Date of Award

2007-01-01

Degree Name

Master of Science

Department

Electrical Engineering

Advisor(s)

HOMER NAZERAN

Abstract

Diabetes is one of most afflicting diseases in society which is caused by the inability of the body to appropriately use insulin to balance sugar levels. Type 1 diabetics require monitoring of their glucose levels to ensure a normal glucose level is maintained between 90 and 130 mg/dL (milligrams per deciliter). Unfortunately, current methods of monitoring are based on invasive and painful sampling of the blood. It was imperative that noninvasive and painless methods of monitoring glucose should be explored. A proven noninvasive technique utilizing infrared light to detect the fluctuations in the heart rate was used to determine if levels of glucose could be accurately determined within regions of safety. The optical absorption properties of blood were used to obtain photoplethysmographic (PPG) data and processed to gather information about the subject's heart rate variations (HRV). The system developed for this study consisted of simultaneously recorded ECG, PPG, and temperature. The data were analyzed on the PDA or through the use of a desktop computer with a custom HRV analysis program written in LabView. Glucose levels were simultaneously measured using a portable glucose monitor. This research study consisted of three primary goals: confirm the reliability of the developed HRV system, determine if HRV data collected from ten subjects without diabetes would correlate to their glucose levels, and finally to determine if HRV data collected from six subjects diagnosed with Type 1 diabetes could be correlated to their glucose levels. The ultimate focus was to develop a complete portable system that could reliably detect changing glucose levels in a non invasive manner. A noninvasive and portable glucose monitor would greatly enhance the lives of diabetics.

The normal blood sugar range of the ten non-diabetic subjects was 70 mg/dL -150 mg/dL. These were the expected values for normal individuals with no history of diabetes or other glycemic ailments. After evaluating the population means and standard deviations, it was concluded that there were very little variations (coefficients of variation less than 7%) in the maximum, mode, minimum, median, average, and approximate entropy when compared to the rest of the HRV parameters Side by side box plots of mean, mode, and median reveal that within five of six subjects with diabetes, the normal and hyperglycemic ranges can be clearly distinguishable. Assuming normality, student t-tests reject the null hypothesis of equal means with 95% confidence with a p value < 0.005. Assuming non normality, Mann-Whitney tests reject the null hypothesis of equal means with 95% confidence with very small p values. Side by side box plots of RMSSD, NN50, and SD1 reveal that within two of six diabetics normal and hypoglycemic ranges can be clearly distinguishable. Assuming normality, student t-tests reject the null hypothesis of equal means with 95% confidence with a p value < 0.01. Assuming non normality, Mann-Whitney tests reject the null hypothesis of equal means with 95% confidence with a p value <0.007.

Language

en

Provenance

Received from ProQuest

File Size

235 pages

File Format

application/pdf

Rights Holder

MARCOS EFREN BOLANOS

Share

COinS