Method of formalizing temporal aspects in events of an electronic
patient record.
[DESCRIPTION]
FIELD OF THE INVENTION
The present invention relates to a method of formalizing temporal
aspects in events of an electronic health record (EHR) .
BACKGROUND OF THE INVENTION
Clinical decision making is a process that assists health
professionals in decision making tasks such as making a diagnosis of
a patient's health condition.
In a healthcare environment such a decision can be deduced from data
available in a single electronic patient record of a patient. It
can also be based on interpreted information originating from large
sets of electronic patent records stored in different forms and
formats in different data repositories.
In order to adequately serve as a basis for decisions, temporal
aspects of such data from patient records is of ultimate importance.
A s these data are used e.g. for comparison and/or statistics and
deduction of data, it is crucial to have accurate knowledge of the
temporal aspects (date information) of the used data.
However, data in the electronic patient records do not always have
accurate or complete temporal information.
Temporal data may be entirely missing or may be incomplete.
Sometimes explicit temporal information is not given but is
implicitly present and could be deduced from other data available in
the electronic patient record.
Events documented in an electronic patient record may thus have
information on the date of the event in several ways. Figure 1
discloses a form in which data regarding an examination are
recorded. This form is a form used in Agfa Healthcare's Clinical
Information Management System, marketed under the trade name ORBIS.
Some events have a very specific date, e.g. complete data are
available on the date of the examination. The examination in this
example was performed on May 31, 2013. This date is used as
reference for this particular form.
Furthermore, the record comprises very specific data on the date of
death of the patient. The patient in this example died on May 31,
2013.
Other events may have no temporal information explicitly recorded.
For example in the form shown in figure 1 , the patient's weight and
height are recorded. No explicit data are however available on the
date on which these data were gathered. For this event without
explicit date, one can infer that the weight and height are measured
before or on the date of the examination.
In any case, if it is desired to be able to use the data for
clinical decision making, there is a need to have this temporal
information in a format that makes comparison of data from different
sources and / or originally represented in different formats,
possible .
SUMMARY OF THE INVENTION
The above-mentioned advantageous aspects are realised by a method as
set out in the appended independent claim.
Further embodiments of the invention are set out in the dependent
claims .
The present invention relates to extraction of data from at least
one event in at least one electronic patient record from at least
one data repository for the purpose of further processing of these
data, e.g. in a clinical decision making application or statistics.
In the context of the present invention the term ^event' refers to
an item (record, and to the data relating to such an event) which
has been stored in an electronic patient record. Examples of such
events are: an examination, laboratory data ... or general
administrative data of a patient. Other types of data may be
envisaged .
The invention specifically relates to temporal aspects of these
events. A s has been mentioned, such temporal aspects play an
important role when data comparison is concerned or when decisions
are made on the basis of these data.
Temporal data (date information such as date of an examination, date
of a medication etc.) may be provided explicit and complete or
explicit and incomplete, or may even be absent.
The invention provides a method in which the temporal data
information is formalized so that it can be used in further
applications.
Further advantages and embodiments of the present invention will
become apparent from the following description and drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 shows data regarding an event extracted from an electronic
patient record,
Figure 2 schematically shows the data flow,
Figure 3 is a schematic representation of a data warehouse.
DETAILED DESCRIPTION OF THE INVENTION
While the present invention will hereinafter be described in
connection with preferred embodiments thereof, it will be understood
that it is not intended to limit the invention to those embodiments.
For the following explanation reference is made to figure 2 which
illustrates the different steps of the present invention.
According to the present invention in a first step data concerning
an event (s) is retrieved from a data repository. The retrieved data
may comprise explicit temporal information (complete or partial) or
no explicit temporal data of the event.
In one embodiment the data are retrieved directly from one or more
database (s) such as relational data bases.
In another embodiment the data are retrieved from a data warehouse
such as the data warehouse described in co-pending un-published
European patent application 12182778.6 filed September 3 , 2012 and
entitled On demand semantic data warehouse' . Figure 3 is a
schematic representation illustrating the composition of such a data
warehouse. Data concerning stored entities become available at a
SPARQL endpoint of the data warehouse.
A semantic data warehouse as described in this patent application
typically comprises a convergence service for executing queries to
connected data sources, converting data from source to domain
semantics and aggregating converted data. The convergence service
is invoked by an entity graph service. This entity graph service on
demand defines a semantic entity representation, the needed queries
and data sources to be queried. It projects and makes available the
resulting data in said semantic entity representation.
An entity representation is stated in RDF (Resource Description
Framework) .
The entity representation is provided by means of a named entity
graph and can be denoted by an URL.
A convergence service is a software system designed to support
interoperable interaction over the world wide web.
The convergence service is invoked by an entity graph service.
The convergence service performs a conversion of data expressed with
data definition ontologies (DDO) as available in the data sources to
data expressed with the domain ontologies (DO) as used by the entity
graphs and aggregates the resulting data.
The conversion service uses formal declarative rules for the
conversion process.
An entity graph can be used as a data graph in the entity graph
SPARQL endpoint to provide answers to queries on the named entity
graph .
In order to be able to provide a user with a unified view of data
from different data sources with each having different local
semantics, an entity graph service is used that on demand produces
an entity graph by specifying which data needs to be retrieved from
identified data sources, invoking the convergence service to
retrieve the data from the different data sources and convert the
data from the local semantics to the domain ontology, and projecting
the result to the model of the defined entity representation.
Entity graphs are constructed on demand based on the use case.
These entity graphs are specific configurable entity representations
with unification of data from different data sources.
The entity graph SPARQL endpoint may provide caching functionality
to cache the generation of the entity representation.
The formal representation of an entity graph can be retrieved by
resolving the URL of the named entity graph.
A specific ETL (Extract-Transform-Load) process can be defined for
each of the targeted data consumer data schemas and the configured
entity graphs.
The data warehouse exposes on demand domain entity graphs.
The data warehouse can be scaled at development time by allowing
development of additional independent plug-ins to expose new entity
graphs. Plug-ins for existing entity graphs do not need to be
adapted.
If the temporal information extracted or retrieved from the data
repository is not represented in a semantic representation, a step 2
is required to convert it into a semantic representation such as an
RDF representation. Alternative applicable semantic representations
are a conceptual graph representation or a topic maps .
E.g. year 2000' entered in a partial date is retrieved as an xsd
long integer ("2000" xsd: long) .
Preferably the formalized data are kept as close as possible to the
original data, to achieve this goal preferably minimal data
interpretation is performed.
Data are preferably retrieved on demand, just in time for use in an
application such as clinical study.
Next, conversion rules are applied to the extracted semantic data.
Conversion rules are applied to convert the semantic representation
of the temporal information into a formal representation. Thus time
entities are generated which are represented with formal ontologies.
E.g. A partial date is converted as an interval.
This conversion is preferably performed upon request.
The converted results could use existing ontologies to represent
intervals and time points.
For example W3C Time ontology can be used as target ontology. The
W3C Time ontology captures the widely used Allen Interval temporal
concepts and provides an instant class to formalize timepoints.
If temporal information is not explicitly present in the data of the
event, an additional step, as described furtheron, is required.
If the event does no comprise explicit temporal information,
temporal information needs to be inferred from other temporal
information available in the data record of the event. Temporal
relationships, which are implicitly embedded in the source data, are
explicitly expressed at this stage.
When the temporal information in all records required for a certain
application, such as for a clinical decision making process, are
converted into a formal representation, the converted data can be
integrated with data from other sources to carry out temporal
reasoning, resulting in knowledge on a relation between the events
to which these temporal data pertain.
Temporal reasoning comprises first order reasoning such as Allen
calculus, which defines possible relations between time intervals
and provides a composition table that can be used as a basis for
reasoning about temporal descriptions of events.
Temporal relationships such as before, during, overlaps, after, etc.
are interpreted.
For example: An event may specify that medication was administered
to a patient in January 2010. Suppose that the event in the
electronic patient record further mentions a health problem
(disease) in February 2010. The temporal information January 2010
as well as February 2010 is incomplete (partial) since only very
rough date information is available.
The method of this invention is then applied to deduce explicit,
complete temporal information with which further reasoning can be
performed.
A first period from Jan. 1 , 2010 to Jan 31, 2010 is known. In this
period medication was administered.
In a second period from February 1 , 2010 to February 28, 2010 the
disease has occurred.
Temporal reasoning will result in the fact that the disease happened
after the administration of the medication.
The method of the present invention is advantageous in that
The method enables relating data records that have temporal
data in incomplete form or that have the temporal data in
different formats. It even provides that data can be used that
do not have explicit temporal information in the record. The
method meets the gap between time related data stored in an
electronic health record and temporal data required for
clinical research.
It allows temporal reasoning between events.
It furthermore allows a very flexible use of data. Data can be
represented by an ontology that best fits a target application.
The reliability of the conversion can be checked by means of a
third party proof checker.
The present invention can be implemented as a computer program
product adapted to carry out the steps set out in the description.
The computer executable program code adapted to carry out the steps
set out in the description can be stored on a computer readable
medium.
[CLAIMS ]
1 . A method for formalizing temporal data in a structured
electronic patient record based on the semantics of the record
comprising the steps of
- (1) extracting data regarding at least one event from a data
repository,
- (2) if the extracted data is not in a semantic representation,
performing a syntactic transformation of extracted data from their
original representation as extracted from said data repository into
a first semantic representation,
- (3) converting extracted and transformed data into a
representation with explicit semantics regarding temporal
information .
- (4) if the extracted data comprises no explicit temporal data,
deducing temporal data from other extracted data pertaining to said
event and attaching deduced temporal data to the event data.
2 . A method according to claim 1 wherein said first semantic
representation is an RDF representation.
3 . A method according to claim 1 wherein said data repository is a
data warehouse.
4 . A method according to claim 1 wherein said conversion is based on
the application of N3 rules.
5 . A method according to claim 1 wherein said explicit semantics
depend on an ontology dictated by a target application in which said
temporal data are used.
. A method according to claim 1 wherein said partial data is
converted to a time period represented with said ontology.
7 . A method according to claim 1 wherein said steps are performed on
request .
8 . A method according to claim 1 wherein interpretation of said
other extracted data in order to deduce temporal data of an event is
minimal so as to remain close to the original available information
of an event.
9 . A computer program product adapted to carry out the method of
claim 1 when run on a computer.
10. A computer readable medium comprising computer executable
program code adapted to carry out the steps of claim 1 .