An Enhanced Multi Layer Neural Network to Detect Early Cardiac Arrests

Patients and doctors are paying increasing attention to health-care automation because it can save a person's life by predicting ailments early. Many people are suffering from chronological diseases as a result of altering eating habits, regardless of age or gender. "Heart Attacks" is...

Full description

Saved in:

Bibliographic Details
Published in:	2021 5th International Conference on Electronics, Communication and Aerospace Technology (ICECA) pp. 1514 - 1518
Main Authors:	Sirisati, Ranga Swamy, Kumar, C. Srinivasa, Latha, A. Gautami, Kumar, B. Narendra, Rao, Kanusu Srinivas a
Format:	Conference Proceeding
Language:	English
Published:	IEEE 02-12-2021
Subjects:	8- Layered Multi Layer Preceptron Architecture Cardiac arrest Classification algorithms Ensemble Algorithm Feature Engineering Greedy Genetic Algorithm Heuristic algorithms Medical services Meta Classifier Predictive models Support vector machines Training
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Patients and doctors are paying increasing attention to health-care automation because it can save a person's life by predicting ailments early. Many people are suffering from chronological diseases as a result of altering eating habits, regardless of age or gender. "Heart Attacks" is a severe ailment that requires attention from time to time. To date, all automated systems have built models using either classical or ensemble machine learning techniques. Overfitting has affected only a few of these systems, such as random forest and SVM algorithms. As a result, the proposed approach has chosen the "Multi Layer Preceptron" neural network technique, which solves the problem of overfitting and generates an accurate number of correct labels linked with the training model. Instead of using all of the variables mentioned in the dataset, the suggested method assists clinicians in predicting a heart attack in a user at an early stage by assessing only 7 top informative attributes. The model was also compared to other classifiers in order to establish the state of the art, which was determined to be "97.23 percent."
DOI:	10.1109/ICECA52323.2021.9675882