Method of adapting in-vitro models to aid in noninvasive glucose determination

1. A computer implemented method for noninvasively estimating an analyte concentration with an in-vivo instrument system, comprising the steps of:

providing a first model;

removing at least one interference from said first model to form a second model;

standardizing said in-vivo instrument system to said second model to generate a third model;

providing an in-vivo test set, comprising:

at least one in-vivo test signal; and

a reference analyte concentration corresponding with said in-vivo test signal;

applying said third model to said in-vivo test signal to generate a test value;

applying a correction to said third model using said test value and said reference analyte concentration to yield a corrected third model;

providing an in-vivo measurement signal; and

estimating and providing for use said analyte concentration using said corrected third model and said in-vivo measurement signal.

2. The method of claim 1 , wherein said first model comprises coefficients generated at least in part with an in-vitro data set.

3. The method of claim 2 , wherein said third model comprises a third set of coefficients.

4. The method of claim 3 , wherein said analyte concentration comprises a glucose concentration.

5. The method of claim 4 , wherein said first model comprises coefficients derived from data comprised of at least twenty percent in-vitro data.

6. The method of claim 5 , wherein said first model comprises coefficients derived from data comprised of at least eighty percent in-vitro data.

7. The method of claim 4 , wherein said step of applying a correction comprises the step of at least one of:

applying an offset; and

applying a scaling factor.

8. The method of claim 7 , further comprising:

repeating said steps of providing an in-vivo test set, applying said second model to said in-vivo test set, applying a correction, and applying an offset.

9. The method of claim 3 , wherein said step of removing comprises projecting said coefficients onto a null space of said interference.

10. The method of claim 9 , wherein said interference comprises at least one of:

a protein signal;

a fat signal;

a water signal;

a salt signal;

a thermal noise;

a specific tissue sample spectrum;

an individual;

a class of subjects; and

a cluster of data.

11. The method of claim 3 , wherein said step of standardizing comprises the step of at least one of:

smoothing;

interpolating;

scaling;

filtering;

performing an offset correction;

performing a bias correction;

normalizing;

performing direct standardization

performing piece-wise direct standardization;

performing a standard normal variate transformation;

performing multiplicative scatter correction;

performing orthogonal signal correction;

re-sampling; and

correcting wavelength.

12. The method of claim 11 , wherein said step of standardizing comprises at least three of:

smoothing;

interpolating;

scaling;

filtering;

performing offset correction

performing bias correction;

normalizing;

performing direct standardization

performing piece-wise direct standardization;

performing standard normal variate transformation;

performing multiplicative scatter correction;

performing orthogonal signal correction;

re-sampling; and

correcting wavelength.

13. The method of claim 3 , wherein said in-vivo test signal comprises a spectrum.

14. The method of claim 3 , wherein said second model comprises coefficients generated on a first instrument and said third model comprises coefficients generated on a second instrument.

15. The method of claim 14 , wherein said first instrument comprises a research grade spectrometer.

16. The method of claim 14 , wherein said second instrument comprises a production grade analyzer.

17. The method of claim 3 , wherein said in-vivo measurement signal comprises a spectrum.

18. The method of claim 3 , further comprising the step of:

housing at least one of said first model, said second model, and said third model in an analyzer.

19. The method of claim 18 , further comprising the step of:

providing an analyzer that comprises:

a base module;

a communication bundle with a first end and a second end, wherein said first end is connected to said base module;

a sample module, wherein said second end of said communication bundle is connected to said sample module; and

a processor.

20. A computer implemented method for noninvasive estimation of a sample constituent property, comprising the steps of:

providing a noninvasive signal;

providing a first model, wherein said first model comprises coefficients that are generated at least in part with an in-vitro data set, wherein said in-vitro data set comprises a spectrum of a tissue phantom having at least one optical parameter representative of said noninvasive signal in terms of photonic scattering and/or absorbance;

standardizing an in-vivo instrument system to said first model, wherein a second model is generated; and

estimating and providing for use said sample property by applying said second model to said noninvasive signal.

21. The method of claim 20 , wherein said step of standardizing comprises the step of at least one of:

smoothing;

interpolating;

scaling;

filtering;

performing offset correction

performing bias correction;

normalizing;

performing direct standardization

performing piece-wise direct standardization;

performing standard normal variate transformation;

performing multiplicative scatter correction;

performing orthogonal signal correction;

re-sampling; and

correcting wavelength.

22. The method of claim 21 , wherein said step of standardizing comprises the step of at least three of:

smoothing;

interpolating;

scaling;

filtering;

performing offset correction

performing bias correction;

normalizing;

performing direct standardization

performing piece-wise direct standardization;

performing standard normal variate transformation;

performing multiplicative scatter correction;

performing orthogonal signal correction;

re-sampling; and

correcting wavelength.

23. The method of claim 20 , wherein said sample property comprises a glucose concentration.

24. The method of claim 20 , wherein said data set comprises data that are collected with a first analyzer, and said noninvasive signal comprises a signal that is collected with a second analyzer.

25. The method of claim 24 , wherein said first analyzer comprises a research grade instrument.

26. The method of claim 24 , wherein said second analyzer comprises a production analyzer.

27. The method of claim 24 , wherein said second analyzer comprises a base module in a first container and a sample module in a second container.

28. The method of claim 20 , wherein said first model is generated with at least eighty percent in-vitro data.

29. An apparatus for noninvasive estimation of a sample constituent property from a noninvasive spectrum, comprising:

an analyzer comprising a base module, a sample module, and a model residing in said analyzer;

wherein said model comprises coefficients generated by standardizing an in-vivo system to an in-vitro data set,

wherein said in-vitro data set comprises a spectrum of a tissue phantom having at least one optical parameter representative of said noninvasive spectrum in terms of photonic scattering and/or absorbance; and

wherein said model is applied to said noninvasive spectrum for estimation of said sample constituent property.

30. The apparatus of claim 29 , wherein said base module resides in a first container and said sample module resides in a second container.

31. A computer implemented method for noninvasive estimation of a sample constituent property, comprising the steps of:

providing a first model, wherein said first model comprises coefficients that are generated at least in part with an in-vitro data set;

standardizing an in-vivo instrument system to said first model to generate a second model, wherein said second model comprises a second set of coefficients;

providing an in-vivo test set, comprising:

at least one in-vivo test signal; and

a reference sample concentration that is correlated with said in-vivo test signal;

applying said second model to said in-vivo test set to generate a test value;

providing an in-vivo measurement signal; and

estimating and providing for use said sample constituent property using said second model and said in-vivo measurement signal, wherein said step of estimating comprises multiplication of said in-vivo measurement signal by both a regression vector and a scaling factor resulting in a product that is adjusted with an offset.

32. The method of claim 31 , wherein said test signal comprises a spectrum.

33. The method of claim 31 , wherein said sample constituent property comprises a glucose concentration.

34. The method of claim 33 , further comprising the step of:

repeating said steps of providing an in-vivo test set, applying said second model to said in-vivo test set, applying a correction, and applying an offset.

35. The method of claim 33 , wherein said estimated glucose concentration is determined according to:

ŷ=xaW+b

where a is said scaling factor, b is said offset, x is said in-vivo test measurement signal, W is said regression vector of said second model, and ŷ is said estimated glucose concentration.

36. The method of claim 31 , wherein said first model comprises coefficients generated using a first instrument and said test signal comprises signals generated on a second instrument.

37. The method of claim 36 , wherein said first instrument comprises a research grade spectrometer.

38. The method of claim 36 , wherein said second instrument comprises a production grade spectrometer.

39. The method of claim 31 , wherein said first model comprises coefficients generated with a research grade spectrometer.

40. The method of claim 31 , wherein said in-vivo instrument system comprises a production grade analyzer.

41. The method of claim 31 , wherein said first model is generated with at least twenty percent in-vitro data.

42. The method of claim 41 , wherein said first model is generated with at least eighty percent in-vitro data.

43. The method of claim 31 , wherein said step of standardizing comprises the step of at least one of:

smoothing;

interpolating;

scaling;

filtering;

performing offset correction

performing bias correction;

normalizing;

performing direct standardization

performing piece-wise direct standardization;

performing standard normal variate transformation;

performing multiplicative scatter correction;

performing orthogonal signal correction;

re-sampling; and

correcting wavelength.

44. A computer implemented method for noninvasively estimating an analyte concentration, comprising the steps of:

providing a first calibration model;

removing at least one interference from said first model to form a second model, wherein said first model comprises coefficients derived from data comprised of at least twenty percent in-vitro data, wherein said step of removing comprises projecting said coefficients onto a null space of said interference;

providing an in-vivo signal; and

estimating and providing for use said analyte concentration using said second model and said in-vivo signal.

45. The method of claim 44 , wherein said first model comprises coefficients derived from data comprised of at least eighty percent in-vitro data.

46. The method of claim 44 , wherein said analyte concentration comprises a glucose concentration.

47. The method of claim 44 , wherein said step of removing comprises subtraction.

48. The method of claim 44 , wherein said interference comprises at least one of:

protein;

fat;

a specific tissue sample;

an individual;

a class of subjects; and

a cluster of data.

49. An apparatus for noninvasive estimation of a sample constituent property from a noninvasive spectrum, comprising:

an analyzer, comprising:

a base module;

a sample module; and

a model residing in said analyzer;

wherein said model comprises coefficients generated by standardizing an in-vivo system to a model of coefficients derived at least in part from an in-vitro data set;

wherein said in-vitro signal comprises a spectrum of a tissue phantom having at least one optical parameter representative of said noninvasive spectrum in terms of photonic scattering and/or absorbance; and

wherein said model is applied to said noninvasive spectrum to generate said sample constituent property.

50. The apparatus of claim 49 , wherein said model further comprises a correction to said model.

51. The apparatus of claim 50 , wherein said correction comprises at least one of:

an offset; and

a scaling factor.

52. The apparatus of claim 49 , wherein said base module resides in a first container and said sample module resides in a second container.

53. The apparatus of claim 52 , further comprising:

a communication bundle;

wherein said communication bundle interfaces said base module to said sample module.

54. A computer implemented method for noninvasively estimating a blood/tissue glucose concentration, comprising the steps of:

providing a noninvasive near-infrared signal;

providing a calibration model;

supplementing said calibration model with an in-vitro signal, wherein said in-vitro signal comprises a spectrum of a tissue phantom having at least one optical parameter representative of said noninvasive near-infrared signal in terms of photonic scattering and/or absorbance; and

estimating and providing for use said blood glucose concentration using said model and said noninvasive signal.

55. The method of claim 54 , further comprising the steps of:

providing an in-vivo test set, comprising:

at least one in-vivo test signal; and

a reference glucose concentration that is correlated with said in-vivo test signal;

applying said model to said in-vivo test signal to generate a test value; and

determining a correction to said model using said test value and said reference glucose concentration.

56. The method of claim 55 , further comprising the step of:

repeating said steps of providing an in-vivo test set; applying said model to said in-vivo test signal to generate a test value; and determining a correction to said model using said test value and said reference glucose concentration.

57. The method of claim 56 , further comprising the step of:

removing at least one interference from said model.

58. A computer implemented method for noninvasive estimation of a sample constituent property from an in-vivo instrument system, comprising the steps of:

providing a noninvasive signal;

providing a model, comprising coefficients generated at least in part with an in-vitro data set, wherein said in-vitro data set comprises a spectrum of a tissue phantom having at least one optical parameter representative of said noninvasive spectrum in terms of photonic scattering and/or absorbance;

standardizing said model to said in-vivo instrument system; and

estimating and providing for use said sample property by applying said model to said noninvasive signal.

59. The method of claim 58 , wherein said step of standardizing comprises the step of at least one of:

smoothing;

interpolating;

scaling;

filtering;

performing offset correction

performing bias correction;

normalizing;

performing direct standardization

performing piece-wise direct standardization;

performing standard normal variate transformation;

performing multiplicative scatter correction;

performing orthogonal signal correction;

re-sampling; and

correcting wavelength.

60. The method of claim 59 , wherein said step of standardizing comprises the step of at least three of:

smoothing;

interpolating;

scaling;

filtering;

performing offset correction

performing bias correction;

normalizing;

performing direct standardization

performing piece-wise direct standardization;

performing standard normal variate transformation;

performing multiplicative scatter correction;

performing orthogonal signal correction;

re-sampling; and

correcting wavelength.

61. The method of claim 58 , wherein said sample property comprises a glucose concentration.

62. The method of claim 58 , wherein said data set comprises data that are collected with a first analyzer and said noninvasive signal comprises a signal that is collected with a second analyzer.

63. The method of claim 62 , wherein said first analyzer comprises a research grade instrument.

64. The method of claim 62 , wherein said second analyzer comprises a production analyzer.

65. The method of claim 62 , wherein said second analyzer comprises a base module residing a first container and a sample module residing in a second container.

66. The method of claim 58 , wherein said model is generated with at least eighty percent in-vitro data.