U.S. patent application number 15/696718 was filed with the patent office on 2017-12-21 for automated assertion reuse for improved record linkage in distributed & autonomous healthcare environments with heterogeneous trust models.
The applicant listed for this patent is KONINKLIJKE PHILIPS N. V.. Invention is credited to RIchard Vdovjak.
Application Number | 20170364639 15/696718 |
Document ID | / |
Family ID | 42029872 |
Filed Date | 2017-12-21 |
United States Patent
Application |
20170364639 |
Kind Code |
A1 |
Vdovjak; RIchard |
December 21, 2017 |
AUTOMATED ASSERTION REUSE FOR IMPROVED RECORD LINKAGE IN
DISTRIBUTED & AUTONOMOUS HEALTHCARE ENVIRONMENTS WITH
HETEROGENEOUS TRUST MODELS
Abstract
An assertion acceptance value matrix (300) indicates the
reliability of assertions, particularly assertions or decisions
whether records match or do not match, made by other medical
institutions in a federation of medical institutions with different
patient record systems and some common patients. Records from
different institutions with a high likelihood of matching or not
matching are automatically matched or not matched. Those that are
ambiguous are manually reviewed. The assertion acceptance value
matrix is used to reduce or expedite the manual review.
Inventors: |
Vdovjak; RIchard;
(Eindhoven, NL) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
KONINKLIJKE PHILIPS N. V. |
Eindhoven |
|
NL |
|
|
Family ID: |
42029872 |
Appl. No.: |
15/696718 |
Filed: |
September 6, 2017 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
13132949 |
Jun 6, 2011 |
|
|
|
PCT/IB09/55183 |
Nov 19, 2009 |
|
|
|
15696718 |
|
|
|
|
61121985 |
Dec 12, 2008 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06Q 50/24 20130101;
G06Q 50/22 20130101; G06F 19/00 20130101; G16H 10/60 20180101 |
International
Class: |
G06F 19/00 20110101
G06F019/00; G06Q 50/24 20120101 G06Q050/24; G06Q 50/22 20120101
G06Q050/22 |
Claims
1. An apparatus for generating reusable comparisons, comprising: an
input which receives a record supplied by an outside party, the
records being patient records of a federation of medical
institutions, the institutions having different medical records
systems, some patients having unmatched patient records in the
medical systems of a plurality of the medical institutions; a
database which stores a plurality of records; at least one computer
processor which: compares the received record with each record
retrieved from the database in order to generate a likelihood ratio
based on a probability that the compared records match; compares
the likelihood ratio to both an accept threshold and a reject
threshold, assigns a record that is between the accept threshold
and the reject threshold to an exception list for a manual
determination whether the records on the exception list match;
receives assertions made by at least one outside party wherein the
assertion comprises whether the records on the exception list were
accepted or rejected as matching; compares the records on the
exception list that were at least one of accepted and rejected both
by the party receiving the record and each outside party in order
to calculate an assertion acceptance value for each outside party,
creating a matrix of assertion acceptance values including at least
one axis representing the party receiving assertions, another axis
representing the party issuing assertions, one cell for each
intersection between the receiving axis and the issuing axis,
wherein each cell containing at least one acceptance value; and
records the assertion acceptance values in the matrix.
2. The apparatus according to claim 1, wherein the matrix is
accessible by each medical institution in the federation of medical
institutions.
3. The apparatus according to claim 1, wherein each cell in the
matrix contains separate assertion acceptance values for a positive
assertion and for a negative assertion.
4. The apparatus of claim 1, where the probability value is within
the in the range from 0 to 1 inclusive, where 0 represents no
acceptance and 1 means the received assertion is the ground
truth.
5. The apparatus according to claim 1, wherein the at least one
computer processor is further programmed to calculate each
assertion acceptance value as a weighted ratio of prior common
assertions determined by the receiving party and each other
party.
6. The method according to claim 1, wherein each cell in the matrix
of assertion acceptance values may contain separate assertion
acceptance values for a positive assertion and one for a negative
assertion.
7. The method according to claim 6, wherein the received record and
an accepted record of one of the medical institutions in the
federation of plurality of medical institutions are assigned a
unique, federation wide medical record number.
8. The method according to claim 1, wherein the acceptance value
matrix is at least one of: used in the likelihood ratio generating
step to increase or decrease the probability in accordance with the
acceptance values; or used in the step of determining whether
records on the exception list should be accepted or rejected.
9. A method of matching patient records for a plurality of medical
institutions with different medical records systems, with some
patients having records in a plurality of the medical records
systems, the method comprising: comparing a selected patient
medical record in a medical records system with a plurality of
patient records from at least one other medical record system with
at least one computer processor; generating likelihood ratios
indicative of a probability that the selected patient medical
record matches each of the compared records with the at least one
computer processor; automatically (1) matching the selected patient
medical record with one of the compared records if the likelihood
ratio exceeds an accept threshold, (2) not matching the selected
patient medical record with compared records that meet a reject
threshold, and (3) if the selected patient medical record and one
or more compared records do not meet either the accept or the
reject threshold, assigning the selected patient medical record to
an exception list for manual review with the at least one computer
processor; receiving an indication whether the selected patient
medical record was matched to one of the compared records by
another of the medical institutions with an input; performing at
least one of the manual review and the generating step in
accordance with an assertion acceptance value matrix indicative of
a reliability of matches made by other medical institutions with
the at least one computer processor; and utilizing the acceptance
value matrix to automatically determine whether accepted or reject
records on the exception list should be accepted or rejected.
10. The method according to claim 9, where the matrix of assertion
acceptance values is created by: placing the name of at least one
other medical institutions which receives assertions on one axis
with the at least one computer processor; placing the name of at
least one other medical institutions which issues the assertions on
a second axis with the at least one computer processor; and
creating one cell for each intersection between the receiving axis
and the issuing axis, wherein each cell contains at least one
assertion acceptance value with the at least one computer
processor.
11. The method according to claim 10, wherein each assertion
acceptance value is calculated as a weighted ratio of prior common
assertions determined by the receiving party and each other
party.
12. The method according to claim 10, wherein each cell in the
matrix of assertion acceptance values may contain separate
assertion acceptance values for a positive assertion and one for a
negative assertion.
13. The method according to claim 9, wherein the acceptance value
matrix is utilized to adjust the accept and reject thresholds.
14. The method of claim 15, where the probability value is within
the in the range from 0 to 1 inclusive, where 0 represents no
acceptance and 1 means the received assertion is the ground
truth.
15. The method according to claim 9, wherein further comprising:
assigning a unique, federation wide medical record number to the
received record and an accepted record of each medical institution
in the federation of medical institutions.
16. The method according to claim 9, further including using the
acceptance value matrix in at least one of: the likelihood ratio
generating step to increase or decrease the probability in
accordance with the acceptance values; or the step of determining
whether records on the exception list should be accepted or
rejected.
17. A non-transitory computer readable medium storing instructions
executable by at least one electronic processor to perform a method
of matching patient records for a plurality of medical institutions
with different medical records systems, with some patients having
records in a plurality of the medical records systems, the method
comprising: comparing a selected patient medical record in a
medical records system with a plurality of patient records from at
least one other medical record system with at least one computer
processor; generating likelihood ratios indicative of a probability
that the selected patient medical record matches each of the
compared records with the at least one computer processor;
automatically (1) matching the selected patient medical record with
one of the compared records if the likelihood ratio exceeds an
accept threshold, (2) not matching the selected patient medical
record with compared records that meet a reject threshold, and (3)
if the selected patient medical record and one or more compared
records do not meet either the accept or the reject threshold,
assigning the selected patient medical record to an exception list
for manual review with the at least one computer processor;
receiving an indication whether the selected patient medical record
was matched to one of the compared records by another of the
medical institutions with an input; performing at least one of the
manual review and the generating step in accordance with an
assertion acceptance value matrix indicative of a reliability of
matches made by other medical institutions with the at least one
computer processor; and utilizing the acceptance value matrix to
automatically determine whether accepted or reject records on the
exception list should be accepted or rejected.
18. The non-transitory computer readable medium according to claim
17, where the matrix of assertion acceptance values is created by:
placing the name of at least one other medical institutions which
receives assertions on one axis with the at least one computer
processor; placing the name of at least one other medical
institutions which issues the assertions on a second axis with the
at least one computer processor; and creating one cell for each
intersection between the receiving axis and the issuing axis,
wherein each cell contains at least one assertion acceptance value
with the at least one computer processor.
19. The non-transitory computer readable medium according to claim
18, wherein each assertion acceptance value is calculated as a
weighted ratio of prior common assertions determined by the
receiving party and each other party.
20. The non-transitory computer readable medium according to claim
17 wherein the acceptance value matrix is utilized to adjust the
accept and reject thresholds.
Description
[0001] This application is a continuation of, and claims the
priority date of, U.S. application Ser. No. 13/132,949, filed Jun.
6, 2011, which a national filing of PCT Application Serial No.
PCT/IB2009/055183, filed Nov. 19, 2009, published as WO
2010/067229A1 on Jun. 17, 2010, which claims the benefit of U.S.
Provisional Application Ser. No. 61/121,985, filed Dec. 12, 2008,
each of which is incorporated herein by reference.
[0002] The present application relates to the art of data
continuity. It finds particular application to the management of
patient records in a medical environment. However, it will also
find use in other types of applications in which data continuity is
of interest.
[0003] It is common for patients to receive care from multiple
healthcare providers which are geographically dispersed at multiple
sites. Using multiple healthcare providers results in a patient
receiving multiple patient identifiers, each patient identifier
local to a specific healthcare provider. Patient data such as
medical images and other relevant medical information is spread
across multiple healthcare provider sites. In order for a
healthcare provider to retrieve patient data records stored among
multiple healthcare provider sites, it is necessary to reconcile
the multiple patient identifiers of the corresponding healthcare
providers and to link the multiple patient identifiers together.
There is currently a need for a system that enables medical records
to follow the patient as the patient moves between multiple
healthcare providers which are dispersed geographically at multiple
sites.
[0004] The present application provides an improved method and
apparatus which overcomes the above-referenced problems and
others.
[0005] In accordance with one aspect, a method is presented of
reusing comparisons, comprising receiving a record supplied by an
outside party, comparing the received record with a plurality of
records currently maintained by a receiving party in order to
determine if any two compared records correspond to a same
customer. This is used in generating a likelihood ratio based on a
probability that the compared records match; and this ration is
used in comparing the likelihood ratio to an accept threshold and
to a reject threshold, the accept threshold being different from
the reject threshold. Then; assigning a record to an exception list
when the record's likelihood ratio a value is both less than the
accept threshold and is also greater than the reject threshold to
an exception list. Then determining whether the records on the
exception list should be accepted as a match or rejected as not
matching and recording the determination; receiving assertions made
by outside parties whether records on the exception list were
accepted or rejected as matching. Then; comparing the records on
the exception list that were at least one of accepted and rejected
by both the party receiving the record and each outside party in
order to calculate an assertion acceptance value for each outside
party. Finally, recording the assertion acceptance values in a
matrix format is performed.
[0006] In accordance with another aspect, an apparatus is presented
for generating reusable comparisons, comprising of an input which
receives a record supplied by an outside party; a database which
stores a plurality of records; at least one processor which
compares the received record with each record retrieved from the
database in order to generate a likelihood ratio based on a
probability that the compared records match. Then the apparatus
compares the likelihood ratio to both an accept threshold and a
reject threshold, assigns a record that is between the accept
threshold and the reject threshold to an exception list for a
manual determination whether the records on the exception list
match. The at least one processor further receives assertions made
by at least one outside party wherein the assertion comprises
whether the records on the exception list were accepted or rejected
as matching; compares the records on the exception list that were
at least one of accepted and rejected both by the party receiving
the record and each outside party in order to calculate an
assertion acceptance value for each outside party. Finally, the
apparatus records the assertion acceptance values in a matrix
format.
[0007] In accordance with a further aspect, a method is proposed of
matching patient records for a plurality of medical institutions
with different medical records systems, with some patients having
records in a plurality of the medical records systems. The method
comprises comparing a selected patient medical record in a medical
records system with a plurality of patient records from at least
one other medical record system, generating likelihood ratios
indicative of a probability that the selected patient medical
record matches each of the compared records. The method engages in
automatically matching the selected patient medical record with one
of the compared records if the likelihood ratio exceeds an accept
threshold, not matching the selected patient medical record with
compared records that meet a reject threshold, and if the selected
patient medical record and one or more compared records do not meet
either the accept or the reject threshold, assigning the selected
patient medical record to an exception list for manual review. Then
receiving an indication whether the selected patient medical record
was matched to one of the compared records by another of the
medical institutions; and finally performing at least one of the
manual review and the generating step in accordance with an
assertion acceptance value matrix indicative of a reliability of
matches made by other medical institutions.
[0008] An advantage resides in the ability to reduce the need to
continually evaluate the same record a plurality of times.
[0009] Another advantage resides in a greater assurance of
accuracy.
[0010] Another advantage resides in identifying inconsistent
results and potential errors.
[0011] An advantage resides in the fact that other healthcare
providers can see how a specific healthcare provider evaluated the
medical records of a particular patient at a previous healthcare
provider and use this evaluation to make their own assessment of
that specific patient's medical records.
[0012] The invention may take form in various components and
arrangements of components, and in various steps and arrangements
of steps. The drawings are only for purposes of illustrating the
preferred embodiments and are not to be construed as limiting the
invention.
[0013] FIG. 1 illustrates two thresholds for distinguishing three
outcomes with respect to record matching.
[0014] FIG. 2 illustrates the assertion acceptance in distributed
heterogeneous environments with autonomous sites.
[0015] FIG. 3 illustrates an assertion acceptance matrix for a
federation with four participating institutions. Note that the
matrix data structure is just one of many possible way to store
this information other approaches (e.g. a list of lists, set of
dictionaries etc.) may be equally applicable.
[0016] FIG. 4 illustrates a flow chart of the method claim.
[0017] FIG. 5 illustrates how the information flows through the
apparatus.
[0018] With reference to FIG. 1, a probabilistic algorithm, which
compares a fixed record with a number of candidates for a match,
computes for each candidate a likelihood ratio or weighted score
that is individually compared with reject 110 or accept 120
thresholds. The reject threshold 110 is used as a basis for a
comparison used to decide whether the record falls below the reject
threshold 110. If the likelihood ratio of a record is less than the
reject threshold 110 the record is rejected 130 as not being a
match 150 and the record is not linked. The accept threshold 120 is
used as a basis for a comparison used to decide which records
exceed the accept threshold. If the likelihood ratio of a record is
greater than the accept threshold 120, then the record 140 is
accepted as a match 170 and the record is linked. If the record is
greater than the reject threshold 110 and also less than the accept
threshold 120, then the computed likelihood falls between the
accept and the reject thresholds. Here, the record might be either
rejected 180 or accepted 190. The decision is not made
automatically, but rather is flagged for a manual review 160 by
qualified personnel. The manual review is made to determine if the
record should be properly rejected 180 or accepted 190 as a match.
The manual review 160 of uncertain matches is of crucial importance
in order to minimize linkage errors, as such errors can have
far-reaching consequences ultimately endangering patient
health.
[0019] As medical providers or health systems associate or
consolidate, it becomes advantageous to share patient records. The
same patient may exist in multiple provider databases with
different medical record numbers (MRN). The flow of information
into and out of the patient record is typically channeled through a
Master Patient Index (MPI) that associates a unique medical record
number (MRN) in the provider's numbering system to each patient
entity when a unit record exists. To obtain an enterprise-wide view
on the patients across distributed data sources, an enterprise
master patient index (EMPI) is implemented with one common,
enterprise wide unique MRN for each patient such that each patient
is known by one and only one MRN, independent of the specific
medical facility such that the same one patient will have the same
unique MRN at each and every medical provider participating in the
present application disclosed system. The EMPI is developed through
integration of the individual MPIs of the entities which together
make up the enterprise. Generally, the integration is achieved by
comparing demographic attributes, such as first and last name,
gender, date of birth, address, and other demographic data to
create the EMPI as a form of an enterprise-level patient identifier
where one single unique MRN would identify the same patient at
every medical facility using the present system. Such an enterprise
level patient identifier is rarely based on a single identifier
shared across the different organizations in the enterprise.
[0020] Probabilistic algorithms can be used to compare a fixed
record with a number of candidates for a match, computing for each
candidate a likelihood ratio or weighted score that is compared to
the chosen accept and reject thresholds as explained above in
connection with FIG. 1. This method is used to decide whether or
not to link the records to a common identifier or enterprise-wide
MRN. When the decision cannot be taken automatically such as when
the computed likelihood falls below the two thresholds, qualified
personnel review or flag the potential matches before they are
accepted and also review the mismatches before they are rejected.
The manual review of uncertain matches is very important in order
to minimize linkage errors, but at the same time it is
time-consuming and expensive.
[0021] With reference to FIG. 2, an assertion-based record linkage
which provides a link between records at two different medical
providers that have a likelihood of pertaining to the same patient,
allows for keeping track of assertions throughout the group of
institutions also referred to as a federation or enterprise,
treating all of the assertions as local ground truth for the
institutions that issued them but not necessarily for other
institutions. An assertion is a belief that two medical records do
o do not refer to the same patient. The customer is assigned a
unique record number to a customer's records. The customer's
demographic data is compared with the demographics data of each
record in a collection of records that are potentially belonging to
the same customer, to derive a likelihood ratio for each compared
record. Each likelihood ratio is compared to a defined accept
threshold and to a defined reject threshold. During the manual
review of the records falling between the two thresholds, an
assertion is made about the pair of records to be matched, based on
the likelihood ratio comparison, stating whether it is believed
that the two records belong to the same patient or notThe local
ground truth approach requires all participating institutions to
make their own manual review of a record in order to resolve the
exception list of records that needed to be manually reviewed. When
the trust level within the group of institutions or federations is
low, each institution does not take assertions of other
institutions as granted. The worst case scenario in which no
institution trusts any other institution is the most time-consuming
to solve.
[0022] The trust level among some of the participating institutions
can be large. For instance, imagine a large well-established
healthcare provider 200 which may have acquired some smaller
institutions with their own patient identification schemes. Let us
assume that hospital A, hospital B, and hospital C, participate
together in one federation which is expanded to add another
institution such as D. Over time, hospitals A, B, and C have
established common procedure and come to have complete confidence
in each other. In this example, when hospital A makes an assertion
210 about two participating identifiers being the same, this
assertion would be accepted as local truth not only as hospital A,
but also hospital B, and hospital C. In such situation, there is no
need for a manual review of hospital B and hospital C. Hospital D
on the other hand may not have developed such a high level of
confidence and is able to locally override hospital A's assertion.
At the same time, the assertion produced by hospital A could play a
role in the automated matching process that determines the cases
for manual review of hospital D. Of course the influence of
hospital A's assertion with respect to hospital D's matching will
be larger or smaller depending on the level of trust hospital D has
toward hospital A. Assertions 240, 250, and 270 made by hospitals
B, C, and D, respectively, are treated analogously.
[0023] The present application proposes to describe a means for
explicitly quantifying a mutual trust pair wise among all
participating institutions within the federation of institutions,
and to use this data to maximize re-use of available assertions in
the manual review phase, thus minimizing the labor intensive task
of performing the assertion each time determination process the
patient record is accessed, while allowing for a heterogeneous
trust model within the federation. To this end, a federation trust
matrix is created and applied to automated assertion re-use, where
the available assertions become a part of the probabilistic formula
that computes the thresholds. This approach allows for a smooth and
efficient assertion handling facilitating the re-use of assertions
in the automated patient identification process within
collaborating heterogeneous healthcare environments.
[0024] With reference to FIG. 3, to allow for assertion re-use in
the context of heterogeneous healthcare multi-enterprise patient
matching, we propose the use of a so-called acceptance matrix 300,
which explicates the level of assertion acceptance among the
participating institutions. The rows 330 in the exemplary matrix
300 represent the institutions that receive the assertions; the
columns 310 represent the institutions that issue the assertions.
The values range from 0 to 1 (e.g., 0% to 100%) and indicate the
percentage or weight which the receiving institution assigns to the
assertion of an institution, where 0 means no acceptance 350, and 1
means treating the incoming assertion as ground truth 370.
[0025] In a variation the assertion matrix 300, the acceptance
percentage comprising the weight of positive and negative
assertions, can be different with respect to assertions from one
issuing institution about a patient ID match and mismatch
respectively; each cell in the matrix variation would contain two
acceptance values, one for the positive assertion and one for the
negative assertion. Here, a cell would contain two values, one
value being the acceptance rate as a probability of a record being
accepted with 0 being no acceptance or rejection and 1 being
certain acceptance; while the second value would be a probability
of being rejected, with 0 being no acceptance and a 1 being a
certain rejection. These two values would reside in the same cell
of a matrix, either side by side, above and below, one on top of
another, diagonally adjacent from each other, or any number of
combinations or in parallel matrices, or can also be stored in
another appropriate type of data structure e.g. list of lists
etc.
[0026] It is noted that the matrix 300 has ones in diagonal matrix
cells, indicating that institutions take their own assertion as
ground facts 380. However, each institution may recognize the
possibility of error and give itself a high percent or weight,
e.g., 0.95, but may not accept it as a ground fact. Also note that
the matrix 300, or variations thereof, do not have to be symmetric
as the trust degree is not necessarily symmetric either, such as
when hospital D accepts an assertion from hospital A with the
degree of 0.5 (360), but hospital A does not take hospital D's
assertion into account at all (no acceptance 350). The value from
the assertion matrix 300 can be embedded into the automated
matching algorithm that is adopted, by simply increasing the
acceptance changes with the indicated percentage, or embedding this
percentage in a chosen probabilistic formula, next to other usual
matching criteria such as a match based on a last name, date of
birth, and the like. So for instance in a case of hospital D, it
will increase the likelihood of a patient ID match by 50% if
hospital D receives an assertion from hospital A about a particular
case. After hospital D receives hospital A's assertion, it then
proceeds with the matching algorithm computing the overall matching
score taking into account all available demographic data plus the
increased likelihood based on hospital A's assertion. Based on the
score of the chosen threshold, the patient IDs are then
automatically linked or proclaimed as different or put for a manual
review.
[0027] The problem of patient identity in a federated environment
in the absence of a global common identifier is one of the key
issues that need to be solved as a prerequisite to being able to
build and deploy a Federated Picture Archiving and Communication
System (PACS) solution. The present application addresses the
manual review phase of the matching process in the context of
autonomous environments.
[0028] With reference to FIG. 4, a flow chart of the method steps
is presented. First a record supplied by an outside party is
received 410. This record is then compared with a plurality of
records 420 currently maintained by the receiving party in order to
determine if any two compared records correspond to the same
customer. From this comparison, a likelihood ratio is generated 430
based on the probability that the compared records match. This
likelihood ratio is then compared 440 with both an acceptable
threshold and a reject threshold. Any record with a likelihood
ratio that is both less than the accept ratio and greater than the
reject ratio 450 is placed on an exception list for manual review.
The records on the exception list are individually evaluated
manually 460 in order to determine if these records should be
accepted or rejected and this determination is recorded. Then the
total number of records evaluated, accepted and rejected is
segmented by an outside party in order to calculate an acceptance
ratio for each individual outside party 470. This acceptance ratio
data is then placed 48 in the matrix 300. In this manner, the
weight given to decisions made by other institutions can change
with experience.
[0029] With reference to FIG. 5, the flow of information through an
apparatus which may be programmed to perform the present
application such as but not limited to a computer. is illustrated
500. Data is entered via an input terminal 510 and formatted 515 in
a common format 500 for storage 540 in a database 530 and for
comparison. The input data is exerpted in any predefined form and
used to create a unique identifier 520. This input identifier is
comprised of demographic data such as, but not limited to name 542,
age 544, and gender, lifestyle, telephone number and address 546.
The input data is used to derive a unique identifier which is used
as a pointer to compare 525 with other data. The database records
are retrieved 535 and compared (one record and one comparison at a
time) 550 until each record in the database 530 has been compared.
The database contains records previously entered and is comprised
of similar types of demographic data such as, but not limited to
name 542, age 544, and gender, lifestyle, telephone number, and
address 546.
[0030] The format of the input data must be the same as the format
of the data stored in and retrieved from the database for an
accurate comparison to occur. The format depicted in FIG. 5 where
age, one letter for gender, one letter for lifestyle, and several
digits for address 520 is not the only format that can be used, but
it must be the same format as the data in the database 535 for the
matching to occur.
[0031] After each record comparison is performed, the results are
processed 548 and a likelihood ratio 554 of a match existing
between the input record 510 and each database record 552 is
calculated for each record to which the input record 510 was
compared against. In one embodiment, a memory 530 contains or
references a correspondence table of patient records that have been
manually accepted by other institutions as matching or manually
rejected as not matching and which institution (5). The memory also
contains the confidence matrix 300. If a hospital with a zero
confidence value in the matrix has made a manual match or
rejection, the prior match or rejection is given zero weight, i.e.,
ignored. If a hospital with a confidence value of 1 has manually
matched the input record to another record, such prior match causes
a likelihood ratio of 100% to be assigned. Confidence values
between 0 and 1 cause the likelihood ratio to be boosted
accordingly. A prior manual rejection by another institution causes
the likelihood ratio to be downgraded analogously. This likelihood
ratio can also be subsequently adjusted by a chosen probabilistic
formula and then compared 556 with an accept threshold 560 and a
predefined reject threshold 562. Matches with a likelihood ratio
greater than the accept threshold are asserted to be a match 564,
and matches with a likelihood ratio below the reject threshold are
asserted to be rejected 568. Records that have a likelihood ratio
less than the accept threshold 560 and higher than the reject ratio
562 are flagged for manual review 566.
[0032] After a manual review is performed, the result is input 570.
Then each of the assertions for each pair of comparisons is broken
down by receiving institution 575 and averaged to calculate a
percentage of input records received 586, which is placed in the
matrix 300. The matrix contains the names of the institution
submitting, supplying, or transmitting a record on one axis and the
institution receiving the record on the other axis. The horizontal
axis 582 may be for the sending or receiving institution while the
vertical axis 584 may receive either the sending or receiving
institution.
[0033] The matrix is then recorded 588 in a database 530 and may be
shared among other institutions who are either members or are not
members of the federation of institutions via a network such as but
not limited to the Internet 590 by way of an Internet connection
590.
[0034] For example, the entered record is for Joe, a 55 year old
male urban dweller. A comparison with a record in the database of
Adam, a 14 year old male urban dweller would produce a low 19%
likelihood ratio of a match due to the great age and address
disparity between the two compared records. If the threshold ratio
of rejection were 20%, then this record with a 19% ratio would fall
below the 20% rejection threshold and would be rejected. These two
compared records are probably not for the same person.
[0035] Another comparison, this time of the entered record of Joe
the 55 year old male urban dweller with the database record of
another different Joe who is 59 years old, a male urban dweller
would produce a higher likelihood ratio of 91% because these two
records are a much closer match in terms of age and address. If the
threshold ratio for accept were 90%, then this record with an
acceptance ratio of 91% would be above the 90% acceptance ratio and
would be accepted as a match. These two compared records are likely
for the same person.
[0036] A comparison with a record for Joan, a 55 year old female
rural dweller would produce a likelihood ratio of 72%. As this 72%
is above the 20% reject ratio and also below the 90% accept ratio,
this record would be placed on an exception list and flagged for
manual review. Only a manual review could determine whether these
records are for the same person.
[0037] Each demographic factor can be, but is not necessarily,
equally weighted. For example, an individual's address can change a
great deal in a short period of time. Therefore this demographic
might be weighted to be of less importance than other demographics.
A demographic that rarely or never changes, such as gender or race,
might be given greater weight because this demographic may be more
reliable as an indicator of a specific person. This could explain
why the likelihood match between Joe 55 M U 123 Oak and Joe 59 M U
998 Balsa is at 91% despite the great difference in addresses
between these two individuals. Here, the similarity in age, gender,
and lifestyle is weighted more heavily than address is
weighted.
[0038] In another embodiment, prior match and rejection
determinations and the institution making such determination is
provided to the manual reviewer. The manual reviewer is also
provided with the institution confidence matrix 300. The manual
review weights the prior decisions in accordance with the
confidence value accorded to the other hospital.
[0039] In another embodiment, the prior match/reject decisions and
the confidence matrix 300 can be used to adjust the accept or
reject thresholds.
[0040] When two institutions make contrary decisions on whether two
records match, a conflict notice is sent to both institutions. The
two records can be automatically sent to the manual reviewers at
both institutions for reconsideration.
[0041] The above-described process is performed on one or more
computers or computer systems with one or more computer software
programs. Computer programs for performing the steps can be stored
on a tangible computer readable medium, such as a disc, computer
memory, or the like.
[0042] The present disclosure has been described with reference to
the preferred embodiments. Modifications and alterations may occur
to others upon reading and understanding the preceding detailed
description. It is intended that the invention be constructed as
including all such modifications and alterations insofar as they
come within the scope of the appended claims or the equivalents
thereof.
* * * * *