Wrapper subset evaluation facilitates the automated detection of diabetes from heart rate variability measures
Year: 2004
Type: conference-paper
Abstract: Diabetes affects almost one million Australians, and is associated with many other conditions such as vision loss, heart failure and stroke. Any improvement in early diagnosis would therefore represent a significant gain with respect to reducing the morbidity and mortality of the Australian population. In this study we apply signal processing and automated machine learning to analyse heart rate variability measures. These data are well suited to the diagnosis of cardiac dysfunction, but here we use the same measures to detect diabetes. By applying appropriate methods we were able to select the most relevant features to use as input to a variety of classifier algorithms. We compare sensitivity and specificity results obtained from these classifier algorithms. Results suggest that the detection of diabetes is feasible from heart rate variability measures . 1. Introduction Both the NSW Department of Health and Commonwealth Government have identified diabetes to be a significant and growing global public health problem with the expected incidence in Australia to increase from 4% to 10% by 2010 [1]. In Australia approximately 1 million people are affected by diabetes and health care costs associated with treatment of complications amounts to approximately $7 billion dollars [2]. Vision loss, heart failure and stroke contribute significantly to the morbidity and mortality of the Australian population. Diseases of the circulatory system such as coronary heart disease and stroke were listed as the underlying cause of death in 55.7% of deaths in 2000 where diabetes was an associated cause [3]. In addition some form of nervous system damage such as cardiac autonomic neuropathy (CAN) that affects the function of the heart and blood vessels, occurs in up to 60-70% of people with diabetes [4]. Diabetes may lead to subtle changes in heart rate variability that become increasingly more compromising, manifesting in arrhythmia and cardiac failure in the majority of individuals with diabetes. Early detection of this pathology would allow timely intervention and thus lessen morbidity and mortality. This would improve the quality of life for people with diabetes and at risk for heart failure and stroke, as well as contributing to a substantial saving to the health care system [5]. To assess CAN the variation in the interval length of the ECG between successive beats can be analyzed in terms of heart rate variability (HRV). From this sequence of intervals, many secondary measures may be derived, making use of fractal and wavelet analysis techniques. It is well known as a tool for detecting cardiac dysfunction [6, 7]. Discrimination of various categories or classes (such as “cardiac dysfunction” or “normal”) is a well-studied class of machine learning problems. Here the key is to determine some relationship between a set of input vectors that represent stimuli, and a corresponding set of values on a nominal scale that represent category or class. The relationship is obtained by applying an CIMCA 2004 Proceedings/ISBN :1740881885 M. Mohammadian (Ed.)12-14 July 2004, Gold Coast - Australia
Language: English
Subfield: Cardiology and Cardiovascular Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W66168305