U.S. patent application number 14/744361 was filed with the patent office on 2015-10-08 for apparatus for health correlation assessment.
This patent application is currently assigned to Welch Allyn, Inc.. The applicant listed for this patent is RICHARD W. NEWMAN, EDWARD O'NEIL, JIEJING QIU, MIN XU. Invention is credited to RICHARD W. NEWMAN, EDWARD O'NEIL, JIEJING QIU, MIN XU.
Application Number | 20150286794 14/744361 |
Document ID | / |
Family ID | 41215652 |
Filed Date | 2015-10-08 |
United States Patent
Application |
20150286794 |
Kind Code |
A1 |
NEWMAN; RICHARD W. ; et
al. |
October 8, 2015 |
APPARATUS FOR HEALTH CORRELATION ASSESSMENT
Abstract
An apparatus, system and method for performing a health
correlation assessment based upon results an eye contrast
sensitivity measurement test of a person. The health correlation
assessment yields a predictive and probabilistic measure of the
likelihood of the person appropriately assigned to a particular
clinical health classification. Such an eye contrast sensitivity
measurement can be used to identify people who are likely to be
within a pre-Alzheimer's disease state or within Alzheimer's
disease state of health and identify those who should seek further
clinical and/or other types of testing to ascertain their actual
health status and to determine an appropriate course of medical
treatment.
Inventors: |
NEWMAN; RICHARD W.; (AUBURN,
NY) ; XU; MIN; (CORTLAND, NY) ; QIU;
JIEJING; (SYRACUSE, NY) ; O'NEIL; EDWARD;
(WEEDSPORT, NY) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
NEWMAN; RICHARD W.
XU; MIN
QIU; JIEJING
O'NEIL; EDWARD |
AUBURN
CORTLAND
SYRACUSE
WEEDSPORT |
NY
NY
NY
NY |
US
US
US
US |
|
|
Assignee: |
Welch Allyn, Inc.
Skaneateles Falls
NY
|
Family ID: |
41215652 |
Appl. No.: |
14/744361 |
Filed: |
June 19, 2015 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
13620486 |
Sep 14, 2012 |
9060728 |
|
|
14744361 |
|
|
|
|
12430202 |
Apr 27, 2009 |
|
|
|
13620486 |
|
|
|
|
12794053 |
Jun 4, 2010 |
8439501 |
|
|
13620486 |
|
|
|
|
12793989 |
Jun 4, 2010 |
8702234 |
|
|
12794053 |
|
|
|
|
12505193 |
Jul 17, 2009 |
8075136 |
|
|
12794053 |
|
|
|
|
12505193 |
Jul 17, 2009 |
8075136 |
|
|
12793989 |
|
|
|
|
11224774 |
Sep 13, 2005 |
7575321 |
|
|
12505193 |
|
|
|
|
10697454 |
Oct 30, 2003 |
7708403 |
|
|
11224774 |
|
|
|
|
61700782 |
Sep 13, 2012 |
|
|
|
61047935 |
Apr 25, 2008 |
|
|
|
Current U.S.
Class: |
705/3 |
Current CPC
Class: |
A61B 5/163 20170801;
A61B 3/022 20130101; A61B 3/113 20130101; A61B 5/14532 20130101;
Y02A 90/26 20180101; A61B 3/1241 20130101; G16H 50/30 20180101;
A61B 5/4088 20130101; A61B 5/4082 20130101; Y02A 90/10 20180101;
A61B 5/4064 20130101 |
International
Class: |
G06F 19/00 20060101
G06F019/00; A61B 5/00 20060101 A61B005/00; A61B 3/02 20060101
A61B003/02 |
Claims
1. An apparatus for performing a probabilistic assessment for one
or more Alzheimer's disease related clinical health classifications
based upon results of a contrast sensitivity eye examination test,
comprising: an eye contrast sensitivity testing device that is
configured to perform an eye contrast sensitivity test with respect
to a first subject, and that is further configured to generate eye
contrast sensitivity test result data, representing a result of
said eye contrast sensitivity test with respect to said first
subject; a data access component for obtaining at least a portion
of eye contrast sensitivity test result data from a first eye
contrast sensitivity test performed by a first subject, from said
eye contrast sensitivity testing device; a health correlation
assessment component that exercises a first algorithm in accordance
with a correlation model, said component outputs at least one
numerical probability value representing an estimated probability
of said first subject being clinically assigned into at least one
of a plurality of Alzheimer's disease related probabilistic health
classifications, said probabilistic health classifications
including at least one classification indicating a presence of an
Alzheimer's disease related state of health; and an output
component that outputs information representing said probabilistic
health classifications.
2. The apparatus of claim 1 wherein said first algorithm employs
mathematical coefficients each having values that are periodically
re-determined from re-processing of a repository of sample eye
contrast sensitivity test data, and wherein said repository of
sample eye contrast sensitivity test data evolves via periodic
addition of said other eye contrast sensitivity test result data
over time.
3. The apparatus of claim 1 wherein said first algorithm further
processes personal attribute data associated with said first
subject, for determining at least one said plurality of Alzheimer's
disease related probabilistic health classifications for said first
subject.
4. The apparatus of claim 1 wherein said probabilistic health
classifications include one probability of a subject being
appropriately classified as having a pre-clinical or clinical stage
of Alzheimer's disease.
5. The apparatus of claim 1, wherein said correlation model is
based, at least in part, upon an amount of time required for said
test subject to complete said eye contrast sensitivity test.
6. The apparatus of claim 1 wherein said probabilistic health
classifications include one probability of a subject being
appropriately classified as having an absence of an Alzheimer's
related disease.
7. The apparatus of claim 1 wherein said at least a portion of a
contrast sensitivity test result includes at least one contrast
sensitivity test parameters of RightDurationTime, LeftDurationTime,
MDLeftEye, MDRightEye, SuperiorLeftEye and SuperiorRight Eye.
8. A system for performing a probabilistic assessment for one or
more Alzheimer's disease related clinical health classifications
based upon results of a contrast sensitivity eye examination test,
comprising: an eye contrast sensitivity testing device that is
configured to perform an eye contrast sensitivity test with respect
to a first subject, and that is further configured to generate eye
contrast sensitivity test result data, representing a result of
said eye contrast sensitivity test with respect to said first
subject; a data access component for obtaining at least a portion
of eye contrast sensitivity test result data from a first eye
contrast sensitivity test performed by a first subject, from said
eye contrast sensitivity testing device; a health correlation
assessment component that exercises a first algorithm in accordance
with a correlation model, said component outputs at least one
numerical probability value representing an estimated probability
of said first subject being clinically assigned into at least one
of a plurality of Alzheimer's disease related probabilistic health
classifications, said probabilistic health classifications
including at least one classification indicating a presence of an
Alzheimer's disease related state of health; and an output
component that outputs information representing said probabilistic
health classifications.
9. The system of claim 8 wherein said first algorithm employs
mathematical coefficients each having values that are periodically
re-determined from re-processing of a repository of sample eye
contrast sensitivity test data, and wherein said repository of
sample eye contrast sensitivity test data evolves via periodic
addition of said other eye contrast sensitivity test result data
over time.
10. The system of claim 8 wherein said first algorithm further
processes personal attribute data associated with said first
subject, for determining at least one said plurality of Alzheimer's
disease related probabilistic health classifications for said first
subject.
11. The system of claim 8 wherein said probabilistic health
classifications include one probability of a subject being
appropriately classified as having a pre-clinical or clinical stage
of Alzheimer's disease.
12. The system of claim 8, wherein said correlation model is based,
at least in part, upon an amount of time required for said test
subject to complete said eye contrast sensitivity test.
13. The system of claim 8 wherein said probabilistic health
classifications include one probability of a subject being
appropriately classified as having an absence of an Alzheimer's
related disease.
14. The system of claim 8 wherein said at least a portion of a
contrast sensitivity test result includes at least one contrast
sensitivity test parameters of RightDurationTime, LeftDurationTime,
MDLeftEye, MDRightEye, SuperiorLeftEye and SuperiorRight Eye.
15. A method for performing a probabilistic assessment for one or
more Alzheimer's disease related clinical health classifications
based upon results of a contrast sensitivity eye examination of a
first subject, comprising the steps of: providing an eye contrast
sensitivity testing device that is configured to perform an eye
contrast sensitivity test with respect to a first subject, and that
is further configured to generate eye contrast sensitivity test
result data, representing a result of said eye contrast sensitivity
test with respect to said first subject; providing a data access
component for obtaining at least a portion of eye contrast
sensitivity test result data from a first eye contrast sensitivity
test performed by a first subject, from said eye contrast
sensitivity testing device; providing a health correlation
assessment component that exercises a first algorithm in accordance
with a correlation model, said component outputs at least one
numerical probability value representing an estimated probability
of said first subject being clinically assigned into at least one
of a plurality of Alzheimer's disease related probabilistic health
classifications, said probabilistic health classifications
including at least one classification indicating a presence of an
Alzheimer's disease related state of health; and providing an
output component that outputs information representing said
probabilistic health classifications.
16. The method of claim 15 wherein said first algorithm employs
mathematical coefficients each having values that are periodically
re-determined from re-processing of a repository of sample eye
contrast sensitivity test data, and wherein said repository of
sample eye contrast sensitivity test data evolves via periodic
addition of said other eye contrast sensitivity test result data
over time.
17. The method of claim 15 wherein said first algorithm further
processes personal attribute data associated with said first
subject, for determining at least one said plurality of Alzheimer's
disease related probabilistic health classifications for said first
subject.
18. The method of claim 15 wherein said probabilistic health
classifications include one probability of a subject being
appropriately classified as having a pre-clinical or clinical stage
of Alzheimer's disease or the absence of Alzheimer's disease.
19. The method of claim 15, wherein said correlation model is
based, at least in part, upon an amount of time required for said
test subject to complete said eye contrast sensitivity test.
20. The method of claim 15 wherein said at least a portion of a
contrast sensitivity test result includes at least one contrast
sensitivity test parameters of RightDurationTime, LeftDurationTime,
MDLeftEye, MDRightEye, SuperiorLeftEye and SuperiorRight Eye.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] This document (Docket No: WAL-002C1) is a continuation of,
and claims priority and benefit to, co-pending (U.S.)
non-provisional utility patent application having Ser. No.
13/620,486 (Docket No. WAL-002CV) (Confirmation No. 4631), which
was filed on Sep. 14, 2012, and that is entitled "Apparatus for
Health Correlation Assessment", and which is incorporated herein by
reference in its entirety.
[0002] The U.S. non-provisional utility patent application having
Ser. No. 13/620,486 (Docket No. WAL-002CV) (Confirmation No. 4631)
is a non-provisional of, and claims priority and benefit to, U.S.
provisional utility patent application Ser. No. (61/700,782)
(Confirmation No. 7817) filed on Sep. 13, 2012, that is entitled
"Apparatus for Health Correlation Assessment", and which is also
incorporated herein by reference in its entirety.
[0003] The U.S. non-provisional utility patent application having
Ser. No. 13/620,486 (Docket No. WAL-002CV) (Confirmation No. 4631),
is also a continuation-in-part of, and claims benefit and priority
to, U.S. non-provisional utility patent application Ser. No.
12/430,202 filed on Apr. 27, 2009, that is entitled "Apparatus and
Method for Diagnosis of Optically Identifiable Ophthalmic
Conditions", which claims priority and benefit to U.S. provisional
utility patent application Ser. No. 61/047,935 that was filed on
Apr. 25, 2008. Priority is claimed to all of the aforementioned
patent applications, which are incorporated herein by reference in
their entirety.
[0004] The U.S. non-provisional utility patent application having
Ser. No. 13/620,486 (Docket No. WAL-002CV) (Confirmation No. 4631)
is also a continuation-in-part patent application of, and claims
priority and benefit to both, U.S. non-provisional patent
application Ser. No. 12/794,053, which has issued as U.S. Pat. No.
8,439,501, and Ser. No. 12/793,989, which has issued as U.S. Pat.
No. 8,702,234, and that were both filed on Jun. 4, 2010, and that
are entitled "Diagnosis of Optically Identifiable Ophthalmic
Conditions".
[0005] Both of the aforementioned U.S. non-provisional patent
applications, Ser. No. 12/794,053 and Ser. No. 12/793,989, are
continuation patent applications of, and claim priority and benefit
to, U.S. non-provisional utility patent application Ser. No.
12/505,193, that was filed on Jul. 17, 2009, has issued as U.S.
Pat. No. 8,705,136, and that is entitled "Apparatus and Method for
Diagnosis of Optically Identifiable Ophthalmic Conditions", and
which is a divisional continuation patent application of, and
claims priority and benefit to, U.S. patent application Ser. No.
11/224,774 that was filed on Sep. 13, 2005, and which has issued
U.S. Pat. No. 7,575,321, and which is a continuation-in-part
application of, and claims the priority and benefit to, U.S. patent
application Ser. No. 10/697,454 that was filed Oct. 30, 2003, and
that has issued as U.S. Pat. No. 7,708,403. Priority is claimed to
all of the aforementioned patent applications, which are each
incorporated herein by reference in their entirety.
JOINT RESEARCH AGREEMENT
[0006] The subject matter of this invention includes the processing
of data obtained from a joint research agreement between the
Assignee of the present application and Andrew Saykin of Indiana
University, Center for Neuroimaging, Dept. of Radiology. Indiana
University School of Medicine, 959 W. Walnut St. R2EH24,
Indianapolis. Ind. and Laura Flashman, Dartmouth College, Dept. of
Psychiatry, Dartmouth College Medical School, Lebanon, N.H.
BACKGROUND OF THE INVENTION
[0007] People residing within certain health classifications, such
as those people afflicted with Alzheimer's disease, require a
substantial amount of costly medical and non-medical care during
the period of time in which they are afflicted with such a
debilitating disease. The amount of required care typically
escalates as the disease progresses. Also, as the disease
progresses, it becomes more likely that the disease will
facilitate, but not necessarily cause, pre-mature 1 death of the
afflicted person.
[0008] Some health care professionals believe that early detection
and treatment of Alzheimer's disease could eventually slow
progression of the disease, reduce the amount and cost of long term
care required to be provided to such a person, and in some
circumstances, enable a person to delay and/or even avoid the most
severe stages of the disease prior to the eventual death of that
person. Technology that can provide early detection of Alzheimer's
disease at a cost that is substantially less than the cost of
existing methods of detection is needed. Such technology could
reduce the massive projected costs of caring for future Alzheimer's
patients.
[0009] The discussion above is merely provided for general
background information and is not intended to be used as an aid in
determining the scope of the claimed subject matter.
BRIEF DESCRIPTION OF THE INVENTION
[0010] The invention relates to an apparatus, system and method for
performing a health correlation assessment based upon results an
eye contrast sensitivity measurement test of a person. The health
correlation assessment provides information conveying a relative
probability (likelihood) of the person being appropriately assigned
into a particular clinical health classification based upon some
standard of clinical testing. In some embodiments, the health
correlation assessment is quantitative and provides a set of one or
more numerical probabilities as a predictive estimate of the
likelihood of the person being appropriately assigned into a
particular clinical health classification.
[0011] Such an eye contrast sensitivity measurement can be used to
identify (predict) those people who are more likely to be
appropriately classified within a pre-clinical Alzheimer's disease
state (e.g. cognitive complaint and mild cognitive complaint) or
within Alzheimer's disease state of health (e.g. mild, moderate,
and severe), and who should seek further clinical testing to
ascertain their actual clinical health status, and to determine a
course of treatment, and to track a rate of progression of the
disease.
[0012] This brief description of the invention is intended only to
provide a brief overview of subject matter disclosed herein
according to one or more illustrative embodiments, and does not
serve as a guide to interpreting the claims or to define or limit
the scope of the invention, which is defined only by the appended
claims.
BRIEF DESCRIPTION OF THE DRAWINGS
[0013] So that the manner in which the features of the invention
can be understood, a detailed description of the invention may be
had by reference to certain embodiments, some of which are
illustrated in the accompanying drawings. It is to be noted,
however, that the drawings illustrate only certain embodiments of
this invention and are therefore not to be considered limiting of
its scope, for the scope of the invention can encompass other
equally effective embodiments. The drawings are not necessarily to
scale. The emphasis of the drawings is generally being placed upon
illustrating the features of certain embodiments of the invention.
In the drawings, like numerals are used to indicate like parts
throughout the various vicws. Differences between like parts may
cause those parts to be indicated with different numerals. Unlike
parts are indicated with different numerals. Thus, for further
understanding of the invention, reference can be made to the
following detailed description, read in connection with the
drawings in which:
[0014] FIG. 1 illustrates communication between an eye contrast
sensitivity test (CST) measurement device and a health correlation
assessment service.
[0015] FIG. 2 illustrates a map of retinal zones employed within an
eye contrast sensitivity test measurement (FDT) device.
[0016] FIGS. 3A-3B list an embodiment of a set of contrast
sensitivity test (CST) output file parameters of the FDT device of
FIG. 1.
[0017] FIGS. 4A-4B illustrate the processing of contrast
sensitivity measurement test (CST) result data by the health
correlation assessment procedure (HCAP) implemented as computer
software.
[0018] FIGS. 5A-5B illustrate a health correlation assessment
procedure algorithm (HCAPPA) coefficient generation process in
which multiple types of data are processed.
[0019] FIGS. 6A-6B illustrate a single CST parameter correlation
model associated with a first embodiment a health correlation
assessment program predictive algorithm (HCAPPA).
[0020] FIGS. 7A-7C illustrate sample data representing a
RightDurationTime and PeripheralRight contrast sensitivity test
(CST) parameter measurements and an age personal attribute with
respect to sample subjects assigned within (4) actual clinical
health classifications.
[0021] FIGS. 8A-8C illustrate graphs of a correlation model
probability of a subject's being appropriately classified as having
one of four health classifications as a function of a
RightDurationTime parameter value obtained from eye contrast
sensitivity test result.
[0022] FIG. 9 is a matrix illustrating a relationship between the
results of the predicted and actual clinical health classifications
based upon the single CST parameter correlation model of FIG.
6A.
[0023] FIG. 10 illustrates a multiple CST parameter correlation
model associated with a second embodiment of the correlation
assessment procedure predictive algorithm (HCAPPA).
[0024] FIG. 11 is a matrix illustrating a relationship between the
results of the predicted and actual clinical health classifications
based upon the multiple CST parameter correlation model of FIG.
10.
[0025] FIG. 12 illustrates systematic interoperation between
contrast sensitivity testing, clinical testing and the health
correlation assessment service.
DETAILED DESCRIPTION OF THE INVENTION
[0026] FIG. 1 illustrates communication between an eye contrast
sensitivity test (CST) measurement device 110 and a health
correlation assessment service (HCAS) 130. The HCAS 130
incorporates computing resources having access to algorithmic
procedures based upon clinical and CST data that support
performance of health correlation assessment. The eye contrast
sensitivity measurement device 110, also referred to as a frequency
doubling technology (FDT) device 110, is designed to project images
onto the eyes of a subject, also referred to herein as the viewer
140. The images are projected during intermittent time intervals
and at various locations within the field of view of the viewer,
and contain contrast patterns that are dynamically altered over
time. See U.S. Pat. No. 8,073,136 which is incorporated herein by
reference. The viewer 140 provides a response to the FDT device 110
upon detecting the presence of a contrast image during a period of
time.
[0027] The results of each contrast sensitivity test (CST) eye
examination are stored into an output file within non-volatile
memory storage, such as a disk storage device, residing within the
FDT device 110. The content of the file can be stored into various
file formats, including such as .FDT, .PDF or .XML file
formats.
[0028] In accordance with the invention, the output file can be
further communicated (copied) to a location that is outside of the
FDT device 110. Such communication can be performed via a
communication path 124a to a network 120, such as the internet,
and/or via a communication path 124b to web accessible sites
located on the internet including a health correlation assessment
service (HCAS) site 130. The FDT device 110 can also have a
communication path 122a to another computer 150 via a direct
connection or via a network, including such as the internet. The
computer 150 can act as an intermediary for access to the HCAS 130
via communication links 122b-122c to the internet. The FDT device
110 can also communicate the output file via a portable memory
device (media), such as via a compact disc or a universal serial
bus (USB) compatible memory device to another computer 150.
[0029] Alternately, in some embodiments, the HCAS 130 and the
computer 150 can be incorporated within the FDT device 110, having
access to the output file that is stored local to the FDT device
110. In other embodiments, the HCAS 130 executes on the computer
150 which has access to a copy of the output file
[0030] FIG. 2 illustrates a map of retinal zones employed within an
eye contrast sensitivity measurement (FDT) device 110. As shown, in
this embodiment, there are (55) zones that occupy the field of view
of each eye of the viewer and that are arranged and numbered from a
viewing perspective of the FDT device 110 towards the subject
(viewer) being tested. Note that other embodiments of contrast
sensitivity measurement may employ different arrangements to map
the field of view of the viewer.
[0031] During a contrast sensitivity examination (test), the FDT
device 110 projects contrast patterns at locations within the
selected (55) zones of the viewer's field of view. The viewer
responds by pressing a button (160) in communication with the FDT
device 110 when a contrast pattern is recognized.
[0032] Upon completion of a contrast sensitivity test, the FDT
device 110 outputs test results into an output file. The output
file includes a set of information quantifying the test performance
of the viewer. The test results include a plurality of individual
parameters. Some of these parameters measure test performance in
association with a particular zone, while other parameters measure
performance in association with a set of more than one zone, or in
association with the contrast sensitivity test as a whole. In some
embodiments, the output file may also include other information,
including personal attributes including such as the name, gender
and date of birth of the viewer.
[0033] FIGS. 3A-3B list an embodiment of a set of contrast
sensitivity test (CST) output file parameters of the FDT device of
FIG. 1. Of the (31) listed CST output file parameters. the last
(24) CST parameters are explicitly defined to directly incorporate
contrast sensitivity threshold values as measured for specific
retinal zones. For example, the value of the "Superior10LeftEye"
CST parameter is defined as an average of measured contrast
sensitivity thresholds for (8) retinal zones 2-3, 7-8, 14-15 and
22-23.
[0034] Conversely, the first listed parameter "Age", is a personal
attribute that is provided by the subject as part of performance of
the contrast sensitivity test procedure and is not itself, a
contrast sensitivity measurement. The "LeftDurationTime" and
"RightDurationTime" values measure time required to complete the
contrast sensitivity test for each of the left and right eyes. The
"MDLcftEye" and "MDRightEye" are mathematical representations (mean
deviation) for contrast sensitivity measurements of each of the
left and right eyes. The "PSDLeftEye" and "PSDRightEye" are
mathematical representations (pattern standard deviation) for
contrast sensitivity measurements of each of the left and right
eyes.
[0035] Embodiments of health correlation assessment procedures
(HCAP) are designed to input and process some or all of the (31)
listed CST output file parameter values and to output a health
correlation assessment of the subject for which the CST parameters
were measured. Other CST parameters can be created and defined
based upon information obtained from the results of the contrast
sensitivity test (Sec FIG. 10)
[0036] FIGS. 4A-4B illustrate the processing of contrast
sensitivity measurement test result data by the health correlation
assessment procedure (HCAP) implemented as computer software. FIG.
4A illustrates the processing of contrast sensitivity measurement
test result data by the health correlation assessment procedure
(HCAP) implemented as computer software 470.
[0037] As shown, results of a contrast sensitivity test 410 are
stored as data for a subject into a contrast sensitivity test (CST)
result output file 412. A health correlation assessment procedure
(HCAP) 470 exercises a health correlation assessment procedure
predictive algorithm (HCAPPA) which incorporates coefficient
values. The HCAP 470 inputs and processes the content of the output
file 412 via a data input component 472, and outputs one or more
probabilistic health classification(s) 422 via a data output
component. The HCAPPA determines a correlation based upon a
correlation model, between the information content of the CST
result output file 412 and one or more probabilistic health
classification(s) 422.
[0038] The HCAPPA coefficient values 424 (a1, b1, a2, b2, a3 and
b3) are incorporated into mathematical expressions (See equations
(1)-(7) of FIG. 6A). The equations and coefficient values
incorporated into those equations quantify a correlation between a
set of one or more particular CST parameter values and a
probabilistic health classification(s) 422, being a predictive
probability of appropriate assignment of each subject receiving the
one or more particular CST parameter values from results of a CST,
to one or more particular clinical health classifications.
[0039] In a first embodiment of FIG. 6A, these coefficients are
employed to compute a probabilistic health classification based
upon a single contrast sensitivity test (CST) measurement
parameter. In the embodiment of FIG. 10, other coefficients are
employed to compute a probabilistic health classification based
upon multiple contrast sensitivity test (CST) measurement
parameters. These HCAPPA coefficients and their values are
numerical values and are not the same and should not be confused
with contrast sensitivity test (CST) parameters, which are each
measurements of eye contrast sensitivity.
[0040] In some embodiments, the probabilistic health
classification(s) 422 include at least one numerical probability of
the subject's being appropriately classified into one or more
health classifications based upon a correlation model. See
equations (0) through (7) of FIG. 6A as an one example of an
embodiment of a correlation model that is expressed via
mathematical equations. This mathematical correlation model
correlates at least a portion of the contrast sensitivity test
result data for a subject with a probability of that subject being
appropriately assigned into one or more clinical health
classifications, in accordance with clinical criteria. In some
embodiments, this mathematical correlation model is based upon a
repository of information including results of a clinical
evaluation for each of the subjects and including contrast
sensitivity test result data for those subjects.
[0041] In some embodiments, the data input 472 component opens and
processes (parses) the information content of the CST output file
412 in accordance with various file formats, including such as
.FDT, .PDF or .XML file formats. The content of the CST output file
412 is parsed to identify each CST parameter value and personal
attribute value residing therein, for input into the HCAP
predictive algorithm (HCAPPA). For example, a CST output file 412
from the Zeiss manufactured FDT device provides CST related
measurement information for the test taking subject, for each eye
of the subject and for each retinal zone of each eye of the
subject. This output file 412 basis, includes a measured contrast
sensitivity threshold value, a total deviation value and a pattern
deviation value expressed in decibels (dB), for each of the (55)
retinal zones (See FIG. 2) of each eye of the test taking subject,
as measured from within a contrast sensitivity test (CST). This
output file 412 can further include some personal attribute
information of the test taking subject, such as test subject name,
age, and gender.
[0042] In some embodiments, the data input component 472 receives
CST parameter values and/or other personal attribute values from
other sources, such as via another data file, or via a
communications channel with other CST related software and/or via a
user interface with a user of the apparatus. For example, the HCAP
data input component 472 may input subject age data from within the
output file 412, but further input ethnicity or family history data
associated with a subject from sources outside of the CST output
file 412, such as via prompting and data entry of a user of the
computer 150 or the FDT device 110.
[0043] In some embodiments, the data output component 474 outputs
and represents probabilistic health classification(s) 422
quantitatively. For example, theses probabilistic health
classification(s) can be expressed as "Alzheimer's Disease: 55%,
"Normal Health: 45%". In some embodiments, probabilistic health
classification(s) 422 are output and expressed qualitatively. For
example, theses probabilistic health classification(s) can be
expressed as "Alzheimer's Disease appears More Probable than Normal
Health", and/or "Further clinical testing for Alzheimer's Disease
is advised".
[0044] Note that in some circumstances, non-Alzheimer's disease
related health information may be accessible to the HCAP 470 and
output via the data output component 474 in combination with
probabilistic health classification(s) 422. For example, the Zeiss
FDT device 110 output file 412 provides information regarding a
Glaucoma Hemi-field Test (GHT). The GHT is an indicator of a
presence of Glaucoma. The Zeiss FDT output file 412 indicates
whether GHT is "Within normal limits" or is not within normal
limits. Such GHT related information indicates a relative
(qualitative) probability of the test subject having Glaucoma in at
least one eye. This GHT is a health classification assessment that
is separate from any Alzheimer's Disease related probabilistic
classification that may be provided by the HCAP 470.
[0045] Note that further clinical testing for glaucoma (e.g. visual
field exam, intra-ocular pressure, and cup/disc ratio) is typically
much less expensive that for Alzheimer's Disease (e.g. a battery of
cognitive tests, MRI, and/or PET scan). In circumstances, where
health classification assessment predicts a relative probability of
the presence of both Glaucoma and Alzheimer's disease, it may be
more cost effective to perform clinical tests for Glaucoma prior to
possible tests for Alzheimer's Disease.
[0046] Like for Glaucoma, in some embodiments, the HCAP 470 will
further provide information relating to a relative probability of
the test subject being appropriately assigned to one or more than
one non-Alzheimer's disease related health classifications
including for example, at least one of glaucoma, Parkinson's
disease, Multiple Sclerosis, and head trauma.
[0047] FIG. 4B illustrates a simplified representation of a host
computer 440 that executes health correlation assessment program
(HCAP) software. In some embodiments, the HCAP is implemented as a
computer software program executing on a host computer 440. In some
embodiments, the host computer 410 is a network accessible computer
that provides computing support as part of the health correlation
assessment service 130. In other embodiments, the host computer is
a personal computer 150 separate from the FDT device 110. In other
embodiments, the host computer 440 is a computer that is embedded
within the FDT device 110.
[0048] In some embodiments (HCAP) software is stored as a computer
program onto portable media, such as a compact disc or universal
serial bus (USB) memory storage device. In some embodiments, the
HCAP software interoperates with an operating system, such as for
example, Microsoft Windows or LINIX operating systems, to access
contrast sensitivity test and other personal attribute data of
subjects to be health correlation assessed, to interact with a
health correlation assessment service (HCAS) via network
communication to access versions of parameters and/or algorithms
for the assessment, and to interact with a user via a user
interface.
[0049] In one embodiment, the HCAP software 470 is configured to
inter-operate with an operating system 460, such as for example, a
Microsoft Version 7 operating system via its application
programming interface. The HCAP software 470 directs operations of
a central processing unit (CPU) 452 that is accessible via a system
bus 450 residing within the architecture of the host computer 440.
User interface 462 hardware is available for the HCAP software 470
to direct interaction with a user and/or administrator of the host
computer.
[0050] Alternatively, the HCAP software 470 is programmed to
interoperate with other operating systems and/or computing
platforms, such as for example, to interoperate with variants of
UNIX, such as Linux or onto other Microsoft supplied operating
systems, various IBM, real-time, or other operating systems.
Preferably, the operating system 460 is designed to manage physical
memory 454 as virtual memory 456 within which the HCAP software 470
resides.
[0051] Optionally, the host computer 440 operates and interfaces
with a computer network 120 via input/output 458 and communication
device hardware 464. The network can be implemented as a local or
wide area network, to enable access to contrast sensitivity result
output files from a variety of locations that are remote to the
host computer 440.
[0052] In some embodiments, the HCAP 470 operates as part of a
health correlation assessment service (HCAS) 130, which is
implemented as a network accessible web site 130 via a private or
public computer network, such as the Internet. In some embodiments,
the method, apparatus or system of the invention, spans across
state or national boundaries and may interoperate with a network
that spans across state or national boundaries. For example, the
host computer 440 can be located in the Canada while contrast
sensitivity result output file data is accessed from locations
within the United States and/or Canada. The HCAP software can be
stored onto portable media, such as for example, a compact disc or
a universal serial bus compatible memory device. In some
embodiments, the host computer 440 is located within the FDT
contrast sensitivity device 110 or within computer 150.
[0053] FIGS. 5A-5B illustrate a health correlation assessment
procedure predictive algorithm (HCAPPA) coefficient generation
process in which multiple types of sample data are processed. In
one embodiment, this process is implemented as computer software,
like the HCAP 470 of FIG. 4B. The HCAPPA coefficient generator
software 510 processes a repository of sample data in order to
generate coefficients for a particular embodiment of an HCAPPA
algorithm.
[0054] In one embodiment, this coefficient generator 510 processes
the sample data in accordance with the correlation model of FIG.
6A, and with equations (0) through (3) of FIG. 6A, and outputs HCAP
algorithm coefficients 426 (a1, b1, a2, b2, a3 and b3) for
incorporation into equations (4) through (7) of FIG. 6A. The
equations and the variables and coefficients incorporated into
these equations quantify a statistical correlation between a
selected CST parameter value (See FIGS. 3A-3B) and a probability of
appropriate assignment of each subject receiving the CST parameter
value, to a particular clinical Alzheimer's related disease health
classification.
[0055] Note that as referred to herein, a Alzheimer's related
disease health classification includes pre-clinical stages of
Alzheimer's disease, such as cognitive complaint(s) (CC), and
clinical stages of the disease, such as mild cognitive impairment
(MCI). The Alzheimer's Disease classification itself includes mild,
moderate, and severe sub-classifications of the Alzheimer's Disease
in accordance with applicable standards associated with the
National Institute on Aging.
[0056] The health correlation assessment procedure predictive
algorithm (HCAPPA) coefficient generator 510 inputs clinical health
classification sample data 520, contrast sensitivity test result
sample data 530 and optionally inputs personal attribute sample
data 540 associated with each of a plurality of CST performing
subjects from the sample data repository 550. The coefficients are
determined so that the equations (0) through (4) incorporating
these coefficients define a mathematical probability function that
best fits with the sample data 530, 540 as plotted. Within these
equations (0) through (4) of FIG. 6A, the variable (x) represents
contrast sensitivity test (CST) values of one CST parameter. In
other embodiments, an alternate correlation model inputs one or
more values of a personal attribute, and/or inputs other CST
parameter values that can be incorporated and designed into
variations of equations like that of FIG. 6A
[0057] FIG. 5B illustrates a representation of the types of sample
data processed by HCAP predictive algorithm (HCAPPA) coefficient
generator 510. Each subject is identified by a unique identifier
532, has associated personal attributes 534, associated contrast
sensitivity test results 536 and an associated actual clinical
health classification 538. These types of sample data for a
plurality of subjects is stored within a repository. From these
sources of sample data, health correlation assessment procedure
algorithm (HCAPPA) coefficients 426 are determined and generated in
accordance with equations (0) through (4) of FIG. 6A via the HCAP
Data Fit algorithm HCAPDFA 512.
[0058] Referring back to FIGS. 1-5B, the HCAP 370 executes within
the virtual memory address space 454 of the host computer 410 where
ever the host computer 410 may reside. The HCAP 470 inputs and
processes information from a contrast sensitivity test result
output file 412, and outputs a probabilistic health classification
422 for the subject (viewer) 140 of the contrast sensitivity test.
The probabilistic health classification is stored into a file 422
which can be displayed onto a user interface 462 of the host
computer 440 and/or displayed onto a user interface of another
computer, such as computer 150 for example, via communication of
the probabilistic health classification to the another computer. In
some embodiments, the another computer can be embedded or directly
connected to the FDT device itself 110.
[0059] The HCAP executes an embodiment of a health correlation
assessment procedure predictive algorithm (HCAPPA). In some
embodiments, the HCAPPA is selectable from a plurality of different
HCAPPA algorithms, via a command directive from within a
configuration file or via a user interface by a user.
[0060] The scope of the invention described herein is not intended
to be limited to any one algorithm or set of algorithms and/or to
any set of procedures, variables and coefficients that are
incorporated into an algorithm. Nor is the invention intended to be
limited to each variable being defined to equal any particular CST
parameter or combination of one or more CST parameters and/or
personal attributes.
[0061] What is intended to be within the scope of the invention is
that the health correlation assessment procedure (HCAP) processes
at least one CST parameter value, of any type or of any combination
of types, to provide a set of one or more probabilistic health
classifications in response to the processed at least one CST
parameter value. Optionally, the at least one CST parameter can be
combined with other non-CST information (represented as non-CST
parameters), such as for example, one or more attributes (age,
gender, ethnicity, family history, work history, genetic or other
types of biomarkers), of any type or of any combination of types,
associated with the CST test taking subject, to provide the one or
more probabilistic health classifications providing at least a
minimum desired utility and accuracy of prediction.
[0062] The probabilistic health classifications provide a relative
probability (prospect) of the first subject being appropriately
assigned into at least one Alzheimer's disease related clinical
health classification or at least one pre-clinical Alzheimer's
disease classification. The relative probability can be expressed
in a quantitative or qualitative manner to users of the health
correlation assessment procedure (HCAP) and of the health
correlation assessment system (HCAS). The relative probability or
prospect provided by the probabilistic health classification is
designed to provide information that reduces uncertainty regarding
an appropriate clinical health classification to which the CST
taking subject should be assigned. This probabilistic health
classification information is provided in a manner that is separate
from other information that could be acquired by clinical testing,
but where such other information at a typically acquired at a much
higher cost to the CST test taking subject.
[0063] In other embodiments, other algorithms can be designed to
process a variety of one or more CST parameters in various
combinations, and optionally in combination with one or more
non-CST parameters such as personal attributes. Each CST parameter
and personal attribute can be represented by one variable (x1, x2,
. . . ) within an algorithm, or CST parameters and/or personal
attribute values can be combined mathematically to define one
variable (x1, x2, . . . ) within the algorithm. As the amount of
Alzheimer's Disease related clinical data grows, it would be
expected that specific algorithms would evolve to better exploit or
attempt to exploit, a repository of Alzheimer's Disease related
data of increasing size. It is foreseeable that many different
types of algorithms can be employed with varying accuracy and
effectiveness.
[0064] In one embodiment, the HCAPPA 424 selects and inputs one or
more contrast sensitivity test (CST) parameters, also referred to
herein as CST measurements, and outputs and a set of probabilistic
health classifications for the subject performing the contrast
sensitivity test. For example, the HCAPPA inputs a
RightDurationTime CST parameter (See FIG. 3A) and outputs a set of
probabilistic health classifications for the subject performing the
contrast sensitivity test, based upon a quantified correlation
between a contrast sensitivity test result value for the
RightDurationTime parameter and clinical health classifications of
sample subjects who also have known associated RightDurationTime
CST parameter measurement values.
[0065] The RightDurationTime is defined as the time required for
the subject to complete an entire contrast sensitivity test of the
right eye. The amount of time required to complete an entire
contrast sensitivity test is variable for each test subject and
each eye. The right eye is typically the first eye to be tested,
while the left eye is typically the second and last eye of the
subject to be tested within the contrast sensitivity test.
[0066] Aside from the selection of a CST parameter, the design of
the HCAPPA can vary with respect to a set of one or more
probabilistic health classification(s) for which to predict. In one
embodiment, the probabilistic health classification(s) include a
first classification of Normal Health, and a second classification
of Cognitive Complaint, third classification of Mild Cognitive
Impairment and a fourth classification of Alzheimer's Disease.
[0067] In one embodiment, the HCAPPA outputs a probability of the
subject's being appropriately classified into each of these (4)
health classifications, based upon the value of the
RightDurationTime contrast sensitivity (CS) measurement parameter.
Tn other embodiments, the HCAA outputs a probabilistic health
classification based upon another set of one or more contrast
sensitivity test (CST) measurements.
[0068] As an example, in one embodiment, the subject may receive a
probabilistic health classification including the following set of
probabilities: 10% of being Normal Health, 20% of being Cognitive
Complaint, 40% of being Mild Cognitive Impairment, and 30% of being
Alzheimer's disease. Alternatively, in another embodiment, the same
probabilistic health classification can be represented by the above
listed probabilities as 10% of being Normal Health and 90% of being
of Non-Normal Health.
[0069] Appropriate assignment of a subject into a clinical health
classification, depends upon other information (indicia) associated
with the subject that is obtained from sources other than from an
eye contrast sensitivity examination. This other information is
obtained from clinical testing. Clinical testing involves
performance of various clinical measurements and tests, including,
for example, a beta-amyloid plaque deposit measurement,
neurofibrillary tangle measurement, a cortex thickness measurement,
volume and density of the hippocampus, and/or one or more cognitive
tests performed upon the subject.
[0070] These clinical measurements and tests are performed in
accordance with applicable standards and the results of which are
interpreted by Alzheimer's disease and/or related disease
specialists to decide appropriate placement of a subject into a
particular clinical health classification. Such appropriate
placement is referred to herein as an actual health classification,
as opposed to a predicted health classification which results from
exercise of the health correlation assessment algorithm (HCAA)
alone.
[0071] The beta-amyloid plaque deposit and neurofibrillary tangle
measurement can be determined via employment of a positron emission
tomography (PET) scan using Pittsburgh Compound B (PiB) as a
fluorescent trace substance. The cortex thickness measurement and
volume and density of the hippocampus can be performed via an MRI
scan. Cognitive tests can include the Mini-Mental State Examination
(MMSE) and what is known as an MMSC cognitive test.
[0072] A combination of the above sources of other information is
interpreted by the specialist(s) to determine whether a subject is
appropriately classified into one ofa set of defined health
classifications. For example, a first set of health classifications
includes Normal Health, Cognitive Complaint, Mild Cognitive
Impairment and Alzheimer's disease health classifications.
[0073] In yet another embodiment, the probabilistic health
classifications include only normal health and non-normal health.
In this embodiment, subjects that would otherwise be classified as
Cognitive Complaint or Mild Cognitive Impairment or as having
Alzheimer's Disease would instead be classified as being of
Non-Normal Health.
[0074] As referred to herein, Normal Health indicates the absence
of clinical evidence for Alzheimer's disease and Mild Cognitive
Impairment and Cognitive Complaint(s). However, a subject referred
to herein as being appropriately classified as having Normal Health
may be possibly afflicted with other non-Alzheimers disease related
ailments, such as Hepatitis C or a broken bone, for example.
However, there is no known strong correlation between a contrast
sensitivity examination and Hepatitis C or a broken bone within the
subject and such non-Alzheimer's disease related ailments are
ignored within the particular disclosed embodiments of the
correlation model.
[0075] In another embodiment of the HCAPPA, the set of
probabilistic health classifications exclude Cognitive Complaint
and include Normal Health, Mild Cognitive Impairment and
Alzheimer's disease. In this embodiment, subjects that would
otherwise be classified as Cognitive Complaint would be instead
classified as being of Normal Health, Mild Cognitive Impairment or
Alzheimer's Disease.
[0076] FIG. 6A illustrates a first embodiment of a correlation
model upon which a first embodiment of a health correlation
assessment procedure predictive algorithm (HCAPPA) is based upon.
As shown, P.sub.1, also referred to herein as P1, represents a
probability of a subject's being clinically classified as actually
having Normal Health. P.sub.2, also referred to herein as P2,
represents a probability of a subject being clinically classified
as actually having Cognitive Complaint(s) (CC). P.sub.3, also
referred to herein as P3, represents a probability of a subject's
being clinically classified as actually having Mild Cognitive
Impairment (MCI). P.sub.4, also referred to herein as P4,
represents a probability of a subject's being clinically classified
as actually having Alzheimer's Disease. The Alzheimer's Disease
classification includes mild, moderate, and severe
sub-classifications of the Alzheimer's Disease in accordance with
applicable standards associated with the Alzheimer's Disease
Association.
[0077] In accordance with this correlation model, the above (4)
probabilities (P1, P2, P3 and P4) are each a function of a selected
CST parameter value, represented by the variable (x). Equation (0)
indicates that P1+P2+P3+P4=1, meaning that for given value of (x),
the (4) probabilities sum to the value 1 and there is no overlap
between the probabilities associated with each of the (4) health
classifications. Each probability represents a predictive
likelihood that a subject (viewer) 140 receiving a CST parameter
value (x), would be appropriately assigned to a particular health
classification based upon clinical test and measurements performed
separately from the contrast sensitivity test (CST).
[0078] In accordance with this correlation model, a subject
(viewer) 140 can only be appropriately assigned to one health
classification of the set consisting of Normal Health. Cognitive
Complaint, Mild Cognitive Impairment and Alzheimer's Disease. A
health care practitioner may advise that the subject should seek
clinical testing to verify that the subject should or should not be
assigned to the classification in which the subject receives the
highest predictive probability from the HCAP. However the subject
may also have other smaller and predictive probabilities of being
appropriately assigned to one or more other health classifications
within this probabilistic set, for which clinical testing may also
be advisable.
[0079] Equation (1) indicates that the logarithm of the ratio of
P2(x) divided by P1(x), which is equivalent to the logarithm of
P2(x) minus the logarithm of P1(x), is expressed as a linear
function. This linear function is defined as (a.sub.1x+b.sub.1).
Notice that within this correlation model, P1(x) and P2(x) are not
limited to being linear functions and are free to be non-linear
functions to better fit contrast sensitivity data. However, this
model does require that the difference in the logarithm of P1(x)
and logarithm of P2(x) be expressed as a linear function.
[0080] Equation (2) indicates that the logarithm of the ratio of
P3(x) divided by P1(x), which is equivalent to the logarithm of
P3(x) minus the logarithm of P1(x), is expressed as a linear
function. This linear function is defined as (a.sub.2x+b.sub.2).
Notice that within this correlation model, P1(x) and P3(x) are not
limited to being linear functions and are free to be non-linear
functions to better fit contrast sensitivity data. However, this
model does require that the difference in the logarithm of P1(x)
and logarithm of P3(x) be expressed as a linear function.
[0081] Equation (3) indicates that the logarithm of the ratio of
P4(x) divided by P (x), which is equivalent to the logarithm of
P4(x) minus the logarithm of P1(x), is expressed as a linear
function. This linear function is defined as (a.sub.3x+b.sub.3).
Notice that within this correlation model, P1(x) and P4(x) are not
limited to being linear functions and are free to be non-linear
functions to better fit contrast sensitivity data. However, this
model does require that the difference in the logarithm of P1(x)
and logarithm of P4(x) be expressed as a linear function.
[0082] In some embodiments, the coefficients a1, b1, a2, b2, a3 and
b3 are computed using a HCAP data fit algorithm HCAPDFA 512
selected as the Newton-Raphson algorithm [1] by fitting the above
defined correlation model (P1(x) through P4(x) probability
equations) to the contrast sensitivity data. The CST (x) values are
plotted with respect to a first probability (P) axis and with
respect to a second (x) value axis, where the first axis and second
axis are orthogonal to each other. CST parameter (x) values that
are known to be associated with subjects known to be clinically
assigned to a particular health classification are plotted to have
a probability of 1 of being associated with a subject (viewer)
assigned to that particular health classification, and other (x)
values for that CST parameter are plotted to have an associated
probability of 0 for being associated with a subject that is
clinically assigned to that particular health classification.
[0083] For example, CST parameter (x) values that are known to be
associated with subjects known to be clinically diagnosed as having
Alzheimer's disease are assigned a probability of 1 for being
associated with a subject that is clinically diagnosed as having
Alzheimer's disease, while all other (x) values for that CST
parameter are plotted to have an associated probability of 0 for
being associated with a subject that is clinically diagnosed as
having Alzheimer's disease.
[0084] In one embodiment, an implementation of such a best fit
algorithm can be performed by a Matlab function, called mnrfit( ),
to estimate the values of the coefficients. The mnrfit( ) function
iteratively determines the coefficients (a1, b1, a2, b2, a3 and b3)
that best fit a set of contrast sensitivity data in combination
with actual clinical health classification information of each
subject associated with the contrast sensitivity data.
[0085] For example, hypothetically, if subject number 0234 was
clinically diagnosed with Alzheimer's disease and has an associated
RightDurationTime CST parameter value of 350 seconds, then the
HCAPPA coefficient generator, via the HCAP Data Fit algorithm
(HCAPDFA) 512, factors this unit of data, along with other units of
sample data, to determine the values of the coefficients (a1, b1,
a2, b2, a3, and b3) that best fit the combination of both the
correlation model defined with equations P1(x), P2(x), P3(x) and
P4(x) and available contrast sensitivity test result data.
[0086] Once these HCAPPA coefficients are determined, the HCAPPA
and the coefficients incorporated into the HCAPPA are employed to
process the CST output file 412 of a new subject (viewer) 140 not
having any known associated clinical health classification. The
HCAPPA outputs a set of probabilistic health classifications which
predict what clinical health classification(s) the subject (viewer)
140 would likely be appropriately assigned to, if clinically
evaluated.
[0087] For example, a new subject would have no known prior testing
with respect to Alzheimer's disease (AD), Mild Cognitive Impairment
(MCI) or Cognitive Complaint (CC), but would have associated
contrast sensitivity (CS) test result data. From this CS test data,
a set of probabilistic health classifications can be determined for
that new subject with respect to Normal Health, Alzheimer's
disease, Mild Cognitive Impairment or Cognitive Complaint based
upon a correlation model as described herein.
[0088] Each of the equations (4) through (7) operate as a portion
of the health correlation assessment procedure predictive algorithm
(HCAPPA). Equation (4) determines the probability of appropriately
classifying a subject as having clinical Normal Health (NH) based
upon the contrast sensitivity test result data for that subject.
Equation (5) determines the probability of appropriately
classifying a subject as having clinical Cognitive Complaint(s)
based upon the contrast sensitivity test result data for that
subject. Equation (6) determines the probability of appropriately
classifying a subject as having clinical Mild Cognitive Impairment
(MCI) based upon the contrast sensitivity test result data for that
subject. Equation (7) determines the probability of appropriately
classifying a subject as having clinical Alzheimer's Disease based
upon the contrast sensitivity test result data for that
subject.
[0089] FIG. 6B illustrates a representation of the types of data
input, processed and generated by the health correlation assessment
procedure 470 HCAP. Each subject is identified by a unique
identifier 612, has associated personal attributes 614, associated
contrast sensitivity test results 616. The aforementioned data is
obtained from CST output file 412 and optionally another source of
supplemental data, such as supplemental personal attribute
data.
[0090] In response to input and processing of data 612-616, the
HCAP 470, employing a selected predictive algorithm (HCAPPA) 424
generates outputs a set of one or more probabilistic health
classifications 618 in accordance with equations (4) through (7) of
FIG. 6A. As shown, (4) probabilistic health classifications are
listed for each subject. For a first subject (Number 0395012), a
set of probabilistic health classifications include Alzheimer's
Disease (AD) having a probability equal to (53%), mild cognitive
impairment (MCI) equal to (28%), cognitive complaint(s) (%15) and
normal health (4%).
[0091] Note that these probabilistic health classifications are not
actual health classifications 518 listed in FIG. 5B). For example,
hypothetically, if first subject (Number 0395012) received clinical
testing and evaluation, this first subject may be actually and
clinically classified as MCI, despite the HCAP predicting that AD
is most probable. However, use of another algorithm by the HCAP
470, or use of the same algorithm with other coefficients that are
generated based upon a larger and or more diverse repository of
sample data, may yield a more accurate prediction that MCI is the
most likely appropriate health classification.
[0092] FIG. 7A illustrates ranges of sample data representing a
RightDurationTime values with respect to (83) subjects assigned to
(4) clinical health classifications. The RightDurationTime is
defined as the elapsed time required to complete a contrast
sensitivity test for the right eye of one subject. This data, also
referred to herein as sample data, is acquired as a portion of a
contrast sensitivity test result for each of the (83) subjects,
also referred to herein as sample subjects. The contrast
sensitivity test result data for each of the (83) sample subjects,
can be represented in other ways, including for example,
represented as a two dimensional distribution, a chi-squared
distribution and/or measured via mean and/or median values,
standard deviation values etc.
[0093] The sample data includes (29) sample subjects clinically
classified being of Normal Health (NH), (18) sample subjects
classified as having Cognitive Complaints (CC), (27) sample
subjects classified as having mild cognitive impairment (MCI) and
(9) sample subjects classified as having Alzheimer's disease. This
sample data is processed to determine the health classification
assessment algorithm (HCAA) parameters (a1, b1, a2, b2, a3, b3, a4
and b4) using a best fit algorithm employing the equations (0)
through (3) of FIG. 6A.
[0094] As shown, a value of the Right Duration Time measurement of
subjects having Normal Health (NH), have a range 710 between
291-333 seconds and have a mean 720 of approximately 312 seconds.
The RightDurationTime measurement for subjects having cognitive
complaints (CC) have a range 712 between 297-346 seconds and have a
mean 722 of approximately 321.5 seconds. The RightDurationTime
measurement for subjects having mild cognitive impairment (MCI)
have a range 714 between 303-355 seconds and have a mean 724 of
approximately 329 seconds. The RightDurationTime measurement for
subjects having Alzheimer's disease have a range 716 between
315-384 seconds and have a mean 726 of approximately 349.5 for
subjects having Alzheimer's disease.
[0095] A shown, the sample subjects that are clinically classified
as actually having Alzheimer's Disease (AD) typically, require more
RightDurationTime on average than that of subjects that are
classified as actually having Mild Cognitive Impairment (MCI), who
typically require more RightDurationTime on average than subjects
that are clinically classified as actually having Cognitive
Complaint(s), who typically require more RightDurationTime on
average than subjects that are clinically classified as actually
having Normal Health (NH).
[0096] Despite the above observation, there is much overlap, and
hence ambiguity, when interpreting this data with the naked eye. A
new subject requiring a RightDurationTime of between 300 and 340
seconds falls within at least (3) different health classifications.
Also for example, a hypothetical subject requiring a
RightDurationTime value 730 of 325 seconds (falls within the range
of sample data for all (4) clinical health classifications. Many
other hypothetical RightDurationTime values fall within the range
of at least (2) clinical health classifications. As a result,
despite any perceived visual correlation, quantifying an accurate
correlation between the RightDurationTime and appropriate
assignment to one of (4) clinical health classifications is not
clear from observation alone with respect to this sample data.
[0097] FIG. 7B illustrates ranges of sample data representing
PeripheryRightEye contrast sensitivity measurement values with
respect to the (83) sample subjects that are each assigned to (4)
clinical health classifications. The PeripheryRightEye measurement
is defined as the average of contrast sensitivity thresholds of
(38) retinal zones that are located within the periphery area of
the right eye (Sec FIGS. 3A-3B). The PeripheryRightEye data for
each of the (83) sample subjects, can be represented in other ways,
including for example, represented as a two dimensional
distribution, a chi-squared distribution and/or measured via mean
and/or median values, standard deviation values etc.
[0098] The sample data includes (29) sample subjects clinically
classified being of Normal Health (NH), (18) sample subjects
classified as having Cognitive Complaints (CC), (27) sample
subjects classified as having mild cognitive impairment (MCI) and
(9) sample subjects classified as having Alzheimer's disease. This
sample data is processed to determine the health classification
assessment algorithm (HCAA) parameters (a1, b1, a2, b2, a3 and b3)
using a best fit algorithm employing the equations (0) through (3)
of FIG. 6A.
[0099] As shown, a value of the PeripheryRightEye measurement of
subjects having Normal Health (NH), have a range 760(between
19.28-28.73 dB and have a mean 770 of approximately 24.00. The
PeripheryRightEye measurement for subjects having cognitive
complaints (CC) have a range 762 between 18.21-30.23 dB and have a
mean 772 of approximately 24.22. The PeripheryRightEye measurement
for subjects having mild cognitive impairment (MCI) have a range
764 between 12.86-28.31 dB and have a mean 774 of approximately
20.59 dB. The PeripheryRightEye measurement for subjects having
Alzheimer's disease have a range 766 between 11.55-23.31 and have a
mean 776 of approximately 17.43 dB for subjects having Alzheimer's
disease.
[0100] A shown, the sample subjects that are clinically classified
as actually having Alzheimer's Disease (AD), have a lower average
PeripheryRightEye CST parameter value than that of subjects that
are classified as actually having Mild Cognitive Impairment (MCI),
who have a lower average PcripheryRightEye CST parameter value than
sample subjects that are clinically classified as actually having
Cognitive Complaint(s), who have approximately the same average
PeripheryRightEye CST parameter as sample subjects that are
clinically classified as actually having Normal Health (NH).
[0101] Despite the above observations, there is much overlap and
uncertainty, when visually interpreting this sample data in
relation with the (4) clinical health classifications. For example,
a new subject having a PcripheryRightEye value of between 18.3 and
23.3 dB falls within (3) different health classifications within
the sample data. Also for example, a hypothetical subject having a
PeripheryRightEye value 780 of 21 dB, falls within all (4) clinical
health classifications. Many other hypothetical PeripheryRightEye
values fall within the range of at least (2) clinical health
classifications. As a result, despite any perceived visual
correlation, accurately quantifying a correlation between any
PeripheryRightEye contrast sensitivity measurement from this sample
data and appropriate assignment to one of (4) clinical health
classifications is not clear from visual observation alone.
[0102] FIG. 7C illustrates ranges of sample data representing an
age personal attribute with respect to the (83) sample subjects
that are each assigned to (4) clinical health classifications. This
age parameter is technically a non-contrast sensitivity test
(non-CST) parameter. However, the age of a test taking subject can
be extracted from a Zeiss FDT output file 412 in the same automated
fashion of a CST parameter.
[0103] The sample data includes (29) sample subjects clinically
classified being of Normal Health (NH), (18) sample subjects
classified as having Cognitive Complaints (CC). (27) sample
subjects classified as having mild cognitive impairment (MCI) and
(9) sample subjects classified as having Alzheimer's disease. This
sample data is processed to determine the health classification
assessment algorithm (HCAA) parameters (a1, b1, a2, b2, a3 and b3)
using a best fit algorithm employing the equations (0) through (3)
of FIG. 6A.
[0104] As shown, a value of the age of subjects having Normal
Health (NH), have a range 780 between 63-86 years of age and have a
mean 770 of approximately 74.5 years. The age for subjects having
cognitive complaints (CC) have a range 782 between 62-83 years of
age and have a mean 792 of approximately 72.5 years. The age for
subjects having mild cognitive impairment (MCI) have a range 784
between 61-88 years and have a mean 794 of approximately 74.5
years. The age for subjects having Alzheimer's disease have a range
786 between 60-83 years of age and have a mean 796 of approximately
71.5 years for subjects having Alzheimer's disease.
[0105] A shown, the sample subjects that are clinically classified
as actually having Alzheimer's Disease (AD), have an age range that
is approximate to that of subjects that are classified as actually
having Mild Cognitive Impairment (MCI), who have an age range that
is approximate to that of subjects that are clinically classified
as actually having Cognitive Complaint(s), who have an age range
that is approximate to that of average sample subjects that are
clinically classified as actually having Normal Health (NH). As a
result, there does not appear to be much of any visual correlation,
between the age of the CST performing subject and their appropriate
assignment to one of (4) clinical health classifications.
[0106] However, a combining and plotting values one or more
individual a CST parameters, with the age value for each of the CST
performing subjects, may reveal a mathematical significant
correlation between the combination of CST parameters and age, for
each particular subject, that can be quantified and exploited
within an accordingly designed embodiment of a HCAPPA
algorithm.
[0107] FIGS. 8A-8C illustrate graphs of a probability of a
subject's being appropriately classified as having one of four
health classifications as a function of a RightDurationTime
contrast sensitivity test parameter value.
[0108] FIG. 8A illustrates a graph 810 of a probability (P1) of a
subject (viewer) being appropriately assigned to a normal health
(NH) classification (having normal health in accordance with a
clinical standard) as a function of a RightDurationTime CST
parameter value. This graph 810 is computed in accordance with the
correlation model equation (4) of FIG. 6A and sample data of (83)
subjects The center line 820 is a function representing a computed
probability, indicated on a vertical axis 814 of the graph 810, of
a subject appropriately being clinically classified as having
Normal Health, also referred to here as a healthy control (HC)
subject.
[0109] This computed probability 820 is a function of a
RightDurationTime value received by a subject and indicated on a
horizontal axis 812 of the graph 810. The error bars vertically
spanning across the center line 820 indicate a 95% confidence range
for each probability value located along an intersection with the
center line 820. In other words, based upon this mathematical model
and its assumptions, there is 95% confidence that the actual
probability for each RightDurationTime CST value indicated along
axis 812, falls within the probability range indicated by an upper
limit 822 and a lower limit 824 of the error bars vertically
spanning across the center line 820.
[0110] The coefficient values for (a1, a2, a3) that are computed
based upon the above described best fit algorithm and the
RightDurationTime sample data are (-0.135, -0.076, -0.050). The
coefficient values for (b1, b2, b3) that are computed based upon
the above described best fit algorithm and the sample data are
(44.330, 25.469, 17.644). Incorporating these aforementioned
coefficient values into equation (4) of FIG. 6A yields the graph of
probability (P1) shown here.
[0111] Graph 830 is of a probability (P.sub.2) of a subject
(viewer) being appropriately assigned to a cognitive complaint (CC)
classification (having cognitive complaints assessed in accordance
with a clinical standard) as a function of a RightDurationTime CST
parameter value. This graph is computed in accordance with the
correlation model equation (5) of FIG. 6A and sample data of (83)
subjects The center line 840 is a function representing a computed
probability, indicated on a vertical axis 834 of the graph 830, of
a subject appropriately being clinically classified as having
Cognitive Complaints.
[0112] This computed probability 840 is a function of a
RightDurationTime value received by a subject and indicated on a
horizontal axis 832 of the graph 830. The error bars vertically
spanning across the center line 840 indicate a 95% confidence range
for each probability value located along an intersection with the
center line 840. In other words, based upon this mathematical model
and its assumptions, there is 95% confidence that the actual
probability for each RightDurationTime CST value indicated along
axis 812, falls within the probability range indicated by an upper
limit 842 and a lower limit 844 of the error bars vertically
spanning across the center line 840.
[0113] FIG. 8B illustrates a graph 850 of a probability (P3) of a
subject (viewer) being appropriately assigned to a mild cognitive
impairment (MCI) health classification (assigned in accordance with
a clinical standard) as a function of a RightDurationTime CST
parameter value. This graph 850 is computed in accordance with the
correlation model equation (6) of FIG. 6A and sample data of (83)
subjects The center line 860 is a function representing a computed
probability, indicated on a vertical axis 854 of the graph 860, of
a subject appropriately being clinically classified as having Mild
Cognitive Impairment (MCI).
[0114] This computed probability 860 is a function of a
RightDurationTime value received by a subject and indicated on a
horizontal axis 852 of the graph 850. The error bars vertically
spanning across the center line 860 indicate a 95% confidence range
for each probability value located along an intersection with the
center line 860. In other words, based upon this mathematical model
and its assumptions, there is 95% confidence that the actual
probability for each RightDurationTime CST value indicated along
axis 852, falls within the probability range indicated by an upper
limit 882 and a lower limit 864 of the error bars vertically
spanning across the center line 860.
[0115] Graph 870 is of a probability (P.sub.4) of a subject
(viewer) being appropriately assigned to an Alzheimer's Disease
(AD) health classification (assigned in accordance with a clinical
standard) as a function of a RightDurationTime CST parameter value.
This graph 870 is computed in accordance with the correlation model
equation (7) of FIG. 6A and sample data of (83) subjects The center
line 880 is a function representing a computed probability,
indicated on a vertical axis 884 of the graph 870, of a subject
appropriately being clinically classified as having Alzheimer's
Disease (AD).
[0116] This computed probability 880 is a function of a
RightDurationTime value received by a subject and indicated on a
horizontal axis 872 of the graph 870. The error bars vertically
spanning across the center line 880 indicate a 95% confidence range
for each probability value located along an intersection with the
center line 880. In other words, based upon this mathematical model
and its assumptions, there is 95% confidence that the actual
probability for each RightDurationTime CST value indicated along
horizontal axis 872, falls within the probability range indicated
by an upper limit 882 and a lower limit 884 of the error bars
vertically spanning across the center line 880.
[0117] FIG. 8C illustrates a graph 890 of probabilities (P1, P2,
P3, P4) of FIGS. 8A-8B superimposed onto one graph 890.
[0118] FIG. 9 is a matrix illustrating a relationship between the
results of the predicted and actual health classifications based
upon a single CST (RightDurationTime) parameter value. As shown,
(46 of 83) subjects (i.e. 26+11+8+1=46) are predicted as more
likely of being clinically classified as having normal health than
likely of being clinically classified within the CC, MCI or
Alzheimer's disease classifications. Also, (0 of 83) subjects are
predicted to be more likely of being clinically classified as
having cognitive complaints (CC) than likely of being clinically
classified as not having cognitive complaints. Also, (34 of 83)
subjects (i.e. 3+7+17+7) are predicted as more likely of being
clinically classified as having mild cognitive impairment (MCI)
than likely of being clinically classified as not having mild
cognitive impairment (MCI). Also, (3 of 83) subjects (i.e. 2+1) are
predicted as more likely of being clinically classified as having
Alzheimer's disease than likely of being clinically classified as
not having Alzheimer's disease.
[0119] From this analysis (26 of 29) subjects that are of actual
(clinical) normal health, are predicted as having normal health,
and (7 of 18) subjects actually having (clinical) cognitive
complaints, are predicted as having cognitive complaints, and (17
of 27) subjects actually having mild cognitive impairment (MCI) are
predicted as having mild cognitive impairment (MCI) and finally,
only (1 of 9) sample subjects actually having clinical Alzheimer's
Disease (AD) are predicted as having Alzheimer's Disease (AD).
[0120] The sensitivity of normal health (NH) prediction is (26/29)
89.6%, the sensitivity of the cognitive complaint (CC) prediction
is (0/18) 0%, the sensitivity of the mild cognitive impairment
(MCI) prediction is (17/27) 62.9%, the sensitivity of the AD
prediction is (1/9) 11.1%. Note that the sensitivity of the
combination of MCI plus AD is (27/36) 75%. The sensitivity of the
combination of CC, MCI plus AD is (34/54)=62.9%.
[0121] What the aforementioned sensitivity values indicate is that
this particular RightDurationTime single CST parameter correlation
model is highly sensitive to subjects having actual normal health,
but is insensitive to subjects with cognitive complaints (CC),
somewhat sensitive to subjects having actual clinical Mild
Cognitive Impairments and much less sensitive to subjects having
actual clinical Alzheimer's disease. However, there is substantial
sensitivity (75%) for the combination of MCI and AD, which is
arguably a combination of Alzheimer's Disease and pre-Alzheimer's
Disease health classifications.
[0122] It is desired that this correlation model evolve to achieve
higher sensitivity for clinically verifiable CC, MCI and AD so that
fewer subjects in these clinically verifiable categories are not
recognized by the HCAPPA. This is desired, even if it comes that
the cost of lower NH sensitivity, given that the cost penalty for
not recognizing normal health is advising and performing an
unnecessary Alzheimer's disease related clinical test. Conversely,
the cost penalty for not recognizing an Alzheimer's related disease
classification can be unnecessary progression of the Alzheimer's
disease within early treatment, which could be much more
costly.
[0123] FIG. 10 illustrates a multiple CST parameter correlation
model and correlation assessment procedure predictive algorithm
(HCAPPA). In this second embodiment, like the first embodiment. P1
(P.sub.1) represents a probability of a subject's being clinically
classified as actually having Normal Health. P2 (P.sub.2)
represents a probability of a subject being clinically classified
as actually having Cognitive Complaint(s) (CC). P3 (P.sub.3)
represents a probability of a subject's being clinically classified
as actually having Mild Cognitive Impairment (MCI). P4 (P.sub.4)
represents a probability of a subject's being clinically classified
as actually having Alzheimer's Disease. The Alzheimer's Disease
classification includes mild, moderate, and severe
sub-classifications of the Alzheimer's Disease in accordance with
applicable standards associated with the Alzheimer's Disease
Association.
[0124] Unlike the first embodiment of the HCAPPA, the above (4)
probabilities (P1, P2, P3 and P4) are each a function of multiple
(5) selected CST parameters, represented by the variables (x1, x2,
x3, x4 and x5). These CST parameters are RightDurationTime (x1),
PeriCenRatioRightEye (x2), SupInfRatioRightEye (x3),
MacularArcRightEye (x4) and SuperiorQuadRatioRightEye (x5).
[0125] As shown, the PeriCenRatioRightEye CST parameter (x2) is a
function of the PeripheryRightEye and CentralRightEye CST
parameters. The SupInfRatioRightEye (x3) CST parameter is a
function of the SuperiorRightEye and InferiorRightEye CST
parameters. The SuperiorQuadRatioRightEye (x5) CST parameter is a
function of the SuperiorQuadRightEye1 and the SuperiorQuadRightEye2
CST parameters.
[0126] Like the first embodiment of the HCAPPA, Equation (0)
indicates that P1+P2+P3+P4=1, meaning that for given set of CST
parameter values of (x1, x2, x3, x4 and x5), the (4) probabilities
sum to the value 1 and there is no overlap between the
probabilities associated with each of the (4) health
classifications. Each probability represents a predictive
likelihood that a subject (viewer) 140 receiving a given set of CST
parameter values (x1, x2, x3, x4 and x5), would be appropriately
assigned to a particular health classification associated with that
probability. based upon clinical test and measurements performed
separately from the contrast sensitivity test (CST).
[0127] In accordance with this correlation model, a subject
(viewer) 140 can be predictably assigned to health classifications
of the set consisting of Normal Health, Cognitive Complaint, Mild
Cognitive Impairment and Alzheimer's Disease. In response to an
output of the HCAPPA, a health care practitioner may advise that
the subject should seek clinical testing to verify that the subject
should or should not be assigned to the classification in which the
subject receives the highest predictive probability from the HCAP.
However the subject may also have other smaller and predictive
probabilities of being appropriately assigned to one or more other
health classifications within this probabilistic set, for which
clinical testing may also be advisable.
[0128] Equation (11) indicates that the logarithm of the ratio of
P2 (x1, x2, x3, x4, x5) divided by P1 (x1, x2, x3, x4, x5), which
is equivalent to the logarithm of P2 (x1, x2, x3, x4, x5) minus the
logarithm of P1 (x1, x2, x3, x4, x5), is expressed as a linear
function. This linear function is defined as
(a.sub.11x.sub.1+a.sub.12x.sub.2+a.sub.13x.sub.3+a.sub.14x.sub.4+a.sub.15-
x.sub.5+b.sub.1). Notice that within this correlation model, P1(x)
and P2(x) are not limited to being linear functions and are free to
be non-linear functions to better fit contrast sensitivity data.
However, this model does require that the difference in the
logarithm of P1(x) and logarithm of P2(x) be expressed as a linear
function.
[0129] Equation (2) indicates that the logarithm of the ratio of
P3(x) divided by P1(x), which is equivalent to the logarithm of
P3(x) minus the logarithm of P1(x), is expressed as a linear
function. This linear function is defined as
(a.sub.21x.sub.1+a.sub.22x.sub.2+a.sub.23x.sub.3+a.sub.24x.sub.4+a.sub.25-
x.sub.5+b.sub.2) Notice that within this correlation model, P1(x)
and P3(x) are not limited to being linear functions and are free to
be non-linear functions to better fit contrast sensitivity data.
However, this model does require that the difference in the
logarithm of P1(x) and logarithm of P3(x) be expressed as a linear
function.
[0130] Equation (3) indicates that the logarithm of the ratio of
P4(x) divided by P1(x), which is equivalent to the logarithm of
P4(x) minus the logarithm of P1(x), is expressed as a linear
function. This linear function is defined as
(a.sub.31x.sub.1+a.sub.32x.sub.2+a.sub.33x.sub.3+a.sub.34
x.sub.4+a.sub.35x.sub.5+b.sub.3). Notice that within this
correlation model, P1(x) and P4(x) are not limited to being linear
functions and are free to be non-linear functions to better fit
contrast sensitivity data. However, this model does require that
the difference in the logarithm of P1(x) and logarithm of P4(x) be
expressed as a linear function.
[0131] The coefficients (a.sub.11, a.sub.12, a.sub.13, a.sub.14,
a.sub.15), (a.sub.21, a.sub.22, a.sub.23, a.sub.24, a.sub.25),
(a.sub.31, a.sub.32, a.sub.33, a.sub.34, a.sub.35) and (b.sub.1,
b.sub.2, b.sub.3) are computed using Newton-Raphson algorithm [1]
by fitting the above defined correlation model to the contrast
sensitivity data. The CST (x.sub.1, x.sub.2, x.sub.3, x.sub.4,
x.sub.5) values are virtually plotted within a multidimensional
space with respect to a first probability (P) axis. A set of CST
parameter (x.sub.1, x.sub.2, x.sub.3, x.sub.4, x.sub.5) values that
are known to be associated with a subject known to be clinically
assigned to a particular health classification, are plotted as
having a probability equal to 1 of being associated with a subject
(viewer) assigned to that particular clinical health
classification, and other sets of (x.sub.1, x.sub.2, x.sub.3,
x.sub.4, x.sub.5) CST parameter values are virtually plotted to
have an associated probability equal to 0 for being associated with
a subject that is clinically assigned to that particular health
classification.
[0132] For example, a set of CST parameter (x.sub.1, x.sub.2,
x.sub.3, x.sub.4, x.sub.5) values that is known to be associated
with a subject known to be clinically diagnosed as having
Alzheimer's disease are assigned a probability of 1 for being
associated with a subject that is clinically diagnosed as having
Alzheimer's disease, while other sets of (x.sub.1, x.sub.2,
x.sub.3, x.sub.4, x.sub.5) CST parameter values are virtually
plotted to have an associated probability of 0, for being
associated with a subject that is clinically diagnosed as having
Alzheimer's disease.
[0133] In one embodiment, an implementation of such a best fit
algorithm can be performed by a Matlab function, called mnrfit( ),
to estimate the values of the coefficients (a.sub.11, a.sub.12,
a.sub.13, a.sub.14, a.sub.15), (a.sub.21, a.sub.22, a.sub.23,
a.sub.24, a.sub.25), (a.sub.31, a.sub.32, a.sub.33, a.sub.34,
a.sub.35) and (b.sub.1, b.sub.2, b.sub.3). The mnrfit( ) function
iteratively determines the coefficients that best fit a set of
sample contrast sensitivity data in combination with actual
clinical health classification information of each subject
associated with the contrast sensitivity data.
[0134] Once these HCAPPA coefficients are determined, the HCAPPA
and the coefficients incorporated into the HCAPPA are employed to
process the CST output file 412 of a new subject (viewer) 140 not
having any known associated clinical health classification. The
HCAPPA outputs a set of probabilistic health classifications which
predict what clinical health classification(s) the subject (viewer)
140 would likely be appropriately assigned to, if clinically
evaluated.
[0135] For example, a new subject would have no known prior testing
with respect to Alzheimer's disease (AD), Mild Cognitive Impairment
(MCI) or Cognitive Complaint (CC), but would have associated
contrast sensitivity (CS) test result data. From this CS test data,
a set of probabilistic health classifications can be determined for
that new subject with respect to Normal Health, Alzheimer's
disease, Mild Cognitive Impairment or Cognitive Complaint based
upon a correlation model as described herein.
[0136] Each of the equations (14) through (17) operate as a portion
of the health correlation assessment algorithm (HCAPPA). Equation
(14) determines the probability of appropriately classifying a
subject as having clinical Normal Health (NH) based upon the
contrast sensitivity test result data for that subject. Equation
(15) determines the probability of appropriately classifying a
subject as having clinical Cognitive Complaint(s) based upon the
contrast sensitivity test result data for that subject. Equation
(16) determines the probability of appropriately classifying a
subject as having clinical Mild Cognitive Impairment (MCI) based
upon the contrast sensitivity test result data for that subject.
Equation (17) determines the probability of appropriately
classifying a subject as having clinical Alzheimer's Disease based
upon the contrast sensitivity test result data for that
subject.
[0137] FIG. 11 is a matrix illustrating a relationship between the
results of the predicted and actual health classifications based
upon the multiple CST parameter correlation model of FIG. 10. As
shown, (38 of 83) subjects (i.e. 24+8+6+0=38) are predicted as more
likely of being clinically classified as having normal health than
likely of being clinically classified as within the CC, MCI or
Alzheimer's disease classifications. Also, (9 of 83) subjects are
predicted to be more likely of being clinically classified as
having cognitive complaints (CC) than likely of being clinically
classified as not having cognitive complaints. Also, (31 of 83)
subjects are predicted as more likely of being clinically
classified as having mild cognitive impairment (MCI) than likely of
being clinically classified as not having mild cognitive impairment
(MCI). Also, (5 of 83) subjects are predicted as more likely of
being clinically classified as having Alzheimer's disease than
likely of being clinically classified as not having Alzheimer's
disease.
[0138] From this analysis (24 of 29) subjects that are of actual
(clinical) normal health (NH), are predicted as having normal
health, and (4 of 18) subjects actually having (clinical) cognitive
complaints (CC), are predicted as having cognitive complaints, and
(18 of 27) subjects actually having mild cognitive impairment (MCI)
are predicted as having mild cognitive impairment (MCI) and
finally, only (4 of 9) sample subjects actually having clinical
Alzheimner's Disease (AD) are predicted as having Alzheimer's
Disease (AD).
[0139] The sensitivity of the normal health (NH) prediction is
(24/29)=82.7%, the sensitivity of the cognitive complaint (CC)
prediction is (4/18) 22.2%, the sensitivity of MCI is
(17/27)=66.6%, and the sensitivity of Alzheimer's Disease (AD)
prediction is (4/9)=55.5%. Importantly, the sensitivity of the MCI
and AD classifications combined is 75% and the sensitivity of the
CC, MCI and AD classifications (unhealthy classifications) combined
is 83.3%.
[0140] Importantly, this multi-parameter embodiment provides
substantially more sensitivity with respect to Alzheimer's Disease
(11.1%->55.5%) and more sensitivity to the CC, MCI and AD
classifications (non healthy classifications) combined
(62.9%->83.3%).
[0141] It is desired that this correlation model evolve to achieve
higher sensitivity for CC, MCI and AD so that fewer subjects in
these categories are not recognized by the HCAPPA. This is desired,
even if such improvement in sensitivity comes that the cost of
lower specificity of normal health (NH), given that the cost
penalty for not recognizing normal health is advising and
performing an unnecessary Alzheimer's disease related clinical
test. Conversely, the cost penalty for not recognizing an
Alzheimer's related disease classification can be unnecessary
progression of the Alzheimer's disease without early treatment,
which could be much more costly in the long term.
[0142] The sensitivity of both the single and multi-parameter
algorithms may further improve with access to more sample data. The
sample data includes only (9 of 83) Alzheimer's Disease (AD)
subjects. Access to more clinically indentified AD subjects to
improve the sensitivity and/or specificity either or both of these
algorithms.
[0143] FIG. 12 illustrates systematic interoperation among contrast
sensitivity testing, clinical testing and the health correlation
assessment service 130. As shown, a plurality of contrast
sensitivity test taking subjects 1202-1206 receive contrast
sensitivity testing 410.
[0144] The subject 1202 performs a first contrast sensitivity test
(CST) 1202a which indicates normal health and performs a second CST
1202b (one year later) which indicates normal health, but that is
significantly more likely of being classified as Mild Cognitive
Impairment. After the second CST, subject 1202 elects to receive
clinical testing 1220. Subject 1202 has an option of permitting a
release of the HCS test(s) and the results of the clinical tests,
to the health correlation assessment service (HCAS) 130.
[0145] A subject has an option to perform many test over time.
Algorithms can be designed to quantity a chronological trend to
measure effectiveness of medical treatment or possible progression
of a disease or medical condition for which the CST is
sensitive.
[0146] The subject 1204 performs a first contrast sensitivity test
(CST) 1204a which indicates substantial likelihood of Mild
Cognitive Impairment and a small likelihood of Alzheimer's disease
and elects to receive clinical testing 1220. Clinical tests
indicate Alzheimer's disease and subject 1204 elects to permit
release of the contrast sensitivity tests and the clinical tests to
the HCAS 130.
[0147] The subject 1206 performs a first contrast sensitivity test
(CST) 1206a which indicates small likelihood of Mild Cognitive
Impairment and elects to receive clinical testing 1220. Clinical
tests indicate normal health and the subject 1206 elects to permit
release of the contrast sensitivity tests and the clinical tests to
the HCAS 130.
[0148] The subject 1208 performs a first contrast sensitivity test
(CST) 1208a which indicates small likelihood of Mild Cognitive
Impairment and elects not to receive clinical testing 1220. There
are no clinical tests performed and subject 1208 elects to permit
release of the contrast sensitivity tests to the HCAS 130 in case
further contrast sensitivity testing and/or clinical testing is
performed sometime in the future.
[0149] At the HCAS 130, specialists scrutinize data received for
subjects from the CST 410 and clinical testing 1220 to decide for
inclusion of the data received into the repository of sample data
550 managed within the HCAS 130. As a result, the repository of
sample data 550 of the HCAS 130 and associated HCAP data fit
algorithms (HCAPDFA) and HCAP predictive algorithms (HCAPPA) can be
revised and enhanced over time to better exploit a larger and more
representative set of sample Alzheimer's disease related data.
[0150] This written description uses examples to disclose the
invention, including the best mode, and also to enable any person
skilled in the art to practice the invention, including making and
using any devices or systems and performing any incorporated
methods. The patentable scope of the invention is defined by the
claims, and may include other examples that occur to those skilled
in the art. Such other examples are intended to be within the scope
of the claims if they have structural elements that do not differ
from the literal language of the claims, or if they include
equivalent structural elements with insubstantial differences from
the literal language of the claims.
* * * * *