Comparative Analysis of the C5.0 Algorithm and Other Machine Learning Models for Early Detection of Multi-Class Heart Disease

Aug 5, 2026

Cardiovascular diseases remain a leading global cause of mortality, making accurate early detection essential for timely clinical intervention. This study performs a rigorous head-to-head comparison of five machine learning algorithms for multiclass heart disease classification using a private dataset of 300 patient records with 11 clinically relevant features. A stratified 5-fold cross-validation strategy evaluates Naïve Bayes, Logistic Regression, Random Forest, a C5.0-analog Decision Tree, and Support Vector Machine with accuracy, precision, recall, and F1-score.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Problem context and motivation

Digital healthcare and ML opportunity

Limitations in prior studies

Methodology

Dataset and features

Cross-validation strategy

Models compared

Results and discussion

Performance using standard metrics

Comparative findings and implications

Conclusion

Comparative Analysis of the C5.0 Algorithm and Other Machine Learning Models for Early Detection of Multi-Class Heart Disease Mardhatillah1*, Hafizh Al-Kautsar Aidilop2**, Asrianda 3* * Teknik Informatika, Universitas Malikussaleh [mardhatillah.180170011@mhs.unimal.ac.id](mardhatillah.180170011@mhs.unimal.ac.id1)[1](mardhatillah.180170011@mhs.unimal.ac.id1), [hafizh@unimal.ac.id](hafizh@unimal.ac.id2)[2](hafizh@unimal.ac.id2), [asrianda@unimal.ac.id](asrianda@unimal.ac.id3)[3](asrianda@unimal.ac.id3) Article history: Received 2025-06-04 Revised 2025-06-27 Accepted 2025-07-03 Keyword: Classification, Decision Tree, Heart Diseases, Early Detection. Cardiovascular diseases represent the leading cause of mortality worldwide, making accurate and early detection a critical factor for effective medical intervention and improved patient prognosis. While machine learning (ML) offers promising tools for predictive diagnostics, many existing studies rely on single-algorithm approaches or less-than-robust validation methods, thereby limiting the generalizability and realworld applicability of their findings.This study aims to conduct a rigorous, head-tohead comparative evaluation of multiple machine learning algorithms for the multiclass classification of heart disease, with the goal of identifying the most effective and reliable model for this complex clinical task.We utilized a private dataset comprising 300 patient medical records, each described by 11 clinically relevant features. To ensure a robust and unbiased evaluation, a stratified 5-fold crossvalidation methodology was employed. Five widely-used classification algorithms were evaluated: Naïve Bayes (NB), Logistic Regression (LR), Random Forest (RF), a C5.0-analog Decision Tree (DT), and Support Vector Machine (SVM) . Model performance was assessed using standard metrics, including accuracy, precision, recall, and F1-score.The comparative analysis revealed that the Naïve Bayes algorithm delivered superior performance, achieving the highest mean accuracy of 43.33%(±4.22%). It also led in other key metrics with a mean precision of 43.40%, recall of 43.64%, and an F1-score of 41.26%. Other algorithms, such as Logistic Regression (40.67% accuracy) and Random Forest (39.33% accuracy), demonstrated competitive performance but were ultimately surpassed by the Naïve Bayes model in this specific multi-class classification context.This research underscores the critical importance of employing robust validation techniques and comprehensive comparative analyses to identify optimal models for clinical applications. The Naïve Bayes algorithm emerges as a strong candidate for developing a reliable clinical decision support system for the early differentiation of various heart conditions, providing a foundation for future data-driven diagnostic tools. This is an open access article under the CC–BY-SA license. Article Info ABSTRACT 1. INTRODUCTION Cardiovascular diseases (CVDs) constitute a paramount global health crisis, consistently ranking as the foremost cause of mortality and morbidity across diverse populations [1, 2] . The spectrum of CVDs, encompassing conditions such as coronary artery disease (CAD), congestive heart failure, and cardiac arrhythmias, imposes a staggering burden on public health systems, national economies, and individual quality of life [4] . The insidious progression of many of these diseases often means that symptoms only manifest at an advanced stage, where treatment options may be limited and less effective. Consequently, the paradigm of modern cardiology has increasingly shifted towards proactive prevention and early detection. Identifying individuals at high risk or diagnosing a condition in its nascent stages is fundamental to enabling timely, effective interventions that can halt disease progression, mitigate severe complications, and significantly reduce mortality rates [8, 9] . The digital transformation of healthcare has ushered in

FAQs

What is the main goal of the study?

To conduct a head-to-head comparative evaluation of multiple machine learning algorithms for multiclass classification of heart disease to identify the most effective and reliable model.

How was model evaluation performed?

A stratified 5-fold cross-validation approach was used to reduce sampling bias and provide robust, unbiased performance estimates.

Which algorithm achieved the best performance, and on which metrics?

Naïve Bayes delivered the highest mean accuracy (43.33% ± 4.22%) and led other metrics as well, including mean precision (43.40%), recall (43.64%), and F1-score (41.26%).

What dataset size and feature set were used?

The study used a private dataset containing 300 patient medical records, each represented by 11 clinically relevant features.

Aug 5, 2026

Upload by Seraphina