Hybrid Approach for Cardiac Arrythmia Prediction and Classification Using Machine Learning

Q: What is the clinical evidence from this study?

Population: Cardiac arrhythmia. Intervention: Weighted K-Nearest Neighbors (KNN) with Extra Trees feature selection and ESP32-based IoT monitoring vs. K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Decision Tree, and Naive Bayes. Primary outcome: Classification accuracy across 13 types of cardiac arrhythmias.

Q: What does this research mean for the field?

Weighted KNN combined with Extra Trees feature selection achieves 98.9% accuracy in classifying 13 types of cardiac arrhythmias. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

P Kiran; Tasleem Ara AF; Suresh H N; Jyothi G C; Jagdeesha S

doi:10.25258/ijddt.16.48s.37

October 5, 2026International Journal of Drug Delivery TechnologyOpen Access

Hybrid Approach for Cardiac Arrythmia Prediction and Classification Using Machine Learning

View Full Paper

Ask AI

Bookmark

Share

Why the study?

This study was conducted to develop a system integrating machine learning with real-time IoT monitoring for the early and accurate detection of cardiac arrhythmias and heart abnormalities.

Does a hybrid machine learning system using Extra Trees feature selection and Weighted KNN improve classification accuracy for 13 types of cardiac arrhythmia compared to other machine learning classifiers?

Key Points

Develop an automated cardiac arrhythmia prediction framework that couples machine learning algorithms with real-time IoT hardware for continuous physiological monitoring.
Analyzed a 278-attribute arrhythmia dataset using Extra Trees feature selection to identify critical cardiac parameters.
Trained and tested multiple machine learning classifiers, including weighted KNN, standard KNN, SVM, logistic regression, and naive Bayes, benchmarked against PCA and random forest feature reduction.
Built a real-time IoT hardware prototype incorporating an ESP32 microcontroller, AD8232 ECG sensor, MAX30100 heart rate and SpO2 sensor, and LM35 temperature sensor streaming data to ThingSpeak and a Flask application.
Weighted KNN achieved the highest diagnostic accuracy at 98.9% among all evaluated models.
The integrated system successfully classified 13 distinct cardiac conditions, including sinus tachycardia, sinus bradycardia, bundle branch blocks, and ventricular premature contractions.

Structured PICO

Does a hybrid machine learning system using Extra Trees feature selection and Weighted KNN improve classification accuracy for 13 types of cardiac arrhythmia compared to other machine learning classifiers?

P

Population

A 278-attribute cardiac arrhythmia dataset from the UCI Machine Learning Repository split 80:20 into training and testing sets.

I

Intervention

Intelligent Cardiac Arrhythmia Prediction System integrating Extra Trees Classifier (as well as PCA and Random Forest) for feature selection with Weighted K-Nearest Neighbors (KNN) machine learning classification and an ESP32-based real-time IoT monitoring prototype (incorporating AD8232 ECG, MAX30100 heart rate/SpO2, and LM35/DS18B20 temperature sensors with ThingSpeak cloud and Flask web application)

C

Comparator

Alternative machine learning classification models including standard K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Decision Tree, Logistic Regression, and Naïve Bayes

O

Outcome

Classification accuracy in predicting and classifying 13 types of cardiac arrhythmiasurrogate

Implication

An integrated system combining Extra Trees feature selection, Weighted KNN classification, and an ESP32-based IoT hardware prototype achieved 98.9% accuracy on a dataset for classifying 13 types of cardiac arrhythmias and demonstrated feasibility for real-time cloud-based physiological monitoring.

Abstract

This project proposes an Intelligent Cardiac Arrhythmia Prediction System that integrates machine learning with realtime IoT monitoring for early and accurate detection of heart abnormalities.The system analyzes a 278-attribute arrhythmia dataset, where Extra Trees is used for feature selection to identify the most important parameters.Among various models evaluated, Weighted KNN achieved the best performance with 98.9% accuracy, successfully classifying 13 types of arrhythmias For real-time monitoring, an ESP32-based IoT module with ECG, heart rate, and temperature sensors continuously collects patient data and uploads it to the ThingSpeak cloud platform.A Flask-based web application automatically retrieves the latest cloud data using Python and generates instant predictions without manual input.The system also provides a secure dashboard with SQLite3 authentication, prediction history, risk assessment, ECG deviation analysis, and personalized health recommendations, making it a scalable and fully automated solution for proactive cardiac care.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Kiran et al. (2026) studied Cardiac arrhythmia. Weighted K-Nearest Neighbors (KNN) with Extra Trees feature selection and ESP32-based IoT monitoring vs. K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Decision Tree, and Naive Bayes was evaluated on Classification accuracy across 13 types of cardiac arrhythmias.

synapsesocial.com/papers/6ae28e0ecbadf70daf6c4074 https://doi.org/10.25258/ijddt.16.48s.37

View Full Paper

Ask AI

Bookmark

Share

Hybrid Approach for Cardiac Arrythmia Prediction and Classification Using Machine Learning

Key Points

Structured PICO

Implication

Abstract

Discussion

Cite This Study

Hybrid Approach for Cardiac Arrythmia Prediction and Classification Using Machine Learning

Key Points

Structured PICO

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading