Source-linked AI summary
Classification of Heart Disease Using K- Nearest Neighbor and Genetic Algorithm
M. A. Jabbar, B. L Deekshatulu, Priti Chandra
TL;DR
Medical datasets may contain redundant and irrelevant attributes, while heart disease creates a need for effective clinical decision support. The paper combines KNN with genetic search to prune and rank attributes before classification, reporting improved accuracy over KNN without genetic algorithms. The resulting model is presented as useful for heart-disease diagnosis with fewer attributes.
Problem
Redundant and irrelevant attributes may reduce classification accuracy in medical datasets, while heart disease motivates improved diagnostic decision support.
Method
The approach combines KNN with genetic search to prune redundant and irrelevant attributes, rank informative attributes, and classify heart-disease records as healthy or sick.
Results
Average accuracy was higher than KNN without genetic algorithms, with Andhra Pradesh heart-disease accuracy improved 5% and weather-dataset accuracy reaching 14.29%.
Takeaways & Limitations
The prediction model is presented as helping doctors diagnose heart disease efficiently using fewer attributes.
Takeaways & Limitations
KNN depends on selecting suitable k values, while genetic algorithms require difficult fitness-function selection, many evaluations, and non-straightforward configuration.
Abstract
from arXiv · showhide
Data mining techniques have been widely used to mine knowledgeable information from medical data bases. In data mining classification is a supervised learning that can be used to design models describing important data classes, where class attribute is involved in the construction of the classifier. Nearest neighbor (KNN) is very simple, most popular, highly efficient and effective algorithm for pattern recognition.KNN is a straight forward classifier, where samples are classified based on the class of their nearest neighbor. Medical data bases are high volume in nature. If the data set contains redundant and irrelevant attributes, classification may produce less accurate result. Heart disease is the leading cause of death in INDIA. In Andhra Pradesh heart disease was the leading cause of mortality accounting for 32%of all deaths, a rate as high as Canada (35%) and USA.Hence there is a need to define a decision support system that helps clinicians decide to take precautionary steps. In this paper we propose a new algorithm which combines KNN with genetic algorithm for effective classification. Genetic algorithms perform global search in complex large and multimodal landscapes and provide optimal solution. Experimental results shows that our algorithm enhance the accuracy in diagnosis of heart disease.
1. Introduction
The paper introduces data mining and classification for medical decision support, then reviews KNN and genetic algorithms as relevant techniques for heart-disease analysis. It describes KNN distance requirements and limitations, alongside genetic-algorithm operators, advantages, and limitations.
- Data mining and healthcare: Data mining automatically extracts knowledgeable information from large datasets and supports classification applications in healthcare.Healthcare organizations use these techniques to address diagnosis and treatment-related questions.
- K-nearest neighbor: Common distance measures for KNN include Euclidean distance, while continuous variables may use Z-score standardization or min-max normalization.Distance functions are expected to satisfy non-negativity, symmetry, and the triangle inequality.
- K-nearest neighbor: KNN classifies samples using neighboring instances but suffers from low efficiency and dependence on selecting suitable k values.The paper identifies improving KNN accuracy through better k selection as requiring further research.
- Genetic algorithm: Genetic algorithms represent candidate solutions with chromosomes and populations, using selection, crossover, mutation, and fitness evaluation.Selection chooses individuals for reproduction, crossover combines parent chromosomes, and mutation alters new solutions during search.
- Genetic algorithm: Genetic algorithms can search broad solution spaces and discover global optima, but require suitable fitness functions, many evaluations, and nontrivial configuration.The paper also notes that genetic algorithms are not good at identifying local optima.
- Heart disease: Heart disease involves blocked or narrowed arteries and is described as a major cause of death in India and Andhra Pradesh.Reported risk factors include high blood pressure, abnormal blood lipids, tobacco use, obesity, inactivity, diabetes, age, gender, and family history.
3. Proposed method
The proposed approach uses genetic search to evaluate and rank attributes, removes lower-ranked attributes, and then applies KNN with GA to classify heart-disease data. Reported comparisons indicate higher accuracy than KNN without GA across evaluated data sets.
- Proposed method: The approach combines genetic search with KNN to improve heart-disease classification accuracy.Genetic search evaluates attributes before classifier construction.
- Proposed method: Genetic search ranks attributes by their contribution, removes lower-ranked attributes, and retains a higher-ranked subset.The resulting classifier is built using the evaluated attributes.
- Proposed method: The classifier is trained to classify heart-disease data as healthy or sick and its accuracy measures correct classification of unknown samples.The workflow loads the data, applies genetic search, selects attributes, and calculates classifier accuracy.
- Proposed method: KNN parameters include varying K values, cross-validation, distance weighting, mean-squared calculation, and disabled normalization.The listed K values range from 1 through N.
- Results: Average accuracy is higher with the proposed approach than with KNN without GA across the compared data sets.The reported comparisons include figures and tables for multiple data sets and evaluation settings.
- Results: Heart-disease accuracy for Andhra Pradesh improves 5% over the classification algorithm without GA.The paper also reports 14.29% accuracy for the weather data set using the proposed approach.
5. Conclusion
The paper presents a heart-disease classification approach evaluated on Andhra Pradesh data and other UCI data sets. The authors report competitive classification performance and potential support for diagnosis using fewer attributes.
- Conclusion: The approach was tested on Andhra Pradesh heart-disease data and other machine-learning data sets from the UCI repository.The evaluation covered 7 data sets.
- Conclusion: The reported experimental results characterize the approach as competitive for classification.This conclusion is based on experiments across the evaluated data sets.
- Conclusion: The prediction model helps doctors with heart-disease diagnosis using fewer attributes.The conclusion also connects heart-disease mortality with identifying major risk factors and developing decision-support measures.