U.S. patent application number 13/935151 was filed with the patent office on 2013-11-07 for method for evaluating a set of measurement data from an oral glucose tolerance test.
This patent application is currently assigned to ROCHE DIAGNOSTICS OPERATIONS, INC.. The applicant listed for this patent is Ortrud Quarder, Arnulf Staib, Gerhard Werner. Invention is credited to Ortrud Quarder, Arnulf Staib, Gerhard Werner.
Application Number | 20130297224 13/935151 |
Document ID | / |
Family ID | 45495886 |
Filed Date | 2013-11-07 |
United States Patent
Application |
20130297224 |
Kind Code |
A1 |
Staib; Arnulf ; et
al. |
November 7, 2013 |
Method for evaluating a set of measurement data from an oral
glucose tolerance test
Abstract
A method is provided for evaluating a set of measurement data
from an oral glucose tolerance test. The method may include
calculating a similarity measure that quantifies the similarity
between a time profile of the series of measured data of the
glucose concentration and a corresponding glucose reference
profile. The method may include calculating a further similarity
measure that quantifies the similarity between the profile of the
series of measured values of the further analyte concentration and
the corresponding analyte sample profile, wherein the data set is
represented by a point in a vector space that comprises coordinate
axes that are formed by the similarity measures, whereby the
coordinates of said point contain the calculated values of the
similarity measures. The method also may include evaluating the
position of the point with respect to reference points, which each
represent a defined state of health, in order to calculate a
parameter that specifies the state of the glucose metabolism of the
patient.
Inventors: |
Staib; Arnulf; (Hepperheim,
DE) ; Quarder; Ortrud; (Heidelberg, DE) ;
Werner; Gerhard; (Weinheim, DE) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Staib; Arnulf
Quarder; Ortrud
Werner; Gerhard |
Hepperheim
Heidelberg
Weinheim |
|
DE
DE
DE |
|
|
Assignee: |
ROCHE DIAGNOSTICS OPERATIONS,
INC.
Indianapolis
IN
|
Family ID: |
45495886 |
Appl. No.: |
13/935151 |
Filed: |
July 3, 2013 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
PCT/EP2011/006500 |
Dec 22, 2011 |
|
|
|
13935151 |
|
|
|
|
Current U.S.
Class: |
702/21 |
Current CPC
Class: |
G16C 99/00 20190201;
G09B 23/28 20130101; G01N 2800/00 20130101 |
Class at
Publication: |
702/21 |
International
Class: |
G06F 19/00 20060101
G06F019/00 |
Foreign Application Data
Date |
Code |
Application Number |
Jan 7, 2011 |
EP |
11000081.7 |
Claims
1. A method for evaluating a set of measurement data from an oral
glucose tolerance test, whereby the set of measurement data
includes a series of measurement data of the glucose concentration
and, in addition, at least one series of measurement data of a
further analyte concentration, the method comprising: calculating,
by at least one computing device, a similarity measure that
quantifies the similarity between a time profile of the series of
measured data of the glucose concentration and a corresponding
glucose reference profile, wherein the calculation of the
similarity measure uses the series of measured data of the glucose
concentration and one each of several predefined glucose reference
profiles; calculating, by the at least one computing device, one
value each of a further similarity measure that quantifies the
similarity between the profile of the series of measured values of
the further analyte concentration and the corresponding analyte
sample profile, wherein the calculation of one value each of a
further similarity measure uses the series of measured values of
the further analyte concentration and one each of several
predefined analyte reference profiles, wherein the data set is
represented by a point in a vector space that comprises coordinate
axes that are formed by the similarity measures, whereby the
coordinates of said point contain the calculated values of the
similarity measures; and evaluating, by the least one computing
device, the position of the point with respect to reference points,
which each represent a defined state of health, in order to
calculate a parameter that specifies the state of the glucose
metabolism of the patient.
2. The method of claim 1, wherein the step of evaluating, by the
least one computing device, the position of the point
characterizing the set of measurement data is evaluated with
respect to the reference points by projecting the point onto a norm
trajectory which follows a disease progression in said vector space
from a healthy normal patient via a pre-diabetic conditions to a
diabetic disease and contains at least a fraction of the reference
points, wherein the length of a section of the trajectory from the
start of the trajectory to the point of the trajectory onto which
the point representing the set of measurement data was projected is
used to determine the parameter specifying the state of glucose
metabolism.
3. The method of claim 2, wherein the vector space comprises
multiple norm trajectories, each of which specify different disease
progressions from a healthy normal patient via a pre-diabetic
condition to an insulin-dependent diabetic disease, whereby the
point characterising the set of measurement data is projected onto
the norm trajectory situated at the smallest distance from it.
4. The method of claim 3, wherein the point characterizing the set
of measurement data is, in addition, also projected onto a second
norm trajectory situated at the second smallest distance from
it.
5. The method of claim 1, wherein the concentration profiles are
normalized before calculating the similarity measures.
6. The method of claim 1, wherein the similarity measures are
calculated as scalar product of vectors, whereby one of the vectors
is determined from the corresponding series of measured values and
the other vector is determined from the corresponding sample
profile.
7. The method of claim 5, wherein the similarity measures are each
calculated as scalar product of two normalized vectors.
8. The method of claim 1, wherein a norm of a vector formed from
the series of measured values of the glucose concentration is used
as a further coordinate of the vector space.
9. The method of claim 1, wherein a norm of a vector formed from
the series of measured values of the further analyte concentration
is used as further coordinate of the vector space.
10. The method of claim 1, wherein at least one coordinate axis of
the vector space specifies the value of a biometric or genetic
variable that is measured independent of a concentration
measurement.
11. The method of claim 10, wherein the biometric variable is the
body mass index, fraction of body fat, waist-to-hip ratio, blood
pressure or heart rate.
12. The method of claim 1, wherein the further analyte
concentration is the concentration of a secretory hormone.
13. The method of claim 1, wherein at least one coordinate axis of
the vector space specifies the concentrations of a metabolite that
shows no or little change on the time scale of an oral glucose
tolerance test.
14. A non-transitory computer-readable medium comprising:
executable instructions such that when executed by at least one
processor cause the at least one processor to: calculate a
similarity measure that quantifies the similarity between a time
profile of the series of measured values of the glucose
concentration and a corresponding glucose reference profile,
wherein the calculation of the similarity measure uses the series
of measured values of the glucose concentration and one each of
several predefined glucose reference profiles; calculate one value
each of a further similarity measure that quantifies the similarity
between the profile of the series of measured values of the further
analyte concentration and the corresponding analyte sample profile,
wherein the calculation of one value each of a further similarity
measure uses the series of measured values of the further analyte
concentration and one each of several predefined analyte reference
profiles, wherein the data set is represented by a point in a
vector space that comprises coordinate axes that are formed by the
similarity measures, whereby the coordinates of said point contain
the calculated values of the similarity measures; and evaluate the
position of the point with respect to reference points, which each
represent a defined state of health, in order to calculate a
parameter that specifies the state of the glucose metabolism of the
patient.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] The present U.S. utility application is related to and
claims the priority benefit to patent cooperation treaty patent
application serial no. PCT/EP2011/006500, filed Dec. 22, 2011,
which claims priority to European patent application no.
11000081.7, filed Jan. 7, 2011. The contents of each of these
applications are hereby incorporated by reference in their entirety
into this disclosure.
TECHNICAL FIELD
[0002] The present specification generally relates to methods for
evaluating a set of measurement data from an oral glucose tolerance
test.
BACKGROUND
[0003] In an oral glucose tolerance test, an oral glucose solution
is administered to a fasting patient and then the glucose
concentration in the blood of the patient is measured at various
time intervals. Usually, the profile of the blood sugar
concentration is measured for a period of 2 hours. However, a
glucose tolerance test can be carried out just as well for a
shorter or longer period of time.
[0004] Determining the glucose concentration profile in response to
the intake of glucose allows anomalies of glucose metabolism to be
recognised. Oral glucose tolerance tests therefore allow impaired
glucose utilisation to be detected and diabetes to be
diagnosed.
SUMMARY
[0005] The present disclosure comprises methods for evaluating a
set of measurement data from an oral glucose tolerance test.
[0006] In at least one embodiment of the present disclosure, a
method for evaluating a set of measurement data from an oral
glucose tolerance test is provided, whereby the set of measurement
data includes a series of measurement data of the glucose
concentration and, in addition, at least one series of measurement
data of a further analyte concentration. In at least one
embodiment, the method comprises calculating, by at least one
computing device, a similarity measure that quantifies the
similarity between a time profile of the series of measured data of
the glucose concentration and a corresponding glucose reference
profile, wherein the calculation of the similarity measure uses the
series of measured data of the glucose concentration and one each
of several predefined glucose reference profiles. Further, the
method comprises calculating, by the at least one computing device,
one value each of a further similarity measure that quantifies the
similarity between the profile of the series of measured values of
the further analyte concentration and the corresponding analyte
sample profile, wherein the calculation of one value each of a
further similarity measure uses the series of measured values of
the further analyte concentration and one each of several
predefined analyte reference profiles, wherein the data set is
represented by a point in a vector space that comprises coordinate
axes that are formed by the similarity measures, whereby the
coordinates of said point contain the calculated values of the
similarity measures. Additionally, the method comprises evaluating,
by the least one computing device, the position of the point with
respect to reference points, which each represent a defined state
of health, in order to calculate a parameter that specifies the
state of the glucose metabolism of the patient.
[0007] In at least one embodiment of the method, the method may
comprise the step of evaluating, by the least one computing device,
the position of the point characterizing the set of measurement
data is evaluated with respect to the reference points by
projecting the point onto a norm trajectory which follows a disease
progression in said vector space from a healthy normal patient via
a pre-diabetic conditions to a diabetic disease and contains at
least a fraction of the reference points, wherein the length of a
section of the trajectory from the start of the trajectory to the
point of the trajectory onto which the point representing the set
of measurement data was projected is used to determine the
parameter specifying the state of glucose metabolism.
[0008] In at least one embodiment of the method, the vector space
comprises multiple norm trajectories, each of which specify
different disease progressions from a healthy normal patient via a
pre-diabetic condition to an insulin-dependent diabetic disease,
whereby the point characterizing the set of measurement data is
projected onto the norm trajectory situated at the smallest
distance from it.
[0009] In at least one embodiment of the method, the point
characterizing the set of measurement data is, in addition, also
projected onto a second norm trajectory situated at the second
smallest distance from it.
[0010] In at least one embodiment of the method, the concentration
profiles are normalized before calculating the similarity
measures.
[0011] In at least one embodiment of the method, the similarity
measures are calculated as scalar product of vectors, whereby one
of the vectors is determined from the corresponding series of
measured values and the other vector is determined from the
corresponding sample profile.
[0012] In at least one embodiment of the method, the similarity
measures are each calculated as scalar product of two normalized
vectors.
[0013] In at least one embodiment of the method, a norm of a vector
formed from the series of measured values of the glucose
concentration is used as a further coordinate of the vector
space.
[0014] In at least one embodiment of the method, a norm of a vector
formed from the series of measured values of the further analyte
concentration is used as further coordinate of the vector
space.
[0015] In at least one embodiment of the method, at least one
coordinate axis of the vector space specifies the value of a
biometric or genetic variable that is measured independent of a
concentration measurement.
[0016] In at least one embodiment of the method, the biometric
variable is the body mass index, fraction of body fat, waist-to-hip
ratio, blood pressure or heart rate.
[0017] In at least one embodiment of the method, the further
analyte concentration is the concentration of a secretory
hormone.
[0018] In at least one embodiment of the method, at least one
coordinate axis of the vector space specifies the concentrations of
a metabolite that shows no or little change on the time scale of an
oral glucose tolerance test.
[0019] In at least one embodiment of the present disclosure, a
non-transitory computer-readable medium is provided that comprises
executable instructions such that when executed by at least one
processor cause the at least one processor to: calculate a
similarity measure that quantifies the similarity between a time
profile of the series of measured values of the glucose
concentration and a corresponding glucose reference profile,
wherein the calculation of the similarity measure uses the series
of measured values of the glucose concentration and one each of
several predefined glucose reference profiles; calculate one value
each of a further similarity measure that quantifies the similarity
between the profile of the series of measured values of the further
analyte concentration and the corresponding analyte sample profile,
wherein the calculation of one value each of a further similarity
measure uses the series of measured values of the further analyte
concentration and one each of several predefined analyte reference
profiles, wherein the data set is represented by a point in a
vector space that comprises coordinate axes that are formed by the
similarity measures, whereby the coordinates of said point contain
the calculated values of the similarity measures; and evaluate the
position of the point with respect to reference points, which each
represent a defined state of health, in order to calculate a
parameter that specifies the state of the glucose metabolism of the
patient.
[0020] These and additional features provided by the embodiments
described herein will be more fully understood in view of the
following detailed description, in conjunction with the
drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
[0021] The embodiments set forth in the drawings are illustrative
and exemplary in nature and not intended to limit the subject
matter defined by the claims. The following detailed description of
the illustrative embodiments can be understood when read in
conjunction with the following drawings, where like structure is
indicated with like reference numerals and in which:
[0022] FIG. 1 shows a flowchart depicting the exemplary steps of at
least one embodiment of the method of the present disclosure;
[0023] FIG. 2 shows an example of a glucose (g) concentration
profile of an oral glucose tolerance test for different groups of
patients;
[0024] FIG. 3 shows an example of a C-peptide (C) concentration
profile in an oral glucose tolerance test for different patient
groups;
[0025] FIG. 4 shows an example of a proinsulin (P) profile in a
glucose tolerance test for different patient groups;
[0026] FIG. 5 shows data points representing the results of oral
glucose tolerance tests for three different patient groups, in a
three-dimensional vector space;
[0027] FIG. 6 shows another view related to FIG. 5; and
[0028] FIG. 7 shows an example of a norm trajectory with data
points from oral glucose tolerance tests of different patient
groups, according to at least one embodiment of the present
disclosure.
DETAILED DESCRIPTION
[0029] For the purposes of promoting an understanding of the
principles of the present disclosure, reference will now be made to
the embodiments illustrated in the drawings, and specific language
will be used to describe the same. It will nevertheless be
understood that no limitation of the scope of this disclosure is
thereby intended.
[0030] Generally, the present disclosure provides methods for
performing an evaluation of measurement data from oral glucose
tolerance tests.
[0031] The efficacy of an oral glucose tolerance test can be
improved by measuring the profile of further analytes in a body
fluid of the patient, usually of blood and/or interstitial fluid in
addition to the glucose concentration profile. Secretory hormones,
such as insulin, pro-insulin, glucagon or C-peptide, are
particularly well-suited. For this reason, a method according to
the present disclosure evaluates a set of measurement data from an
oral glucose tolerance test, where the set of measurement data
includes a series of measurement data of the glucose concentration
and, in addition, at least one series of measurement data of a
further analyte concentration.
[0032] Turning to FIG. 1, a flowchart is depicted which shows the
steps of an exemplary embodiment of a method 100 of the present
disclosure. Initially, a similarity measure may be calculated from
the series of measured values of the glucose concentration and one
each of several predefined glucose reference profiles (exemplary
calculating step 110). The similarity measure quantifies the
similarity between the profile of the series of measured values of
the glucose concentration and the corresponding glucose reference
profile.
[0033] For example, the expected curve of the glucose concentration
profile in a glucose tolerance test of a patient with a certain
state of health can be used as glucose reference profile.
Corresponding reference profiles can also be determined by means of
testing sample patients who have already been diagnosed by other
means, in particular for a fully healthy state of health (H-NG), a
patient with type II diabetes (DT2), a normoglycaemic sample
patient with metabolic syndrome (MS-NG), a sample patient with
metabolic syndrome and impaired glucose tolerance (MS-IGT), a
sample patient with metabolic syndrome and impaired fasting glucose
(MS-IFG), and a sample patient, in whom the elevated fasting
glucose is combined with impaired glucose tolerance (MS-CGI). The
mathematic description of said profile curves is quite laborious
and the evaluation therefore requires a major computational
effort.
[0034] Exemplary glucose reference profiles may also be functions
that are linear over time, or at least over sections thereof. Even
simple functions of this type alone allow sections of the time
profile of the glucose concentration to be characterized for a
certain state of health. In particular, functions that are linear
or linear over sections thereof and describe, only approximately,
the increase or decrease of the glucose concentration during a part
of the time period of a glucose tolerance test for a certain state
of health can be used.
[0035] However, the similarity of the reference profiles to an
actual profile is not obligatory, since, for example, it is also
feasible to use a set of sufficiently different functions to
approximate realistic time profiles, for example by linear
combination, such as in the form of a Fourier series. A suitable
set of functions can consist, for example, of polynomials, in
particular polynomials of the type of x.sup.n.
[0036] The series of measured values of the glucose concentration
and a glucose reference profile may be used according to the
present disclosure to calculate a value of a similarity measure.
Basically, any calculation rule is suitable that yields a result
describing the coincidence between the series of measurement data
of the glucose concentration and the reference profile that is
being considered.
[0037] For example, a correlation coefficient, in particular a
Pearson product-moment correlation coefficient, can be used as a
similarity measure. However, the similarity measure may also be
calculated as the scalar product of two vectors, wherein one of the
two vectors is determined from the corresponding series of measured
values and the other vector is determined from the corresponding
reference profile. For instance, the individual measured values,
g.sub.1, g.sub.2, g.sub.3 to g.sub.n, which were determined for
consecutive time points t.sub.1, t.sub.2 to t.sub.n, can be used as
components of a vector. In like manner, a vector can also be formed
from a reference profile in that the concentration value of the
reference profile at the relevant time point t.sub.1, t.sub.2 to
t.sub.n is used as vector component.
[0038] In this context, normed concentration profiles, for example
as scalar product of normed vectors, may be used to calculate the
similarity measures. By this means, the mathematic effort can be
simplified and the descriptive value of the similarity measure can
be increased, since only the relative glucose concentration profile
and thus the shape of the profile curve is being considered.
[0039] Turning back to FIG. 1, according to at least one embodiment
of the present disclosure, the series of measured values of a
further analyte concentration and each of several predefined
analyte reference profiles may be used to calculate a value of a
further similarity measure that quantifies the similarity between
the profile of the series of measured values of the further analyte
concentration and the corresponding analyte reference profile
(exemplary calculating step 120). The profile curve of the analyte
concentration of a patient with a certain state of health expected
in a glucose tolerance test can be used as analyte reference
profiles. In at least one exemplary embodiment, the analyte
reference profiles use functions that are linear, at least over
sections thereof.
[0040] In this context, the similarity measure with respect to the
glucose concentration profile and the profile of the further
analyte concentration are not necessarily calculated for the same
and/or all available states of health, because significant
differences in the profile of the analyte concentration between all
states of health are not observed with all analytes. Pre-diabetic
states of health and early stages of a diabetic disease can
coincide over large parts of the analyte or glucose concentration
profile such that it is feasible to forego the calculation of a
similarity measure for the corresponding sections of the
concentration profile and the corresponding states of health
without lessening the significance of the robustness.
[0041] According to at least one embodiment of the present
disclosure, the set of measured data from an oral glucose tolerance
test and various reference profiles are first used to calculate a
set of similarity measures, i.e. a set of variables which each are
a measure of the similarity between the profile of the series of
measured data and a reference profile.
[0042] In a further step of at least one exemplary method of the
present disclosure, a vector space comprising coordinate axes that
are formed by the similarity measures is considered. The data set
of an individual with unknown state of health is characterized by a
point in the vector space, whereby the coordinates of said point
contain the calculated values of the similarity measures.
Accordingly, if, for example, a first coordinate axis of the vector
space is formed by the similarity measure of the time profile of
the series of measured values of the glucose concentration and a
first glucose reference profile, the value of that similarity
measure is the first coordinate of the point representing the set
of data.
[0043] In at least one exemplary method of the present disclosure,
the position of this point may in at least one embodiment be
evaluated with respect to reference points, which each represent a
defined state of health (exemplary evaluating step 130). The
reference points can be determined by glucose tolerance tests on
subjects whose respective state of health is exactly known through
an independent diagnosis. Reference points that can be used
include, for example, the states of health of a fully healthy
subject (H-NG), of a patient with insulin-dependent type II
diabetes (DT2), of a normoglycemic sample patient with metabolic
syndrome (MS-NG), of a sample patient with metabolic syndrome and
impaired glucose tolerance (MS-IGT), of a sample patient with
metabolic syndrome and impaired fasting glucose levels (MS-IFG),
and of a sample patient, in whom the elevated fasting glucose is
combined with impaired glucose tolerance (MS-CGI).
[0044] In order to minimize the influence of individual variations
and particularities, a reference point may also be determined by
averaging of oral glucose tolerance tests of multiple subjects with
the same state of health.
[0045] In an instance where the state of health of the subject is
unknown, the position of the point that represents the data set of
a subject can be evaluated, for example, by calculating each
distance to the various reference points. The unknown state of
health can then be assigned to the reference point for which the
smallest distance was determined The magnitude of the distance can
be a measure of the reliability of the assignment made.
Accordingly, the distance can be used as a parameter that specifies
the status of the glucose metabolism of the patient.
[0046] To evaluate the position of the point that represents the
data set of an individual whose state of health is unknown, the
point may be projected onto a norm trajectory that reflects a
progression of disease in the vector space, from a healthy normal
patient via one of the pre-diabetic conditions to a serious
diabetic disease. A norm trajectory of this type contains at least
a fraction of the above-mentioned reference points and can be
determined from the measured data from oral glucose tolerance tests
of a considerable number of subjects whose respective state of
health is known from an independent diagnosis. Then, a point in the
vector space can be calculated for each data set of a subject.
Neglecting measuring errors and natural variation--said points are
situated on a line that connects a data point of a healthy subject
to the data point of a subject afflicted by serious diabetic
disease. Points of subjects in various stages of disease are
situated between the starting point and end-point of the
trajectory.
[0047] The trajectory, although needed for evaluating the set of
measured data from an oral glucose tolerance test of the present
disclosure, does not need to be re-calculated for each new
evaluation. Rather, it is sufficient to determine a norm trajectory
of this type just once by evaluation of a large number of measured
data of a multitude of subjects with different, and known, states
of health with pre-defined reference profiles. Therefore, if the
method according to the present disclosure is implemented using a
computer program, the trajectory can be pre-defined. A user of the
program then can operate without influencing the profile of the
trajectory or its values.
[0048] Under ideal conditions, the point representing the set of
measured data that is to be evaluated is situated on the norm
trajectory. Due to inevitable measuring inaccuracies and natural
variations, it must be expected in real-life that the point
representing the set of measured data may be situated at a shorter
or larger distance from the norm trajectory. In this case, the
point of the norm trajectory that is closest to the point
representing the set of measured data to be evaluated may be
determined in a further step of the method. Accordingly, the point
of the norm trajectory may be determined through projecting the
point representing the set of measured data onto the norm
trajectory.
[0049] Said point of the norm trajectory subdivides the norm
trajectory into two sections, namely a starting section from the
start of the norm trajectory to said point, and an end-section from
said point of the trajectory to the end of the trajectory.
[0050] The length ratios of these two sections of the trajectory
contain the information regarding the distance of the state of
health of the patient from a fully healthy status without diabetes
and from a fully manifest serious diabetic disease. Accordingly,
the length of a section of the trajectory from the start of the
trajectory to the point of the trajectory onto which the point
representing the data set was projected can therefore be used to
determine a parameter that quantifies the extent of an impairment
of glucose metabolism. This parameter can therefore specify the
stage or risk of diabetic disease.
[0051] The parameter can, in at least one embodiment, be specified
as the ratio of the length of the starting section of the
trajectory and the overall length of the trajectory. In a fully
healthy patient, the starting point of the norm trajectory is
closest to the point representing the set of measured data to be
evaluated such that the length of the starting section is equal to
or close to zero. In a patient with fully manifest diabetic
disease, though, the end of the trajectory is closest to the point
representing the set of measured data to be evaluated, such that
the length of the starting section of the trajectory divided by the
overall length of the trajectory is equal, or close, to one.
[0052] It is significant in this context that multiple norm
trajectories may exist in the considered vector space, whereby each
specifies a different disease progression from a healthy normal
patient via a pre-diabetic condition to an insulin-dependent
diabetic disease. This is because diabetic disease may develop in a
variety of ways and can be based on different causes or damage. For
example, a diabetic disease can commence with the number of
insulin-producing cells in the pancreas decreasing and the body's
inherent insulin production coming to a standstill as a result.
Alternatively, a diabetic disease may commence with muscle cells
taking up glucose in progressively worse manner (insulin
resistance) such that increased amounts of insulin or treatment
with insulin sensitizers is required. Both the therapeutic options
and the stages of disease on the way to the final stage of a
diabetic disease are different in these two cases. Therefore, a
separate trajectory can be considered in the vector space for each
of these cases. Whereas the starting points and end-points of these
trajectories coincide, their interim course differs.
[0053] If multiple norm trajectories are defined in the vector
space, the point characterizing the data set is projected onto the
norm trajectory situated at the shortest distance from it. In
addition, the point characterizing the data set can also be
projected onto a second norm trajectory situated at the second
shortest distance from it. By this means, the second norm
trajectory can be used as the basis for determination of a second
parameter describing the status of glucose metabolism based on the
length of a section of the trajectory from the start of the
trajectory to the point of the trajectory onto which the point
representing the data set was projected. Valuable information in
this context can be obtained from the distance of the point
representing the data set from the first norm trajectory and the
distance from the second norm trajectory, e. g. from the
relationship of the two distances. This is the case, because these
distances indicate the reliability of the assignment to a norm
trajectory that was made and thus the reliability of further
disease progression to be expected. Cases, in which an unequivocal
assignment to a norm trajectory cannot be made, may be a sign of
multiple parallel damage mechanisms acting on the glucose
metabolism, in particular in a pre-diabetic subject.
[0054] All coordinate axes of the vector space considered according
to the invention can be provided by similarity measures. In this
case, all coordinates of a point representing a data set from an
oral glucose tolerance test can be specified as values of
similarity measures. However, the vector space considered can just
as well comprise one or more additional dimensions, i.e. have
further coordinate axes for which coordinate values are calculated
independent of the similarity measures. One further coordinate axis
of the vector space can be formed, for example, by a norm of a
vector formed from the series of measured values of the glucose
concentration or analyte concentration. In this case, the point
representing the data set in the vector space comprises the value
of the corresponding norm as a further coordinate. The norm may be
calculated in customary manner as Euclidian vector norm, i.e. as
the root of the sum of squares of the individual vector
components.
[0055] Just as well, biometric or genetic variables can be added as
dimensions of the vector space. Examples of biometric variables
include body mass index, fraction of body fat, and waist-to-hip
ratio. Parameters such as blood pressure or heart rate can also be
used as coordinates. Concentrations of metabolites showing no or
little change on the time scale of an oral glucose tolerance test
may also be used, such as cholesterol, cholesterol fractions (LDL
and HDL), HbA1c or renal function parameters.
[0056] A method according to the present disclosure can be
implemented efficiently through the use of a data processing
device, i.e. a computer. Embodiments of the present disclosure
therefore also relate to a computer program for implementing a
method according to the present disclosure. A computer program
product according to the present disclosure can be loaded into the
memory of a computer and executes the steps of an embodiment of the
method according to the present disclosure when the computer
program product is run on a computer. The present disclosure also
relates to a machine-readable storage medium on which a computer
program product of this type is stored, i.e. a program that
implements a method according to the invention when it is being run
on a computer.
[0057] FIG. 2 shows, for different patient groups, the mean profile
of the glucose concentration (g) in units of mg per d1 plotted
versus time (t) during a glucose tolerance test for a period of 120
min. The individual patient groups are fully healthy people (H),
normoglycaemic patients with metabolic syndrome (MS-NG), patients
with metabolic syndrome and impaired glucose tolerance (MS-IGT),
and patients with metabolic syndrome and impaired fasting glucose
levels (MS-IFG). Moreover, the use of the method according to the
present disclosure for other risk groups--risk of type II
diabetes--, e.g. adipose people or individuals with a genetic
predisposition, is also contemplated.
[0058] FIG. 3 shows the average profile of the C-peptide
concentration during the oral glucose tolerance test for the
patient groups of FIG. 2. In this context, rather than the absolute
C-peptide concentration being plotted in FIG. 3, the change
.DELTA.C of said concentration with respect to a baseline value is
plotted. In like manner, FIG. 4 shows the change of the proinsulin
concentration .DELTA.P with respect to a baseline value during the
glucose tolerance test.
[0059] In the example shown in FIGS. 2 to 4, analyte concentrations
were measured in each case at the start of the glucose tolerance
test (t=0), after 15 min, after 30 min, after 60 min, and after 120
min Measured values of the glucose concentration are denoted
g.sub.n hereinafter, whereby n indicates the time in minutes at
which the concentration value was measured. In like manner,
measured values of the C-peptide concentration, or to be more
precise of the change with respect to a baseline value, are denoted
c.sub.n and measured values of the proinsulin concentration, or to
be more precise of the change with respect to a baseline value, are
denoted p.sub.n.
[0060] The measured values of an analyte concentration can be used
to form a vector, in that each concentration value measured is used
as a vector component. Accordingly, a vector g=(g.sub.0, g.sub.15,
g.sub.30, g.sub.60, g.sub.120) can be formed from the measured
values of the glucose concentration in the present case. In like
manner, for example, the measured values of the C-peptide
concentration and of the proinsulin concentration can be used to
form vectors c and p, respectively, which represent the respective
series of measured values.
[0061] For the further evaluation, it is useful to normalize the
vectors thus formed. A suitable norm is, in particular, the common
quadratic definition, i.e. the Euclidian vector norm,
.parallel.g.parallel.=(g.cndot.g).sup.1/2
[0062] Reference profiles of the glucose concentration and of the
further analyte concentrations are then defined for the further
evaluation of the series of measured values and/or the vectors
formed from them. Then vectors are formed for each of these sample
profiles, which can represent a section of the expected time
profile of the analyte concentration during an oral glucose
tolerance test for a given state of health.
[0063] In the present illustrative embodiment, the following
vectors are used as vectors for reference profiles, whereby M is
the dimension of the vector space in each case:
[0064] The vector of the normalized body diagonal N.sub.m=(1/ M, 1/
M, . . . 1/ M). Accordingly, if the vector includes five
components, as is the case in a profile according to FIGS. 2 to 4,
the resulting vector is N=(1, 1, 1, 1, 1)/ 5.
[0065] A vector L.sub.M=(1, 2, 3, . . . ,
M)/.parallel.L.sub.M.parallel., i.e. a linearly increasing sample
profile. For the example of FIGS. 2 to 4, L=(1, 2, 3, 4, 5)/
55.
[0066] A triangular (uneven M) or trapezoidal (even M) sample
profile with the vector D.sub.M=(1, 2, . . . , M/2, M/2-1, . . . ,
1)/.parallel.D.sub.M.parallel., whereby M/2 is to be rounded up if
M is uneven. For the case of FIGS. 2 to 4, D=(1, 2, 3, 2, 1)/
19.
[0067] A similarity measure is calculated in a further step of the
method and specifies the similarity of the concentration profile
and the corresponding reference profile. The similarity measure can
be calculated, for example, by calculating the scalar product of
the respective vectors. In this context, the scalar product can, in
turn, also be used as similarity measure or the angle formed by the
two vectors can be calculated from the scalar product.
[0068] For example the following angles can be used as similarity
measure of the glucose concentration profile:
.alpha..sub.g=arc cos(N.cndot.g), .beta..sub.g=arc cos(L.cndot.g),
.gamma..sub.g=arc cos(D.cndot.g).
[0069] Accordingly, the following angles can be used as similarity
measure for the profile of the C-peptide concentration and/or
proinsulin concentration:
.alpha..sub.c=arc cos(N.cndot.c), .beta..sub.c=arc cos(L.cndot.c),
.gamma..sub.c=arc cos(D.cndot.c) and/or
.alpha..sub.p=arc cos(N.cndot.p), .beta..sub.p=arc cos(L.cndot.p),
.gamma..sub.p=arc cos(D.cndot.p).
[0070] Each data set from an oral glucose tolerance test can be
described by similarity measures of this type. A norm of the
individual vectors representing the concentration profiles can be
used just as well to supplement the characterisation of a set of
measured data from a glucose tolerance test: for example, the
Euclidian norm of the vectors, g, c, p. One thus obtains for a set
of measured data a set of variables, for example the variables,
.alpha..sub.g, .beta..sub.g, .gamma..sub.g, .alpha..sub.c,
.beta..sub.c, .gamma..sub.c, .alpha..sub.p, .beta..sub.p,
.gamma..sub.p, .parallel.g.parallel., .parallel.c.parallel.,
.parallel.g.parallel..
[0071] It is to be noted in this context that a larger or smaller
number of reference profiles and, in particular, that other
reference profiles can be used just as well. In particular,
different reference profiles can be used for each individual
analyte concentration. A set of measured data can therefore also be
characterised through a different set of variables with a larger or
smaller number of variables.
[0072] The values of the individual variables of the set of
variables calculated from the measured data, for example the values
of the variables .alpha..sub.g, .beta..sub.g, .gamma..sub.g,
.alpha..sub.c, .beta..sub.c, .gamma..sub.c, .alpha..sub.p,
.beta..sub.p, .gamma..sub.p, .parallel.g.parallel.,
.parallel.c.parallel., .parallel.g.parallel., can be used as
coordinates in a vector space. By this means, each set of measured
data from a glucose tolerance test can be represented by one point
in a vector space. The coordinate axes of the vector space are then
given by one of the variables each, whereby the value of said
variable specifies the respective coordinate.
[0073] FIGS. 5 and 6 show a schematic view in different viewing
angles of a simplified example of a vector space of this type, in
which points are marked, which each represent a set of measured
data from an oral glucose tolerance test. Since only three
dimensions can be shown graphically, the variables,
.parallel.g.parallel., .alpha..sub.c, and .gamma..sub.c, were
selected from the above-mentioned variables for FIGS. 5 and 6 for
purposes of illustration. However, in a practical implementation of
the method, a vector space of a higher dimension is used, i.e. a
larger number of variables.
[0074] It is evident even from the simplified example of FIGS. 5
and 6 that the data points of various patient groups are clearly
separated from each other in this vector space. Data points of
healthy humans (H) are indicated by +, data points of patients with
metabolic syndrome and impaired glucose tolerance (MS-IGT) are
indicated by x, and data points of patients with type II diabetes
(DT2) are indicated by .diamond.. The plot of FIG. 6 shows that the
data points are situated approximately in the same plane and thus
form a line in FIG. 6. Data points of patients with type II
diabetes (DT2) are situated in the left upper part of said line.
Data points of healthy samples (H) are situated in the right lower
part. Accordingly, proceeding from right to left along the line,
there are data points of type H first, then there are increasingly
more data points of the MS-IGT-type, and lastly there are data
points of the DT2-type. Accordingly, even in a simplified vector
space, a line can be recognised that indicates how the data
obtained from an oral glucose tolerance test change upon disease
progression.
[0075] The use of values of the individual variables of the set of
variables calculated from the measured data, i.e., for example, the
values of the variables, .alpha..sub.g, .beta..sub.g,
.gamma..sub.g, .alpha..sub.c, .beta..sub.c, .gamma..sub.c,
.alpha..sub.p, .beta..sub.p, .gamma..sub.p, .parallel.g.parallel.,
.parallel.c.parallel., .parallel.g.parallel., as coordinates in a
vector space therefore defines a vector space, in which there is a
trajectory that specifies a typical disease progression with
increasing impairment of glucose metabolism. The trajectory
therefore starts at a point that is expected for a fully healthy
status and progresses via points representing pre-diabetic
conditions or early diabetic disease, up to a point whose
coordinates occur as values of variables in a patient with an
insulin-dependent diabetic disease.
[0076] Said trajectory can be referred to as norm trajectory, since
it describes the normal progression of disease. A norm trajectory
can be obtained by evaluating the data of a considerable number of
oral glucose tolerance tests of subjects whose state of health is
known.
[0077] A point in said vector space is obtained from each set of
measurement data by evaluating data from oral glucose tolerance
tests for each stage of a diabetic disease. These points should
coincide if the states of health are identical. However, it cannot
be presumed that the state of health of two patients is exactly
identical. Hence, some scattering of the points is to be expected.
Despite this scattering, a norm trajectory can be calculated from a
sufficiently large number of data points, for example by
calculating the mean. Preferably, the mean of the values of a
variable is calculated for data sets measured on patients with
identical state of health, i.e. the center of gravity of a point
cloud. However, it is feasible just as well to calculate a mean at
an earlier stage of data analysis, for example a mean of the
individual concentration profiles can be calculated in order to
determine a typical profile of the concentration of glucose or
other analyte for the respective state of health in a glucose
tolerance test.
[0078] FIG. 7 shows in exemplary manner a norm trajectory with data
points from oral glucose tolerance tests of various patient groups,
namely healthy patients H (.cndot.), patients with metabolic
syndrome and impaired glucose tolerance (MS-IGT) (x) and
insulin-dependent type II diabetics DT2 (.diamond.).
[0079] In order to determine the state of health of the patient
from a set of measurement data of an oral glucose tolerance test,
it needs to be determined which point of the norm trajectory has
the smallest distance from the point representing the set of
measurement data from the glucose tolerance test. Thus, the point
representing the measurement data from a glucose tolerance test is
being projected onto the norm trajectory. The point of the norm
trajectory thus determined subdivides the norm trajectory into two
sections, namely a starting section and an end-section. The length
of the section of the trajectory from the start of the trajectory
to the point of the trajectory onto which the point representing
the data set was projected is then used to determine a parameter
that quantifies the extent of an impairment of glucose metabolism
and thus indicates the disease stage of the patient. The parameter
can, for example, be the ratio of the length of the starting
section to the overall length of the trajectory.
[0080] The method also enables, in particular, a differentiation to
be made within a given state of disease and/or health. Certain
phases of disease can be seen with type II diabetes: treatment
involving diet, treatment involving diet plus oral medication (e.g.
metformin), treatment involving oral medication plus supplemental
insulin, as well as fully insulin-dependent type II diabetes. This
progression can be described in a vector space by a norm
trajectory. The data point of a diabetic that has been tested can
be assigned to a point of the norm trajectory and the disease stage
can be recognized from its position on the norm trajectory.
Further, embodiments of the method of the present disclosure allow
a user to recognize when the time has come to switch from one
treatment method to the next.
[0081] Systems and methods of the present disclosure for
automatically displaying patterns in biological data may include
one or more processors, and machine readable instructions. The
machine readable instructions can cause the one or more processors
to divide biological data into segments of interest. The one or
more processers can transform, automatically, each of the segments
of interest into a set of features according to a mathematical
algorithm. Further, the one or more processers can cluster,
automatically, the segments of interest into groups of clustered
segments according to a clustering algorithm. The segments of
interest can be grouped in the groups of clustered segments based
at least in part upon the set of features. A cluster center can be
associated with one of the groups of clustered segments. Moreover,
the one or more processers can present, automatically, the cluster
center on a human machine interface.
[0082] Further, in describing representative embodiments, the
disclosure may have presented a method and/or process as a
particular sequence of steps. However, to the extent that the
method or process does not rely on the particular order of steps
set forth herein, the method or process should not be limited to
the particular sequence of steps described. Other sequences of
steps may be possible. Therefore, the particular order of the steps
disclosed herein should not be construed as limitations of the
present disclosure. In addition, disclosure directed to a method
and/or process should not be limited to the performance of their
steps in the order written. Such sequences may be varied and still
remain within the scope of the present disclosure.
[0083] Having described the present disclosure in detail and by
reference to specific embodiments thereof, it will be apparent that
modifications and variations are possible without departing from
the scope of the disclosure defined in the appended claims. More
specifically, although some aspects of the present disclosure are
identified herein as preferred or particularly advantageous, it is
contemplated that the present disclosure is not necessarily limited
to these preferred aspects of the disclosure.
* * * * *