Apparatus and methods for mitigating the effects of foreign interferents on analyte measurements in spectroscopy

• US 7,756,558 B2
• Filed: 12/19/2005
• Issued: 07/13/2010
• Est. Priority Date: 05/24/2004
• Status: Active Grant

• Current Assignee: Rockley Photonics Limited
• Original Assignee: TruTouch Technologies, Inc.
• Inventors: Hendee, Shonn, Ridder, Trent
• Primary Examiner(s): WINAKUR, ERIC FRANK
• Application Number:
  US11/305,964
• Publication Number:
  US 20070142720A1
• Time in Patent Office: 1,667 Days
• Field of Search: 600309-344
• US Class Current: 600/310
• CPC Class Codes: A61B 5/14532 for measuring glucose, e.g. by tissue impedance measurement
• CPC Class Codes: A61B 5/7264 Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems neural networks per se G06N3/00; expert systems per se G06N5/00

Abstract

The present invention includes apparatuses and methods for mitigating the effects of foreign interferents on analyte measurements. The present invention comprises several interferent mitigation steps. Examples include sample cleaning procedures, detection of the presence of interferents, determination of the identity of interferents, and modification or selection of a multivariate calibration model to mitigate the effects of one or more interferents on analyte measurements. The interferent mitigation steps of the present invention can be applied individually, and in some embodiments can be applied in combination. Some examples of relevant analyte measurements include the noninvasive determination of the presence or concentration of alcohol, glucose, urea, byproducts of alcohol metabolism, and substances of abuse.

Claims

1. A method for determining an analyte property in a biological subject, comprising:

a) Determining a spectrum, comprising information related to the response of a portion of the subject to incident radiation;

b) Determining from the spectrum without a contemporaneous reference value whether an interferent affected the spectrum (an “

interferent condition”

), and, if so, performing an interferent mitigation action;

c) Determining the analyte property in the biological subject from the spectrum after steps a) and b).

2. A method as in claim 1, wherein determining from the spectrum whether an interferent affected the spectrum comprises determining an outlier metric of the spectrum, and determining whether the outlier metric has a value corresponding to an interferent condition.

3. A method as in claim 1, wherein determining from the spectrum whether an interferent affected the spectrum comprises determining the analyte property in the biological subject, and determining whether the analyte property has a value corresponding to interferent conditions.

4. A method as in claim 1, wherein determining from the spectrum whether an interferent affected the spectrum comprises applying to the spectrum a model relating a spectrum to presence of an interferent, and from the result determining whether the spectrum indicates an interferent condition.

5. A method as in claim 1, wherein the spectrum comprises response of a portion of the subject at first and second paths through the portion of the subject, and wherein determining from the spectrum whether an interferent affected the spectrum comprises determining an interferent condition from comparison of the spectrum at the first path with the spectrum at the second path.

6. A method as in claim 1, wherein determining the analyte property in the biological subject comprises determining the analyte property from a model relating a spectrum to the analyte property, and wherein the interferent mitigation action comprises selecting a model for determination of the analyte property that is resistant to interferent effects.

7. A method as in claim 1, wherein determining the analyte property in the biological subject comprises determining the analyte property from a model relating a spectrum to the property and wherein the interferent mitigation action comprises changing a model for determination of the analyte property in a way that decreases the effect of the interferent on the determination of the analyte property.

8. A method as in claim 1, wherein the analyte property comprises the concentration or presence of alcohol, the biological subject comprises human tissue, and the spectrum comprises the response of the tissue to near-infrared light.

9. A method as in claim 8, wherein determining the analyte property in the biological subject after the interferent mitigation action comprises applying a multivariate model relating alcohol concentration or presence to a spectrum of response to near-infrared light.

10. A method of determining an analyte property in a biological subject when there is no interferent present, comprising:

a) Determining a spectrum of a portion of the subject, where the spectrum comprises information regarding the response at a plurality of wavelengths of the portion of the subject to incident radiation;

b) Determining from the spectrum without a contemporaneous reference value whether an interferent is present, where an interferent is any foreign substance that unacceptably compromises the quality of the determination of the characteristic;

c) If there is no interferent present, then determining the analyte property from the spectrum.

11. A method as in claim 10, wherein determining from the spectrum whether an interferent is present comprises at least one of:

a) Determining whether the spectral F-ratio of the spectrum relative to a plurality of calibration spectra has a value that indicates an interferent is present;

b) Determining whether a multidimensional distance between the spectrum and a calibration spectrum has a value that indicates an interferent is present;

c) Determining whether the Mahalanobis distance from the spectrum to a plurality of calibration spectra has a value that indicates an interferent is present.

12. A method as in claim 10, wherein determining from the spectrum whether an interferent is present comprises applying to the spectrum at least one of the following methods of analysis:

a neural network, residuals of principal component analysis, and k-nearest neighbors.

13. A method as in claim 10, further comprising classifying the interferent, if one is determined to be present.

14. A method as in claim 12, further comprising communicating the classification of the interferent to a user.

15. A method of determining an analyte property in a biological subject, comprising:

a) Determining a spectrum of a portion of the subject, where the spectrum comprises information regarding the response at a plurality of wavelengths of the portion of the subject to incident radiation;

b) Determining from the spectrum whether an interferent is present, where an interferent is any foreign substance that unacceptably compromises the quality of the determination of the property;

c) If an interferent is present, then classifying the interferent from the spectrum;

d) Selecting a method of determining the analyte property from the determination of the presence of an interferent and, if an interferent is present, also from the classification of the interferent;

e) Determining the analyte property from the spectrum and from the selected method.

16. A method as in claim 15, wherein classifying the interferent comprises at least one of:

k-nearest neighbors, principal components analysis, and a neural network.

17. A method as in claim 15, wherein selecting a method comprises using a multivariate model to determine the analyte property, wherein, if an interferent is detected, a multivariate model resistant to the effect of the classified interferent is used.

18. A method as in claim 17, wherein the multivariate model resistant to the effect of the classified interferent comprises a model generated from data including the presence of the classified interferent.

19. A method as in claim 17, wherein the multivariate model resistant to the effect of the classified interferent comprises a multivariate model modified using orthogonal signal correction.

20. A method as in claim 17, wherein the multivariate model resistant to the effect of the classified interferent comprises a multivariate model modified using a model modification procedure.

21. A method as in claim 15, wherein selecting a method comprises using a multivariate model to determine the analyte property, wherein, if an interferent is detected, a base multivariate model is modified before use to be resistant to the effect of the classified interferent.

22. A method as in claim 21, wherein the base multivariate model is modified using orthogonal signal correction.

23. A method as in claim 21, wherein the base multivariate model is modified using a model modification procedure.

24. A method as in claim 15, wherein selecting a method of determining the analyte property from the determination of the presence of an interferent comprises selecting a method without a contemporaneous reference value of the analyte property.