Combinative multivariate calibration that enhances prediction ability through removal of over-modeled regions

• US 6,871,169 B1
• Filed: 08/01/2000
• Issued: 03/22/2005
• Est. Priority Date: 08/14/1997
• Status: Expired due to Fees

• Current Assignee: GLT Acquisition Corporation
• Original Assignee: Sensys Medical, Inc. (Sensys Medical Limited)
• Inventors: Thennadil, Suresh, Ruchti, Timothy L., Hazen, Kevin H.
• Primary Examiner(s): Broda, Samuel
• Application Number:
  US09/630,201
• Time in Patent Office: 1,694 Days
• Field of Search: 703/2, 703/12, 702 22- 32, 356/39
• US Class Current: 703/2
• CPC Class Codes: G01N 21/274 Calibration, base line adjustment, drift correction
• CPC Class Codes: G01N 21/3577 for analysing liquids, e.g. polluted water
• CPC Class Codes: G01N 21/359 using near infrared light

Abstract

A novel multivariate model for analysis of absorbance spectra allows for each wavelength or spectral region to be modeled with just enough factors to fully model the analytical signal without the incorporation of noise by using excess factors. Each wavelength or spectral region is modeled utilizing its own number of factors independently of other wavelengths or spectral regions. An iterative combinative PCR algorithm allows a different number of factors to be applied to different wavelengths. In an exemplary embodiment, a three-factor model is applied over a given spectral region. The residual of the three-factor model is calculated and used as the input for an additional five-factor model. Prior to the additional five factors being applied, some of the wavelengths are removed. This leads to a three-factor model over the first region and an eight-factor model over the second region. This analysis of residuals can be repeated such that a one to n factor model could be applied to any given wavelength, or rather any number of factors may be employed to model any given frequency or spectral region. A method of predicting concentration of a target analyte from sample spectra applies a calibration developed using the inventive PCR algorithm to a matrix of sample spectral to generate a vector of predicted concentrations for the target analyte.

Claims

1. A method of developing a calibration for predicting concentration of a target analyte in sample spectra, said method employing factor-based multivariate techniques, said method comprising the steps of:

A. providing a matrix of calibration spectra;

B. modeling at least one spectral region using a first selected number of factors;

C. removing said modeled regions from said spectral matrix, so that, a residual matrix is generated;

D. modeling at least one region of said residual matrix using a further selected number of factors, wherein said further selected number of factors may be equal to or different from said first selected number;

E. repeating steps B through D, using said residual matrix as an input matrix for step B, until the number of factors employed for all iterations is equal to an optimal number of factors required to model at least a portion of said sample spectrum; and

F. generating a vector of calibration coefficients wherein said vector constitutes said calibration;

wherein a specific number of factors models at least one region of a spectrum independently of the number of factors used to model other spectral regions, thus minimizing prediction error.

2. The method of claim 1, wherein said multivariate techniques include Partial Least Squares (PLS) analysis.

3. The method of claim 1, wherein said multivariate techniques include Principal Component Regression (PCR).

4. The method of claim 3, wherein said factors are selected from a matrix of loading vectors and a matrix of scores, said matrix calibration spectra=X₁, said matrix of loading vectors=T, and said scores matrix=V, so that

X₂=TV^T,

and wherein using Principal Component Analysis (PCA) is employed to decompose X₁ to generate T, and wherein X₁ comprises I samples and k wavelengths.

5. The method of claim 4, wherein said scores matrix, V, is generated by PCA.

6. The method of claim 5, wherein the number of factors used in the current iteration=x_n, with z=the number of factors used for an iteration and n=the current iteration, so that said first number of factors=z₁;

and wherein step B comprises the steps of;

G. setting T_n=T_u(;

, 1;

z_n); and

H. setting V_n=V_u(;

, 1;

z_n), where u=the number of factors in an optimal model.

7. The method of claim 6, wherein step C comprises the step of:

I. calculating a residual spectral matrix X₂ from X₁ according to;

X_n+1=X_n−

T_nV_n^T.

8. The method of claim 7, wherein step I further comprises the step of:

J. eliminating at least one high-noise region from said residual spectral matrix.

9. The method of claim 8, wherein step J comprises the optional step of:

K. calculating a mean residual spectrum according to;

Resi=mean(X₁−

T₁V₁^T).

10. The method of claim 9, wherein step J further comprises any of the steps of:

L. squaring said mean residual spectrum and removing at least one large value area; and

M. removing at least one region where raw spectra have high absorbance, wherein noise is related to absorbance.

11. The method of claim 10, further comprising the step of:

N. generating a final scores matrix by concatenating all previous scores matrices according to;

T_all=[T₁Y₂T₃ . . . T_n].

12. The method of claim 11, wherein set F comprises:

O. generating a vector of calibration coefficients, s, using multiple linear regression, wherein said vector s, constitutes said calibration, according to;

y=Ts, where y is a vector of i concentrations;

T^Ty=T^Ts;

(T^YT)<sup>−

1</sup>T^Ty=s; and

S=(T^YT)<sup>−

1</sup>y^Ty.

13. The method of claim 1, wherein said prediction error is due to overmodeling of noisy spectral regions.

14. A method of predicting concentration of a target analyte from a prediction data set based on a multivariate calibration comprising steps of:

A. providing a prediction data set, said prediction data set comprising a matrix of sample spectra;

B. generating a calibration by modeling a calibration data set according to an iterative, combinative algorithm, wherein a specific number of factors models at least one region of a spectrum independently of the number of factors used to model other spectral regions; and

C. applying said calibration to said prediction data set, so that a prediction of a target analyte concentration is produced;

wherein prediction error is minimized.

15. The method of claim 14, wherein step B comprises steps of:

D. modeling at least one spectral region from said calibration matrix using a first selected number of factors;

E. subtracting said modeled regions from said calibration matrix, so that a residual matrix is generated;

F. modeling at least one region from said residual matrix using a further selected number of factors, wherein said further selected number may be equal to or different from said first selected number;

G. repeating steps D through F, using said residual matrix as an input matrix for step B, until the number of factors employed for all iterations is equal to an optimal number of factors required to model at least a portion of said entire spectrum.

16. The method of claim 15, wherein said factors are selected from a matrix of loading vectors and a matrix of scores, said matrix calibration spectra=X₁, said matrix of loading vectors=T, and said-scores matrix=V, so that

X₁=TV^T,

and wherein using Principal Component Analysis (PCA) is employed to decompose X₁ to generate V, and wherein X₁ comprises i samples and k wavelengths.

17. The method of claim 16, wherein said scores matrix, T, is generated by regressing said matrix of loading vectors, V.

18. The method of claim 17, wherein the n umber of factors used in the current iteration=Z_n, with n=the current iteration and z=the number of factors used in an iteration, and wherein step E comprises the steps of:

H. setting T_n=T_u(;

, 1;

Z_n); and

I. setting V_n=V_u(;

, 1;

Z_n);

where u=the number of factors in an optimal model.

19. The method of claim 18, wherein step E comprises the step of:

J. calculating a residual spectral matrix X₂ from X₁ according to;

X_n+1=X_n−

T_nV_n^T.

20. The method of claim 19, wherein step E further comprises the step of:

K. eliminating at least one high-noise region from said residual spectral matrix.

21. The method of claim 20, wherein step K comprises the optional step of:

L. Calculating a mean residual spectrum according to;

Resi=mean(X_n−

T_nV_n^T).

22. The method of claim 21, wherein step K comprises any of the steps of:

M. squaring said mean residual spectrum and removing at least one large value area; and

N. removing at least one region where raw spectra have high absorbance, wherein noise is related to absorbance.

23. The method of claim 22, further comprising the step of,

O. generating a final scores matrix by concatenating all previous scores matrices according to:

T_all=[T₁T₂T₃ . . . T_s].

24. The method of claim 23, further comprising the step of:

P. generating a vector of calibration coefficients, s, using multiple linear regression, wherein said vector s, constitutes said calibration model, according to;

y=Ts, where y is a vector of ‘

I’

concentrations;

T^Ty=T^Ts;

(T^TT)<sup>−

1</sup>T^Ty−

s; and

s=(T^TT)<sup>−

1</sup>T^Ty.

25. The method of claim 24, wherein stop C comprises the steps of:

Q. generating a scores matrix T′

₁ for said matrix of sample spectra, where said sample matrix=X′

, according to;

T′

₁=X′

V₁;

R. computing a residual matrix for said sample matrix according to;

X′

₁=X′

₋

T′

₁V₁^T;

S. removing at least one high noise region from said residual of said sample matrix;

T. repeating steps R through T, using said residual of said sample matrix with said high noise regions removed as an input matrix for step R, until a total number of factors employed for all iterations is equal to the said optimal number of factors employed for said calibration set.

26. The method of claim 25, wherein step C further comprises the step of:

U. generating a final scores matrix, T_all, for said samples matrix by concatenating all previous scores matrices according to;

T_all=[T₁T₂T₃ . . . T″

_1V].

27. The method of claim 26, wherein step C further comprises the step of:

V. predicting concentration of said target analyte, Y_pred, according to;

y_pred=(T′

_all)s.

28. The method of claim 14, wherein said algorithm comprising an iterative, combinative PCR algorithm.

29. The method of claim 14, wherein said algorithm comprises an iterative, combinative PLS algorithm.

30. The method of claim 14, wherein said prediction error is due to overmodeling of noisy spectral regions.

31. An iterative, combinative algorithm for modeling a set of data, said algorithm comprising the steps of.

A. providing a data matrix containing an analytical signal, said data matrix comprising a plurality of samples and a plurality of variables;

B. modeling at least one region of said data matrix, a region constituting at least one of said plurality of values, using a selected number of factors;

C. subtracting said modeled regions from said data matrix, so that a residual matrix is generated;

D. modeling at least one further region from said residual matrix after removal of a further selected number of values, wherein said further selected number equal to or different from said first number; and

E. repeating steps B through D, using said residual matrix as an input matrix for step B, until the number of factors employed for all iterations is equal to an optimal number of factors required to model said entire data set;

wherein a specific number of factors models at least one region of said data matrix independently of the number of factors used to model other regions of said data matrix so that said analytical signal is modeled.

32. The algorithm of claim 31, wherein said factors are selected from a matrix of loading vectors and a matrix of scores, said data matrix of =X₁, said matrix of loading vectors=T, and said scores matrix=V, so that

X₁=TV^T,

and wherein using Principal Component Analysis (PCA) is employed to decompose X₁ to generate V, and wherein X₁ comprises i samples and k variables.

33. The algorithm of claim 32, wherein said scores matrix, V, is generated by regressing said matrix of loading vectors, T.

34. The algorithm of claim 33, wherein the number of factors used in the current iteration=z_n, with z=the number of factors used in an iteration, and n=the current iteration, so that said first number of factors=z₁, and wherein step B comprises the steps of:

F. setting T_n=T_u(;

, 1;

z_n); and

G. setting V_n=V_u(;

, 1;

z_n), where u=the number of factors in an optimal model.

35. The algorithm of claim 34, wherein step C comprises the step of:

H. calculating a residual matrix X₂ from X₁ according to;

X_n+z=X_n−

T_nV_n^T.

36. The algorithm of claim 35, wherein step H further comprises the step of:

I. eliminating at least one high-noise region from said residual matrix.

37. The algorithm of claim 36, wherein step I comprises the optional step of:

J. calculating a mean residual matrix according to;

Resi=mean(X_n−

T_nV_n^T).

38. The algorithm of claim 37, wherein step I further comprises the step of:

.

K. squaring said mean residual matrix and removing at least one large value region.

39. The algorithm of claim 37, further comprising the step of:

L. generating a final scores matrix by concatenating all previous scores matrices according to;

T_all=[T₁T₂T₃ . . . T_s].

40. The algorithm of claim 39, further comprising the step of:

M. generating a vector of calibration coefficients, s, using Multiple Linear regression, wherein said vectors, s, constitutes a calibration, according to;

y=Ts, where y is a vector of I known values related to a predetermined parameter within said data set, said data set constituting a calibration data set;

T^Ty=s;

(T^TT)<sup>−

1</sup>T^Ty=s; and

S=(T^TT)<sup>−

1</sup>T^Ty.

41. The algorithm of claim 40, further comprising the step of:

N. applying said calibration to a prediction data set, said prediction data set constituting a matrix, wherein actual values for said predetermined parameter are unknown, to generate a vector, y_pred, of predicted values for said parameter.

42. The algorithm of claim 41, wherein step N comprises the steps of:

O. generating a scores matrix T′

₁ for said matrix of prediction data, where said matrix of prediction data=X′

, according to;

T′

₁=X′

V₁;

P. computing a residual matrix for said matrix X′

according to;

X′

₁=X′

₋

T₁V₁^T;

Q. removing at least one high noise region from said residual of said matrix X′

; and

R. repeating steps O through Q, using said generated residual with said high noise regions removed as an input matrix for step V, until a total number of factors employed for all iterations is equal to the said optimal number of factors employed for said calibration set.

43. The a algorithm of claim 42, wherein step N further comprises the step of:

S. generating a final scores matrix, T′

_all, for said prediction data set by concatenating all previous scores matrices according to;

T′

_all=[T′

₁T′

₂T′

₃ . . . T′

_s].

44. The algorithm of claim 43, wherein step N further comprises the step of:

T. generating said vector, Y_pred, according to;

y_pred=(T′

_all)s.

45. The method of claim 31, wherein overmodeling of noisy regions present in said data matrix is minimized.

46. The method of claim 31, wherein said analytical signal comprises a spectroscopic signal.