System and method for estimating blood analyte concentration
There is provided a system and method for estimating blood analyte concentration using a non-invasive medical device. The method includes detecting light from a plurality of light sources and generating signals representative of observed absorption of the light from the plurality of light sources. Blood analyte concentrations are then estimated using support vector regression analysis.
1. A system for estimating blood analyte concentration comprising:
a sensor comprising:
a plurality of light sources configured to operate at different wavelengths; and
a detector configured to generate signals based on light from the plurality of light sources; and
a monitor coupled to the sensor configured to:
receive the signals generated by the detector;
determine the different wavelengths at which the plurality of light sources operate;
select a support vector regression function for the different wavelengths at which the plurality of light sources operate;
estimate blood analyte concentration using the selected support vector regression function in a support vector regression analysis of the received signals; and
output the estimated blood analyte concentration.
2. The system of claim 1 , wherein the plurality of light sources comprises three light sources operating at different wavelengths.
3. The system of claim 2 , wherein the three light sources comprise:
a first light source operating in the red region of the electromagnetic spectrum; and
second and third light sources operating in the IR region of the electromagnetic spectrum.
4. The system of claim 1 , wherein the monitor is configured to estimate the blood analyte concentration using an equation having the form
y
=
∑
i
α
(
i
)
κ
(
w
(
i
)
,
χ
)
+
b
,
where y is the blood analyte concentration and η represents the signals generated by the detector, the monitor comprising a memory configured to store support vectors w (i) , a kernel k, and calibration parameters α and b.
5. The system of claim 4 , wherein the kernel k a nonlinear kernel.
6. A method for non-invasively estimating blood analyte concentration comprising:
operating a plurality of light sources at different wavelengths;
detecting light from the plurality of light sources;
generating signals representative of observed absorption of the light from the plurality of light sources;
determining the different wavelengths at which the plurality of light sources are operating;
selecting a support vector regression function for the different wavelengths at which the plurality of light sources are operating; and
estimating blood analyte concentration using the selected support vector regression function in a support vector regression analysis of the generated signals.
7. The method of claim 6 , wherein the estimating blood analyte concentration comprises solving for an equation having a form:
y
=
∑
i
α
(
i
)
κ
(
w
(
i
)
,
χ
)
+
b
,
where y is the blood analyte concentration and x represents the signals representative of the observed absorption of the light, w (i) represents the support vectors, k is a kernel, and α and b are calibration parameters.
8. The method of claim 6 , wherein operating the plurality of light sources comprises operating three light sources comprising a first light source operating in the red region of the electromagnetic spectrum, and second and third light sources operating in the infrared region of the electromagnetic spectrum.
9. The method of claim 7 , wherein the kernel is a polynomial kernel.
10. The method of claim 7 , wherein the kernel is a non-linear kernel.
11. The method of claim 7 , wherein the kernel is a sigmoidal kernel.
12. The method of claim 7 , wherein the kernel is a Gaussian kernel.
13. A monitor comprising:
a memory configured to store one or more support vector regression functions for different wavelengths;
a processor configured to receive signals generated by a sensor, determine different wavelengths at which a plurality of light sources within the sensor operate, select a support vector regression function for the different wavelengths at which the plurality of light sources operate, and estimate blood analyte concentration using the selected support vector regression function in a support vector regression analysis of the received signals; and
a display configured to display the estimated blood analyte concentration.
14. The monitor of claim 13 , wherein the plurality of light sources comprises three light sources operating at different wavelengths.
15. The monitor of claim 14 , wherein the three light sources comprise a first light source operating in the red region of the electromagnetic spectrum, and second and third light sources operating in the IR region of the electromagnetic spectrum.
16. The monitor of claim 14 , wherein the monitor is configured to estimate the blood analyte concentration using an equation having the form
y
=
∑
i
α
(
i
)
κ
(
w
(
i
)
,
χ
)
+
b
,
y is the blood analyte concentration and η represents the signals generated by the sensor, and the memory is configured to store support vectors w (i) , a kernel k, and calibration parameters α and b.
17. The monitor of claim 16 , wherein the kernel k is a nonlinear kernel.
18. The monitor of claim 16 , wherein the kernel k is a polynomial kernel.
19. The monitor of claim 16 , wherein the kernel k is a sigmoidal kernel.
20. The monitor of claim 16 , wherein the kernel k is a Gaussian kernel.