SanteMPI Matches
SanteMPI provides a robust Matching Enginewhich can be used to determine whether new registrations or existing registrations are duplicate / identical patients.
The matching dashboard is used to show the registered matches (i.e. those which have been detected and flagged) within the SanteMPI linkage tables.
Nomenclature
The nomenclature used in SanteDB is described in detail in Master Data Storage, when describing the patients in the user interface documentation the data will be referred to as:
LOCAL_A
- The source record submitted from a third party system, representing the$input
recordMASTER_A
- The master/golden record established by SanteMPI's master data management layer forLOCAL_A
(note there may be multiple locals attached to this record)LOCAL_B
- The source record submitted from a third party system, representing the$block
recordMASTER_B
- The master/golden record established by SanteMPI's master data management layer forLOCAL_B
(note: there may be multiple locals attached to this record)
Patient Matching
The first portion of the match dashboard shows the duplicate detection panel. This panel allows data administrators to quickly review duplicates which have been detected by the MPI software.
Candidate Report - Downloads the
Detected Duplicates
table in excel formatLinkage Report - Downloads an excel report with the entirety of the SanteMPI database linkages provided.
Candidate Patient A/B - The pair of patient records which were detected as a duplicate.
Patient A - The inbound or
$input or LOCAL_A
patient which was being registeredPatient B - The duplicate or
$block or MASTER_B
patient which was scored
Normalized Score - The confidence (
0.0 .. 1.0
) that the patient represented in Patient A is the same as Patient B.View Match Detail - Opens the detail screen for the match (see: )
Quick Resolve - Resolves the match pair
Quick Ignore - Ignores the match pair
Quick Resolve
The quick resolution option allows the data administrator to quickly instruct the SanteMPI MDM layer that Patient A has been verified as Patient B. This action:
Detaches
LOCAL_A
fromMASTER_A
(i.e. detaches the patient in Patient A column from its current master / golden record)Attaches
LOCAL_A
toMASTER_B
(i.e. attaches the Patient A column to the Patient B master / golden record)If no further local or source records exist for
MASTER_A
thenMASTER_A
is obsoleted.
The SanteMPI MDM user interfaces do not MERGE records, they change linkages. In SanteDB a MERGE is a different operation which is performed only by source systems when they send an appropriate ADT^A40
or PMIR Merge
Quick Ignore
The quick ignore option allows the data administrator to quickly instruct SanteMPI MDM layer that Patient A has been verified NOT to be Patient B. This action:
Removes the current candidate record between
LOCAL_A
andMASTER_B
Adds a new
Ignore
instruction which instructs all future detections betweenLOCAL_A
andMASTER_B
to be ignored.
View Match Detail
Clicking on the View
option, will display a more detailed analysis of the match pair. This view assists data administrators in determining the validity of the match, such that the match can be verified or ignored.
Match Summary Information Tab - Shows a side-by-side comparison of basic attributes on both the A and B records.
Either patient record can be opened using the
View
option.
A/B Comparison - Shows the A and B comparison of all demographics for the patient match pair.
Scoring - Shows a report of the match score between the two records based on configurations which are active in SanteMPI.
Match Classification - Indicates the overall quality of the match
A/B Compare
The Matched Entities
tab shows summary information of the two patients which have been flagged as potential matches. However, there are times when a more holistic view of a patient match pair is desired. Some reasons the A/B compare may be used:
To see other relationships both the A and B records have (such as citizenship, employment, marital status, multiple-birth, etc.)
To view next of kin records on both A and B records.
To view additional identification information (from third party plugins) which would otherwise be rendered in the patient dashboard.
The A/B compare tab takes the contents of the SanteMPI Patient Detail (which is a common place for plugins to extend the UI) and shows the panels as an A/B comparison.
Scoring
The scoring tab presents a list of match configurations which are active on the SanteMPI server and presents each configured attribute, the weights, and calculated score.
This panel is used by data administrators to understand why the particular match pair came to be flagged. This information is helpful when determining whether a configuration needs to be changed, or to understand why a match is a match.
Classification - Indicates the match pair's classification as determined by the configuration.
Match Method - The method or algorithm used to score the matches
Weighted - The matching engine used a weighted scoring for the match
Deterministic - The matching engine used only a deterministic score for the match pair
Identifier - The matching engine made the determination based on identification of the source and target resources
Absolute Score - The numeric score assigned to the match pair. This score is a sum of all weighted scores for all attributes.
Strength - The normalized score which indicates on a scale of
0 .. 1
what the match agreement was.
The detail table illustrates the calculated (and weighted) score of each object (see: Matching Engine) for more details.
E | Indicates whether the attribute was even evaluated. If the attribute was not evaluated, then the | ✔️ - The attribute was evaluated and scored
✖️ - The attribute was not evaluated. The |
Attribute | Indicates the attribute identifier or name which was scored. | The path to the property in A and B values. |
A Value | The value of the property in record A which was evaluated. | |
B Value | The value of the property in record B which was evaluated. | |
M / U | Identifies the match probability (M) and the uncertainty (U) of the attributes. | M - The likelihood that if the A value and B value agree, then the record represents a true match. U - The likelihood that if the A value and B value agree, that they agree on a coincidence. |
Score | The numeric, calculated score given to the record (before partial scoring applied) | |
Weight | The final score applied to the overall score for the record. |
Related Topics
Master Data Management TabMatching ConfigurationMatching EngineLast updated