Page 69 - Kaleidoscope Academic Conference Proceedings 2024

PIMA DIABETES PREDICTION USING MACHINE LEARNING AND QUANTUM
MACHINE LEARNING TECHNIQUES

Vimal, Dixit
Jawaharlal Nehru University, New Delhi, India

ABSTRACT Type I diabetes evokes the stage of autoimmune destruction
of pancreatic cells of islets beta cells and it is the most
Quantum Machine Learning (QML) which is the integration common type of diabetes in childhood or adolescence. In
of Quantum Mechanics (QM) and Machine Learning (ML), contrast, type 2 diabetes, as the most widely ramified one
comes with immense computational capacity due to principles in the world, anticipates both insulin resistance and relative
of superposition and entanglement. The PIMA diabetes insulin deficiency, which are usually connected with irregular
dataset is used in this research as a test bed for comparing lifestyle and overweight tendency [2]. Gestational Diabetes
the results obtained from ML and QML approaches. The Mellitus (GDM) is also caused by hormonal changes that
assessment shows that Support Vector Machine (SVM) make the maternal insulin resistance prevalent, as these pose
has better results compared to Logistic Regression (LR) danger to both mother and fetus.
with an accuracy of 0.76. On the other hand, among
QML algorithms; the Quantum Support Vector Classifier Besides this, several types of diabetes are distinct, such
(QSVC) has a higher accuracy than the Variational Quantum as MODY [3] that gifts diabetes at a young age, diabetes
Classifier (VQC) with 0.74. The findings of the given research secondary to pancreatic pathologies and induced drug effects,
imply that the use of QML methods may be valuable for further and endocrine disruptions that affect insulin secretion or
developments in diabetes prediction in the future.
action. Accurate classification of the types of diabetes and
their origin is fundamental for the use of individualized
Keywords - ML, QML, LR, SVM, VQC, QSVC management strategies, and timely interventions for the
prevention and mitigation of risk of consequences. These
1. INTRODUCTION strategies aim at improving patients’ prognosis. By
investigating the pathogenesis, subtle differentiations of
Diabetes is among the primary causes of death globally as
clinical features, and idiosyncrasies between types of
well as in India. As per the report of the World Health
diabetes, healthcare specialists can then give patients care
Organization (WHO) [1], the death rate due to diabetes
matching their unique needs, help them get their blood sugar
mellitus by the age group 55-59 is the lowest and highest
under control, and empower them to live healthier lives.
in 85 and above, as depicted in Figure 1. Whereas it is
Figure 2 depicts the Prevalence rate of diabetes in adults
very low in the age group of 0-29 and low to moderate in
according to WHO.
the age group of 30-54. Diabetes, or metabolic disorder, is
a multi-faceted category that includes several sub-types that
have their unique cause, pathophysiology, and presentation.

Figure 2 – WHO, Prevalence rate of diabetes in adults, 2014.

The world is facing an increasing number of people suffering
from diabetes, so the search for better-predicting tools and
efficient methods of treatment becomes imperative. ML
Figure 1 – Death Rate due to diabetes mellitus by age group.
and consequently QML technologies being among the most

978-92-61-39091-4/CFP2268P @ITU 2024 – 25 – Kaleidoscope