EARLY DETECTION OF HUNTINGTONS DISEASE (HD) BY MACHINE LEARNING USING GENETIC DATA
By
Arunesh Dutt¹*, Komal Singh1,2, Pradeep Kumar Tiwari1,2 , Abhishek Pathak3 , Ashish Khare1, Vijaya Nath Mishra3 and Narendra Kumar Shukla3
1Department of Electronics and Communication Engineering, University of Allahabad, Prayagraj, Uttar Pradesh, India - 211002
2Centre of Computer Eduction and Training, Institute of Professional Studies, University of Allahabad, Prayagraj, Uttar Pradesh, India-211002
3Department of Neurology, Institute of Medical Sciences (IMS), Banaras Hindu University, Varanasi, Uttar Pradesh, India - 221005
*Corresponding Author
Email: aruneshdutt19@gmail.com, s.komalcs@gmail.com, pradeeptiwari@allduniv.ac.in, abhishekpathakims@gmail.com, khare@allduniv.ac.in, vnmishraneuro@bhu.ac.in, nkshukla@allduniv.ac.in
(Received : August 01, 2026; Revised : February 17, 2026; Accepted : February 20, 2026)
DOI: https://doi.org/10.58250/jnanabha.2026.561P-3
Abstract
A genetic condition known as Huntington’s disease (HD) impairs voluntary movement by causing the death of brain cells called neurons in various brain regions. It is challenging to identify early-stage HD in clinical practice since there aren’t many effective, simple-to- administer cognitive tests that are sensitive to organized data. Life expectancy has risen steadily as modern treatment has advanced. The prevalence of HD has also increased and is anticipated to increase over the next several decades as a result of escalating healthcare expenses. Early HD detection is therefore essential. Through this work, we used a clinical and genetic dataset of 10 HD patients. The present approach introduces a novel support vector machine (SVM)-based machine Learning (ML) algorithm for the detection of HD patients. The results of a classification study comparing various techniques indicate that SVM -based classification is more accurate at identifying subjects, even across varying stages of development. The detection capabilities of the suggested method were thoroughly examined using clinical and genetic data. With two-fold cross-validation, the maximum accuracies for early HD achieved by Gaussian SVM, Cubic SVM, Quadratic SVM, Naive Bayes, and Linear SVM are 70%, 100%, 100%, 80%, and 60%, respectively. These findings show that a useful and promising strategy for HD identification is the use of supervised classification techniques to genetic characteristics. The primary conclusions of this research are: 1) To detect substantial HD, a genetic test will be necessary. 2) The quadratic and cubic SVMs are the most effective predictors for HD. 3) According to experimental findings, quadratic and cubic SVMs outperform other classification algorithms for HD diagnosis as well. Even with a small dataset, the non-linear SVM approach improves diagnosis accuracy.
2020 Mathematical Sciences Classification: 00B99.
Keywords and Phrases: Huntington’s disease (HD), symptoms, classification issues, and machine learning (ML) approaches.