Intan Kumalasari

Implementasi Kinerja Metode Fuzzy Sugeno Untuk Mendeteksi Penyakit Diabetes

Introduction

Implementasi kinerja metode fuzzy sugeno untuk mendeteksi penyakit diabetes. Deteksi penyakit diabetes akurat dengan metode Fuzzy Sugeno. Sistem pakar ini atasi ketidakpastian diagnosis DM, klasifikasi risiko berdasarkan gula darah, tekanan darah, dan BMI untuk diagnosis cepat.

Abstract

Diabetes Mellitus (DM) is a chronic disease caused by decreased insulin secretion, resulting in an increase in blood glucose levels in humans. This disease is still difficult to cure until now. Furthermore, diabetes is the main trigger for blindness, heart and kidney disease which can cause premature death almost all over the world. The problem faced is because of this large number, of course, every hospital and even the existing clinic needs a system that can help them quickly determine the results of all diabetes tests and classify which patients have the risk of diabetes, prediabetes even those affected by hypoglycemia or very low blood glucose levels. The main advantage of the fuzzy method is its ability to overcome uncertainty in the diagnosis of diabetes. Expert systems with the fuzzy method can handle varying variations in symptoms and severity by assigning weight or degree of membership to each symptom or condition.This allows the expert system to generate more accurate diagnoses based on the severity of the patient's symptoms and condition. The input variables used consisted of fasting blood sugar levels, sugar levels after meals, blood pressure, and body mass index. The output is in the form of detection results of positive and negative patients with diabetes. In the testing stage, the researcher performs calculations manually and implements the performance of the Fuzzy Sugeno method using the Java programming language. Where the data used, sourced from the PT. Primary. It is hoped that the output of this study will be an accredited journal.    

Review

This manuscript proposes an implementation of the Fuzzy Sugeno method for detecting diabetes, a highly relevant and critical health issue globally. The abstract clearly articulates the problem of diabetes diagnosis and the potential for an expert system to assist healthcare providers in rapidly classifying patients. The selection of Fuzzy Sugeno is justified by its inherent ability to manage uncertainty in diagnostic scenarios, assigning degrees of membership to various symptoms and conditions. The stated input variables (fasting blood sugar, post-meal sugar, blood pressure, BMI) are appropriate for diabetes detection, and the aim to produce positive/negative diabetes detection results is a practical objective for a clinical support system. The practical motivation to develop a system for hospitals and clinics is a strong point, highlighting the potential impact of this research. However, several critical aspects require further elaboration and clarification to fully assess the methodological rigor and scientific contribution. While the Fuzzy Sugeno method is introduced, the abstract provides no detail on how the fuzzy rules are formulated, what expert knowledge (if any) is leveraged for membership function design, or the specific defuzzification strategy employed. The data source, "PT. Primary," is mentioned but lacks essential details regarding its nature (clinical records, public dataset, synthetic), size, demographic characteristics, and how it was collected or preprocessed. Furthermore, the description of the testing stage, involving "manual calculations" and Java implementation, is vague regarding the validation methodology. There is no mention of performance metrics (e.g., accuracy, sensitivity, specificity, AUC) or a comparison against existing diagnostic standards or other machine learning approaches, which are crucial for establishing the system's efficacy and reliability. The abstract also presents a slight discrepancy, initially mentioning classification into "risk of diabetes, prediabetes, even hypoglycemia," but later narrowing the output to simply "positive and negative patients with diabetes." In conclusion, the proposed research addresses a significant and impactful problem with a relevant methodological approach. The concept of utilizing Fuzzy Sugeno for diabetes detection holds promise due to its ability to handle diagnostic uncertainty. However, the abstract currently lacks the necessary technical depth and validation detail expected from a robust scientific contribution. To merit publication, a full manuscript would need to provide comprehensive details on the fuzzy system's design, a thorough description of the dataset, and a rigorous, quantitative evaluation of its performance, including comparison with established benchmarks. Significant revisions addressing these gaps would be necessary to elevate this work to an publishable standard.

Full Text

Comments

You need to be logged in to post a comment.