and serve as a decision-support tool for healthcare providers, particularly in facilities with limited access to urology specialists.