Patent Application
20140149140
This invention relates generally to a system and method for a patient identification index. More specifically, a system and method for matching a plurality of patient identification formats with a particular patient and allowing for the integration of such various formats into a single system.
Within the healthcare setting, there are a large number of computerized systems, diagnostic systems and therapy devices. Most of these systems require a patient identifier in order to enable certain data storage, safety, or tracking features within the system. In many cases, the safe and effective use of the device depends on proper entry of the patient's identification.
Recent studies have found that identification errors within these systems are significant. Because there is not a standard in the US for how a patient will be identified, each hospital and device provider is free to choose from any of a number of possible IDs. Any of patient name, bed number, medical record number, episode of care, social security number, date of birth, or any similar identifier, can be used to identify a patient. This practice makes it very difficult to integrate information across multiple systems for use in real time diagnosis and treatment, or use in retrospective analysis for process improvement. The practice also results in a significant number of repeated tests or lost results due to ID errors.
Common causes of data entry error include worn out/stained barcodes, multiple wristband ID formats and/or accounts, scanning barcodes from other hospital facilities, using transient/arbitrary account numbers (e.g. 999-999-999), manually entering identification in an outpatient facility with no access to barcoded wristbands, or selecting of wrong account/episode number when wristbanding a patient. Such errors are likely to result in a delay in the delivery of care and other similar problems.
Accordingly, there is a need for a system and method that facilitates the integration of various forms of patient identification. Furthermore, there is a need for a system and method that minimizes common causes of data entry errors in such patient identification systems. The present invention fulfills these needs and provides other related advantages.
The present invention is directed to a system and method that facilitates the integration of various forms of patient identification. The inventive system and method also minimizes common causes of data entry errors in such patient identification systems.
In the present invention, a server on a hospital network is configured to receive and process all Admit Discharge Transfer (ADT) messages that are generated from any system in the hospital. As messages are received, the server stores all messages and builds a reference table that includes fields for all known elements of patient ID used in the hospital. A server application continually checks to match active patients (those who have been admitted but not yet discharged) with known previous patients (those treated and discharged in the past). The server application also checks to match active patients where two or more separate records using different elements/forms of patient ID may have been opened. As matches are made, the records are updated to link all different elements of ID so that the identification problems described above are eliminated.
To create the reference table, each hospital identifies all aspects of ID that are possibly used in the hospital. This will generate the initial list of fields for the database. In addition to the ID attributes, an association with certain assets (equipment), disease states and/or clinicians may also be included.
The present invention is directed to a process for creating a patient identification index for a patient care environment. The process includes the step of receiving a plurality of ADT (Admit Discharge Transfer) records corresponding to a respective plurality of patients in the patient care environment, wherein each ADT record comprises one or more patient identifiers. The ADT records are stored in an electronic database and a reference table having separate data fields for different forms of patient identifiers used in the patient care environment is built. Each ADT record is processed so as to assign each of the patient identifiers to a corresponding one of the data fields in the reference table. The reference table is continually updated with additional ADT records as they are received. A first ADT record for an identified patient is matched with a second ADT record for the identified patient based upon one or more common patient identifiers from the first and second ADT records.
The matching includes matching a first patient identifier from the first ADT record to a second patient identifier from the second ADT record and linking the first patient identifier to the second patient identifier in the reference table. The matching step also includes linking all common patient identifiers from all available ADT records for the identified patient. The patient identifiers include last name, first name, medical record number, accession number, social security number, date of birth, insurance number, in-patient wristband number, or temporary admission number. All forms of patient identifiers used in the patient care environment are preferably listed to assist in building the reference table.
The reference table may also include separate data fields for equipment, disease states, clinicians, or episodes of care. In this case, each ADT record may also include one or more asset identifiers for equipment, disease states, clinicians, or episodes of care. Where asset identifiers are used, the matching step further involves creating an association between the patient identifiers of the first ADT record for the identified patient and one or more asset identifiers for equipment, disease states, clinicians, or episodes of care from a third ADT record for the identified patient.
The matching also includes cross-referencing the first and second ADT records for the identified patient with a fourth ADT record for the identified patient corresponding to a prior admission of the identified patient to the patient care environment.
The matching step also includes calculating a matching score for the one or more common patient identifiers from the first and second ADT records for the identified patient and determining whether the matching score exceeds a predetermined threshold value for a likeliness of match. The matching step also involves identifying conflicts between patient identifiers of the first ADT record for the identified patient and patient identifiers of a fifth ADT record for either the identified patient or another patient.
The present invention is also directed to a system for creating a patient identification index for a patient care environment. The system includes a computer processor and electronic database connected to a network. The computer processor includes a data analysis module and the electronic database includes a reference table. The reference table has separate data fields for different forms of patient identifiers used in the patient care environment. The data analysis module is programmed to receive a plurality of ADT records corresponding to a respective plurality of patients in the patient care environment, wherein each ADT record comprises one or more patient identifiers. The ADT records are stored in the reference table on the electronic database. Each ADT record is processed so as to assign each of the patient identifiers to a corresponding one of the data fields in the reference table. The reference table is continually updated with additional ADT records as they are received. A first ADT record for an identified patient is matched with a second ADT record for the identified patient based upon one or more common patient identifiers from the first and second ADT records.
The data analysis module may also be programmed to match a first patient identifier from the first ADT record to a second patient identifier from the second ADT record and link the first patient identifier to the second patient identifier in the reference table. The data analysis module also matches the first ADT record with the second ADT record by linking all common patient identifiers from all available ADT records for the identified patient. The data analysis module is further programmed to list all forms of patient identifiers used in the patient care environment. The patient identifiers comprise last name, first name, medical record number, accession number, social security number, date of birth, insurance number, in-patient wristband number, or temporary admission number.
The reference table also includes separate data fields for equipment, disease states, clinicians, or episodes of care, and wherein each ADT record comprises one or more asset identifiers for equipment, disease states, clinicians, or episodes of care. The data analysis module is further programmed to create an association between the patient identifiers of the first ADT record for the identified patient and one or more asset identifiers for equipment, disease states, clinicians, or episodes of care from a third ADT record for the identified patient.
The data analysis module is further programmed to cross-reference the first and second ADT records for the identified patient with a fourth ADT record for the identified patient corresponding to a prior admission of the identified patient to the patient care environment. The data analysis module may also calculate a matching score for the one or more common patient identifiers from the first and second ADT records for the identified patient and determine whether the matching score exceeds a predetermined threshold value for a likeliness of match.
The data analysis module also identifies conflicts between patient identifiers of the first ADT record for the identified patient and patient identifiers of a fifth ADT record for either the identified patient or another patient.
Other features and advantages of the present invention will become apparent from the following more detailed description, taken in conjunction with the accompanying drawings, which illustrate, by way of example, the principles of the invention.
The accompanying drawings illustrate the invention. In such drawings:
The present invention is concerned with a system and method for a patient identification index. More specifically, a system and method for matching a plurality of patient identification formats with a particular patient and allowing for the integration of such various formats into a single system.
In a preferred embodiment, the server 12 on a hospital network 10 is configured to receive and process all Admit Discharge Transfer (ADT) messages 20 that are generated from any system in the hospital. Such ADT messages 20 are well established in the industry and are preferably formatted according to the HL7 standard for data interchange.
As messages 20 are received, the server 12 stores all messages 20 and builds a reference table 22 that includes fields for all known elements of patient ID used in the hospital. A server application continually checks to match active patients (those who have been admitted but not yet discharged) with known previous patients (those treated and discharged in the past). The server application also checks to match active patients where two or more separate records using different elements/forms of patient ID may have been opened. As matches are made, the records are updated to link all different elements of ID so that the identification problems described above are eliminated.
To create the reference table 22, each hospital will identify all aspects of ID that are possibly used in the hospital. This will generate the initial list of fields for the database. In addition to the ID attributes, an association with certain assets (equipment), disease states and/or clinicians may also be included.
The same server 12 also provides a look-up capability for all systems in the hospital so that identification of a patient can be verified prior to use by one of those systems. Additional ID and demographic information may also be supplied. For example, prior to an EKG test the EKG machine would send the patient name that had been entered. The server 12 would verify that the name corresponded to a valid active patient and could return both the validation and additional information about the patient such as date of birth 28 or Insurance information 36.
In
A partial list of the types of devices that might query for Patient ID Info include:
The server would also have the ability to scan previously discharged patients in order to gather and populate missing data. The data could be automatically updated or alternately presented to a person for validation.
Intelligent algorithms enable the server to calculate a score for the “likeliness of match” based on data elements that are available. Scores above a certain threshold cause the server 12 to automatically update ID data while lower scores will cause the data to be presented to a human for validation.
Several known algorithms that can accomplish this calculation include the Boyer-Moore string search algorithm, the Knuth-Morris-Pratt string search algorithm and the Rabin-Karp string search algorithm. Boyer-Moore searches a string of data for a pattern in the data by using information gathered during a pre-search step to skip section of data that do not fit the pattern. This functionality results in a lower constant factor compared to other string algorithms. Knuth-Morris-Pratt searches for occurrences of a particular word in a main text string by observing when a mismatch occurs to determine where the next match could begin. This allows the algorithm to skip re-examination of previously matched characters. Rabin-Karp uses hashing (mapping large data sets of variable length to smaller data sets of fixed length) to find any one of a set of pattern strings in a text.
Each of these algorithms or any other that is used can be augmented by domain knowledge in the priority and weighting of certain factors. Such factors include a medical record number being given the highest weight followed by episode number or accession number. Other factors such as last name, first name, and date of birth would be indicators but not certain identifiers in the algorithm. Room and bed numbers would be the lowest level weight since they can be more easily changed than any of the other factors. The system includes tools and configurations that would allow each patient care environment to configure the priority and weighting of the algorithms.
Although several embodiments have been described in detail for purposes of illustration, various modifications may be made without departing from the scope and spirit of the invention. Accordingly, the invention is not to be limited, except as by the appended claims.
