Chemistry: molecular biology and microbiology – Measuring or testing process involving enzymes or... – Involving glucose or galactose
Reexamination Certificate
2001-07-23
2003-11-25
Chin, Christopher L. (Department: 1641)
Chemistry: molecular biology and microbiology
Measuring or testing process involving enzymes or...
Involving glucose or galactose
C204S400000, C204S403060, C422S082010, C422S082020, C435S025000, C435S176000, C435S287100, C435S817000, C436S149000, C436S150000, C436S151000, C436S518000, C436S525000, C436S806000
Reexamination Certificate
active
06653091
ABSTRACT:
FIELD OF THE INVENTION
The invention relates generally to a method and device for measuring the concentration of target chemical analytes present in a biological system. More particularly, the invention relates to a method and monitoring systems for predicting a concentration of an analyte using a series of measurements obtained from a monitoring system and a Mixtures of Experts (MOE) algorithm.
BACKGROUND OF THE INVENTION
The Mixtures of Experts model is a statistical method for classification and regression (Waterhouse, S., “
Classification and Regression Using Mixtures of Experts,
October 1997, Ph.D. Thesis, Cambridge University). Waterhouse discusses Mixtures of Experts models from a theoretical perspective and compares them with other models, such as, trees, switching regression models, modular networks. The first extension described in Waterhouse's thesis is a constructive algorithm for learning model architecture and parameters, which is inspired by recursive partitioning. The second extension described in Waterhouse's thesis uses Bayesian methods for learning the parameters of the model. These extensions are compared empirically with the standard Mixtures of Experts model and with other statistical models on small to medium sized data sets. Waterhouse also describes the application of the Mixtures of Experts framework to acoustic modeling within a large vocabulary speech recognition system.
The Mixtures of Experts model has been employed in protein secondary structure prediction (Barlow, T. W.,
Journal Of Molecular Graphics,
13(3), p. 175-183, 1995). In this method input data were clustered and used to train a series different networks. Application of a Hierarchical Mixtures of Experts to the prediction of protein secondary structure was shown to provide no advantages over a single network.
Mixtures of Experts algorithms have also been applied to the analysis of a variety of different kinds of data sets including the following: human motor systems (Ghahramani, Z. and Wolpert, D. M.,
Nature,
386(6623):392-395, 1997); and economic analysis (Hamilton, J. D. and Susmel, R.,
Journal of Economnetrics,
64(1-2):307-333, 1994).
SUMMARY OF THE INVENTION
The present invention provides a method and device (for example, a monitoring or sampling system) for continually or continuously measuring the concentration of an analyte present in a biological system. The method entails continually or continuously detecting a raw signal from the biological system, wherein the raw signal is specifically related to the analyte. A calibration step is performed to correlate the raw signal with a measurement value indicative of the concentration of analyte present in the biological system. These steps of detection and calibration are used to obtain a series of measurement values at selected time intervals. Once the series of measurement values is obtained, the method of the invention provides for the prediction of a measurement value using a Mixtures of Experts (MOE) algorithm.
The raw signal can be obtained using any suitable sensing methodology including, for example, methods which rely on direct contact of a sensing apparatus with the biological system; methods which extract samples from the biological system by invasive, minimally invasive, and non-invasive sampling techniques, wherein the sensing apparatus is contacted with the extracted sample; methods which rely on indirect contact of a sensing apparatus with the biological system; and the like. In preferred embodiments of the invention, methods are used to extract samples from the biological sample using minimally invasive or non-invasive sampling techniques. The sensing apparatus used with any of the above-noted methods can employ any suitable sensing element to provide the raw signal including, but not limited to, physical, chemical, electrochemical, photochemical, spectrophotometric, polarimetric, calorimetric, radiometric, or like elements. In preferred embodiments of the invention, a biosensor is used which comprises an electrochemical sensing element.
In one particular embodiment of the invention, the raw signal is obtained using a transdermal sampling system that is placed in operative contact with a skin or mucosal surface of the biological system. The sampling system transdermally extracts the analyte from the biological system using any appropriate sampling technique, for example, iontophoresis. The transdermal sampling system is maintained in operative contact with the skin or mucosal surface of the biological system to provide for continual or continuous analyte measurement.
In a preferred embodiment of the invention, a Mixtures of Experts algorithm is used to predict measurement values. The general Mixtures of Experts algorithm is represented by the following series of equations: where the individual experts have a linear form:
An
=
∑
i
=
1
n
An
i
w
i
(
1
)
wherein (An) is an analyte of interest, n is the number of experts, An
i
is the analyte predicted by Expert i; and w
i
is a parameter, and the individual experts An
i
are further defined by the expression shown as Equation (2)
An
i
=
∑
j
=
1
m
a
ij
p
j
+
z
i
(
2
)
wherein, An
i
is the analyte predicted by Expert i; P
j
is one of m parameters, m is typically less than 100; a
ij
are coefficients; and z
i
is a constant; and further where the weighting value, w
i
, is defined by the formula shown as Equation (3).
w
i
=
ⅇ
d
i
[
∑
k
=
1
n
ⅇ
d
k
]
(
3
)
where e refers to the exponential function and the d
k
(note that the d
i
in the numerator of Equation 3 is one of the d
k
) are a parameter set analogous to Equation 2 that is used to determine the weights w
i
. The d
k
are given by Equation 4.
d
k
=
∑
j
=
1
m
α
jk
P
j
+
ω
k
(
4
)
where &agr;
jk
is a coefficient, P
j
is one of m parameters, and where &ohgr;
k
is a constant.
Another object of the invention to use the Mixtures of Experts algorithm of the invention to predict blood glucose values. In one aspect, the method of the invention is used in conjunction with an iontophoretic sampling device that provides continual or continuous blood glucose measurements. In one embodiment the Mixtures of Experts algorithm is essentially as follows: where the individual experts have a linear form
BG=w
1
BG
1
+w
2
BG
2
+w
3
BG
3
(5)
wherein (BG) is blood glucose, there are three experts (n=3) and BG
i
is the analyte predicted by Expert i; w
i
is a parameter, and the individual Experts BG
i
are further defined by the expression shown as Equations 6, 7, and 8
BG
1
=p
1
(time)+
q
1
(active)+
r
1
(signal)+
s
1
(
BG|cp
)+
t
1
(6)
BG
2
=p
2
(time)+
q
2
(active)+
r
2
(signal)+
s
2
(
BG|cp
)+
t
2
(7)
BG
3
=p
3
(time)+
q
3
(active)+
r
3
(signal)+
s
3
(
BG|cp
)+
t
3
(8)
wherein, BG
i
is the analyte predicted by Expert i; parameters include, time (elapsed time since the sampling system was placed in operative contact with said biological system), active (active signal), signal (calibrated signal), and BG/cp (blood glucose value at a calibration point); p
i
, q
i
, r
i
, and s
i
are coefficients; and t
i
is a constant; and further where the weighting value, w
i
, is defined by the formulas shown as Equations 9, 10, and 11
w
1
=
ⅇ
d
1
ⅇ
d
1
+
ⅇ
d
2
+
ⅇ
d
3
(
9
)
w
2
=
ⅇ
d
2
ⅇ
d
1
+
ⅇ
d
2
+
ⅇ
d
3
(
10
)
w
3
=
ⅇ
d
3
ⅇ
d
1
+
ⅇ
d
2
+
ⅇ
d
3
(
11
)
where e refers to the exponential function and d
i
is a parameter set (analogous to Equations 6, 7, and 8) that are used to determine the weights w
i
, given by Equations 9, 10, and 11, and
d
1
=&tgr;
1
(time)+&bgr;
1
(active)+&ggr;
1
(signal)+&dgr;
1
(
BG|cp
)+&egr;
1
(12)
d
2
=&tgr;
2
(time)+&bgr;
2
(active)+&ggr;
2
(signal)+&dgr;
2
(
BG|cp
)+&egr;
2
(13)
d
3
=&tgr;
3
(time)+&bgr;
3
(ac
Dunn Timothy C.
Jayalakshmi Yalia
Kurnik Ronald T.
Lesho Matthew J.
Oliver Jonathan James
Chin Christopher L.
Cyngnus, Inc.
Fabian Gary R.
McClung Barbara G.
LandOfFree
Method and device for predicting physiological values does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and device for predicting physiological values, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and device for predicting physiological values will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3116218