KR101793185B1 - Method for identifying patient personal information - Google Patents

Abstract

The method for identifying patient personal information according to an embodiment of the present invention updates the identification case priority used in performing the pattern analysis process through the step of performing the learning process so as to perform the pattern analysis process more quickly and accurately And the accurate matching result can be calculated accordingly.

Description

A method for identifying a patient '

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a method of identifying a patient personal information, and more particularly, to a method of identifying a patient personal information by using basic information of a patient to exchange medical information of the patient between hospitals.

Conventionally, a hospital information system is operated for each hospital, thereby improving the efficiency of the medical service in the hospital. However, it is difficult to exchange medical information between hospitals because the collection information and format of the hospital information system operated by each hospital are different. As a result, when the same disease is treated at another hospital, This causes various time and money waste.

In developed countries such as the United States, MPI (Master Patient Index) algorithms have been developed and used to identify and identify individual patients when using patient information between hospitals by comparing and analyzing patient information between different hospitals. While increasing quality and reducing waste of time and money.

In Korea, there is a unique identification number such as a resident registration number. However, due to the protection of personal information which is a global issue and personal information related laws in Korea, such unique identification number can not be used for patient identification.

In addition, since the algorithm of the currently disclosed MPI system is developed on the basis of a 1-byte English-language environment, it is difficult to directly apply it to a domestic environment using a 2-byte character and a combination character.

In order to solve such a problem, a patient personal information identification program suitable for the domestic situation in which a patient individual is identified by matching other information other than a unique identification number such as a resident registration number is needed.

Patent Document 1: Published Patent Application No. 10-2009-0126241

SUMMARY OF THE INVENTION The present invention has been made in order to solve the above problems, and it is an object of the present invention to provide a patient information management system and a patient information management system for identifying patient information by using other information except a unique identification number such as a patient's resident registration number Method.

A method for identifying a patient personal information according to an exemplary embodiment of the present invention includes: inputting (S210) a patient identification item through an input function unit as an input API capable of responding to various types of input by a module unit; The module unit preprocessing the patient identification item by the normalization rule DB of the database unit (S220); A module analyzing step (S230) of analyzing and analyzing the input type of the patient identification item to select a group of MPI identification cases matching the identification case with the highest matching rate in the master table of the database unit; Performing the exact matching on the MPI identification case group (S240); Determining whether the patient identification information for identifying the patient individual is unique (S250); In step S250, the module unit performs probability matching on the MPI identification case group in accordance with the absence of unique patient identification information in the patient identification information determination step S250. A step S270 of outputting the matching result information of the performing step S240 and the matching result information of performing the probability matching step S260, and generating and outputting a unique ID code; And the module unit learns information of the master table which is newly generated and accumulated in the patient information and the identification case information which is finally selected to identify the patient, and performs a learning process of updating the weight of the priority determination rule of the MPI identification case (S280).

The method of identifying a patient personal information according to an exemplary embodiment of the present invention may include a step of inputting the patient identification item in step S210 in which the name, date of birth, sex, address, mobile phone number, Type, primary care department, and date of occurrence.

In the patient personal information identification method according to an embodiment of the present invention, the preprocessing step (S220) includes loading (S221) the data input in the step S210 of receiving the patient identification item; A step (S222) of loading a pre-processing rule corresponding to the loaded input data through a pre-processing rule DB; Confirming (S223) the item-specific data type and length of the loaded input data; (S224) normalizing the data confirming the data type and the length into a predetermined regular expression format according to the pre-processing rule loaded through the pre-processing rule DB; And storing the normalized data as pre-processing result information (S225).

In the patient personal information identification method according to an embodiment of the present invention, the normalizing step S224 is a method of separating the beginning, the middle, and the ending by using Unicode representing Hangul and calculating an index and normalizing the index do.

In the patient personal information identification method according to an embodiment of the present invention, the pattern analysis step S230 may include a step S231 of checking whether there is data in the normalized input item through the normalizing step S224; Checking the input type of data existing in the normalized input item (S232); Accessing an identification case priority DB of the database unit and loading an identification case priority (S233); Selecting an identification case having the highest priority corresponding to the confirmed input type (S234); And storing the corresponding identification case sequence number and the matching rate threshold value (S235).

In the patient personal information identification method according to an embodiment of the present invention, the correct matching step S240 may include: (S241) starting the accurate matching by connecting the module unit to the master table of the database unit; A step (S242) of loading the identification case and the corresponding threshold in the step S230 of connecting the module to the master table and performing the pattern analysis process; A step (S243) of loading the preprocessed input data in the pre-processing step (S220); Loading a matching algorithm (S244); Comparing the date of occurrence of the input data of the patient identification item with the application date of the target record, and selecting the data before the occurrence date as the target (S245); Calculating a matching rate in the order of the identification case items using the matching algorithm (S246); And comparing the matching rate calculated in the order of the identification case item with a predetermined matching rate threshold value to determine unique identification information (S247).

The performing step S260 of performing the probability matching in the patient personal information identifying method according to the embodiment of the present invention may include the step S261 of loading the priority of the identified case in step S230 of performing the pattern analysis process, ; Accessing the identification case priority DB of the database unit and obtaining an identification case of the next priority used in the unidentified after the accurate matching (S262); Comparing the generation date of the input data inputted in the step S210 of receiving the patient identification item with the application date, and judging the application date record before the generation date as the target (S263); A step (S264) of calculating a matching rate in the order of the identification case items using the matching algorithm; And comparing the matching rate calculated in the order of the identification case item with the set matching rate threshold value to determine the unique identification information (S265).

The step S270 of generating and outputting the unique ID code in the patient personal information identification method according to the embodiment of the present invention includes performing the correct matching process S240 and performing the probability matching process S260. A step S271 of loading the processed matching results through the step S271; And determining (S272) whether the loaded matching result is a unique matching value, a case where the matching result is at least two matching values, or a case where there is no matching value at step S272 .

In the method for identifying patient personal information according to the embodiment of the present invention, when the matching result loaded in the step S272 of discriminating the three cases is the unique matching value in the master table, And outputs the same.

In the method for identifying patient personal information according to an embodiment of the present invention, when the number of the loaded matching results is two or more in step S272, the two or more matching results are displayed on the screen (S274-2); And a step (S275-2) of receiving a selection of a user for the output results of the two or more matching operations.

In the patient personal information identification method according to the embodiment of the present invention, when there is no matching result loaded in the step S272 of discriminating the three cases, it is determined whether to generate a new MRN number according to a user's input (S274-3); A step (S275-3) of generating a new MRN number according to a user's input; And a step (S276-3) of additionally storing and outputting information related to the generated MRN number to the master table.

In the patient personal information identification method according to an embodiment of the present invention, the step of performing the learning process (S280) includes loading (S281) a matching result selected for patient identification; Counting the successful cases of identification by connecting to the master table and restoring the identification case priority (S283); And updating the identification case priority DB of the database unit according to the re-calculated identification case priority (S284).

The features and advantages of the present invention will become more apparent from the following detailed description based on the accompanying drawings.

Prior to this, terms and words used in the present specification and claims should not be construed in a conventional, dictionary sense, and should not be construed as defining the concept of a term appropriately in order to describe the inventor in his or her best way. It should be construed in accordance with the meaning and concept consistent with the technical idea of the present invention.

The method for identifying patient personal information according to an embodiment of the present invention updates the identification case priority used in performing the pattern analysis process through the step of performing the learning process so as to perform the pattern analysis process more quickly and accurately And the accurate matching result can be calculated accordingly.

1 is a block diagram of an apparatus for performing a patient personal information identification method according to an embodiment of the present invention;
2 is an overall flowchart of a method for identifying patient personal information according to an embodiment of the present invention.
FIG. 3 is a flowchart illustrating a preprocessing step of a patient personal information identification method according to an embodiment of the present invention. FIG.
FIG. 4 is a flowchart illustrating a pattern analysis step of a patient personal information identification method according to an embodiment of the present invention. FIG.
FIG. 5 is a flowchart illustrating an accurate matching step of a patient personal information identification method according to an embodiment of the present invention; FIG.
FIG. 6 is a flowchart for explaining a probability matching step of a patient personal information identification method according to an embodiment of the present invention; FIG.
FIG. 7 is a flowchart illustrating an output step of a patient personal information identification method according to an embodiment of the present invention; FIG.
FIG. 8 is a flowchart for explaining a learning step of a patient personal information identification method according to an embodiment of the present invention; FIG.
FIG. 9 is a graph showing the accuracy of a patient personal information identification method according to an embodiment of the present invention. FIG.
FIG. 10 is a graph illustrating sensitivity of a patient personal information identification method according to an embodiment of the present invention. FIG.
11 is a graph showing the specificity of a patient personal information identification method according to an embodiment of the present invention.
12 is a graph showing the processing speed of a patient personal information identification method according to an embodiment of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS The objects, particular advantages and novel features of the present invention will become more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which: FIG. It should be noted that, in the present specification, the reference numerals are added to the constituent elements of the drawings, and the same constituent elements are assigned the same number as much as possible even if they are displayed on different drawings. Also, the terms first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another. In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. 1 is a block diagram of an apparatus for performing a patient personal information identification method according to an embodiment of the present invention.

An apparatus 100 for performing a patient personal information identification method according to an embodiment of the present invention includes a module unit 110 and a database unit 120. [

1, the module unit 110 includes an input function unit for inputting the patient identification item information, a preprocessing function unit for comparing the entered patient identification item information with the preprocessing rule DB, A probability matching function unit for increasing the identification rate of similar results after accurate matching, and an output function for outputting the final identification result And a learning function unit for performing a learning function for improving the accuracy in accordance with the accumulation of the partial and identification completed data.

The database unit 120 includes a master table for storing and managing a personal unique identification code, a preprocessing rule database (DB) for storing a preprocessing rule for normalizing input data, and an identification case priority rule for setting a group of identifiable items And an identification case priority DB storing the IDs.

At this time, the input function unit of the module unit 110 may include, for example, a name, a date of birth, a sex, an address, a mobile phone number, a fixed telephone number, an email address, insurance Type, main medical department, date of occurrence, and the like. Here, the type of insurance means an insurance qualification code of the patient (for example, a workplace subscription, a regional subscription, a medical care, etc.), and the occurrence date is a date when the data of the input item is actually generated Date of the time point).

The preprocessing function unit of the module unit 110 can normalize the entered patient identification items by formatting and calibrating them according to the normalization rule.

The pattern analysis function unit of the module unit 110 may check the input type of the patient identification item and select the identification case having the highest priority in the identification case priority DB of the database unit 120. [

The accurate matching function of the module unit 110 performs matching between the data input through the input function unit and the data stored in the master table based on the identification case item selected by the pattern analysis function unit, Patient identification information can be selected.

If the probability matching function unit of the module unit 110 fails to select a unique identification record in the accurate matching function unit, matching items of the identification case corresponding to the next priority are added one by one in the identification case to increase the matching probability It is possible to carry out the function of sorting.

The output function unit of the module unit 110 outputs the corresponding information if the unique patient identification information is selected after performing the matching process. If the information exceeds the matching rate threshold value, the output function unit outputs all of the information, If there is no information exceeding the selection matching rate threshold value, a unique identification code may be newly generated and output.

The learning function unit of the module unit 110 learns information of the master case table in which the identification case information and the patient information newly selected for patient identification are newly generated and stored and updates the weight of the priority order determination rule of the identification case can do.

The apparatus 100 for performing the method for identifying patient personal information according to the embodiment of the present invention identifies and identifies the patient identification information through the patient information identification algorithm using the module unit 110 and the database unit 120, You can improve accuracy by updating.

Hereinafter, a patient personal information identification method performed by the apparatus 100 for performing a patient personal information identification method according to an embodiment of the present invention will be described with reference to FIG. 2 to FIG. 3 is a flowchart illustrating a preprocessing step of a patient personal information identification method according to an embodiment of the present invention. FIG. 4 is a flowchart FIG. 5 is a flow chart for explaining an accurate matching step of a patient personal information identification method according to an embodiment of the present invention, and FIG. 6 7 is a flowchart illustrating an output step of a patient personal information identification method according to an embodiment of the present invention, and FIG. 8 Is a flowchart for explaining a learning step of a patient personal information identification method according to an embodiment of the present invention.

2, the method for identifying patient personal information according to an exemplary embodiment of the present invention includes a step of searching for a patient in the master table of the database unit 120 for the information received through the input function unit of the module unit 110, And a learning process for improving the identification accuracy in the data matching process.

First, a module 110 of an apparatus 100 for performing a patient personal information identification method according to an embodiment of the present invention is an input API capable of handling various types of input, for example, a name, a date of birth, A patient identification item such as a sex, an address, a mobile phone number, a fixed telephone number, an e-mail address, an insurance kind, a main medical care department, an occurrence date, and the like is inputted (S210).

Here, the input function unit of the module unit 110 receives the user input based on the data length defined in each of the patient identification items through the input API, and when the occurrence date is entered in blank, the input function unit of the module unit 110 And then stores the finally input data.

The module unit 110 performs a preprocessing process of formatting and calibrating the received patient identification items based on the normalization rule DB of the database unit 120 and normalizing them (S220).

After performing the preprocessing process, the module 110 confirms and analyzes the input type of the patient identification item and selects the MPI identification case group matching the identification case having the highest matching rate in the master table of the database unit 120 An analysis process is performed (S230).

After performing the pattern analysis process, the module 110 performs an accurate matching process for analyzing the patient identification information that is equal to or greater than the accurate matching rate threshold value in the MPI identification case group (S240).

In accordance with this accurate matching process, the module unit 110 determines whether the patient identification information for identifying the individual patient is unique (S250).

If there is no unique patient identification information in the patient identification information determination step S250, the module unit 110 performs a probability matching process for performing probability matching with the MPI identification case group (S260).

In the probability matching process, matching items are added to the MPI identification case group that is not judged in the accurate matching process (S240), and the patient identification information having high similarity is analyzed.

Calculates a probability matching rate by adding data not corresponding to the identification case, outputs the patient identification information when the probability matching rate threshold value of the identification case is equal to or greater than the threshold value, through the output function unit of the module unit 110, (S270), the unique ID code is newly generated and output.

Then, the module unit 110 learns the information of the master table, which is newly generated and stored in the case information and the patient information selected to finally identify the patient, and updates the weight of the priority determination rule of the MPI identification case (S280).

The process for identifying the patient personal information according to the embodiment of the present invention including the above process will be described in detail.

3, the step S220 of loading the input data in the step S210 of inputting the patient identification item (S221), the step of loading the pre-processing rule corresponding to each loaded item (S222), a step of checking the data type and length of each item with respect to the loaded input data (S223), a step of checking the data type and the length of the data by the preprocessing rule DB loaded in the preprocessing rule DB (Step S224), and storing the normalized pre-processing data as the pre-processing result information (step S225).

Here, the data input in the step of inputting the input data (S221) can be loaded as described in [Table 1].

number field type Length Explanation One name varchar 20 No spaces, no special characters, no numbers, only Hangul characters 2 date of birth varchar 8 No spaces, no special characters, only numbers
'YYYYMMDD' 3 gender varchar One English only, M or F 4 address varchar 100 Use of lot number address system, Hangul complete letter only
'xxxxxxxxxxxxxxxxxxxxxxaddress' 5 Cell Phone Number varchar 11 No spaces, only numbers 6 Home phone number varchar 11 No spaces, only numbers 7 e-mail varchar 50 Without blanks 8 Type of insurance varchar One Only numbers 9 Department of Primary Care varchar 2 English only 10 Date of occurrence varchar 8 No spaces, no special characters, only numbers 'YYYYMMDD'

In order to normalize the loaded input data in the regular expression format previously set in accordance with the pre-processing rule loaded in the step of normalizing (S224), for example, in the pre-processing rule DB for the 'name' You can execute the function to delete, the function to erase special characters, the function to erase numbers, the function to leave only Hangul, and the function to delete if there are more than 20 characters.

In addition, since a combination of 2 bytes is used in the case of Hangul, a restriction occurs when a matching algorithm is applied. In order to solve this problem and convert it into 1 byte like English and use it freely in the matching algorithm, it is possible to separate and normalize the Korean character into the initial, neutral, and longitudinal.

In this case, since the range of Unicode expressing Hangul is represented by 11,172 characters between 0xAC00 and 0xD7A3, it is possible to divide the beginning, the middle and the end and calculate the index by using the following Equation 1 using Unicode.

The step S230 of performing the pattern analysis process may include a step S231 of checking whether there is data in the normalized input item through the step S224 of normalizing the module 110 as shown in FIG. A step S232 of checking the input type of the existing data when the data exists in the input item, a step S233 of accessing the identification case priority DB and loading the identification case priority (S233) A step S234 of selecting an identification case having the highest ranking, and a step S235 of storing the corresponding identification case sequence number and the matching rate threshold value.

For example, when the patient information of the address, the date of birth, and the mobile phone number is inputted, as shown in [Table 2], among the identification cases corresponding to the input items in the priority DB of the identification case, Number 'case and a matching rate threshold of 80%, it is possible to store a corresponding threshold value of 10 and 80%.

Priority case Match rate threshold One Name, date of birth ((Name length -1) + (date of birth -1))
/ Total length of the item 2 Name, date of birth, cell phone number ((Name length -1) + (date of birth -1) + (cell phone number length -1)) / 3 Name, date of birth, address ((Name length -1) + (date of birth -1) + (address length -)) / total length of the item 4 Name, date of birth, home phone number ((Name length -1) + (date of birth -1) + (home phone number length -1)) / 5 Name, date of birth, email ((Name length -1) + (date of birth -1 length) + (length of e-mail -2)) / 6 Cell phone number, main department ((Cell phone number length -1) + (2))
/ Total length of the item ... ... ...

In step S240, the module 110 performs an accurate matching based on the input case and the identification case item selected in the step S230 of performing the pattern analysis process to select the patient identification information It is the process of issuing.

First, as shown in FIG. 5, the module unit 110 connects to the master table of the database unit 120 to start accurate matching (S241).

Thereafter, the module unit 110 loads the identification case selected in the step S230 of performing a pattern analysis process by connecting to the master table, a step S242 of loading a corresponding threshold value, a step S220 of performing a preprocessing process, A step S244 of loading the pre-processed input data in step S243, a step S244 of loading a matching algorithm, a step of comparing the generation date of the input data with the application date of the target record, S245), calculating a matching rate in the order of the identification case items using the matching algorithm (S246), comparing the matching rate calculated in the order of the identification case items with the predetermined matching rate threshold value to determine the unique identification information S247).

In this case, if the number of pieces of information in which the matching rate is equal to or higher than the threshold value is one in accordance with the step of determining the unique identification information (S247), the MPI information matched to the input value is recognized as one piece of the searched information S248-1), and performs step S249-1 of generating MPI information matched with the input value as unique identification information.

On the other hand, if it is determined in step S247 that the unique identification information is not found, if there is no record exceeding the threshold value or if the number of records exceeding the threshold value is two or more, step S249-2 is generated as the probability matching information. . Here, the probability matching information generated in step S249-2 of generating the probability matching information is used to calculate the final result value in step S260 of performing the probability matching process.

The process of calculating the matching rate of the matching algorithm in step S240 of performing the exact matching process will be described below in Korean, English, and numbers.

For example, when the input information entered for the objective record is as follows,

Purpose Record: Hong Kil Dong, gdhong@naver.com , 01047515661

Input information: Hong Gil, et al., Gghong@naver.com , 01047515231

If the matching algorithm is applied to the case where the priority order matching the input item is applied to the identification case priority DB, for example, the name, e-mail, and cell phone number are selected as matching items, the similarity to each item is as follows.

s (x, y) = 1 d (x, y) / max (length (x), length

String str1 = "ㅎ ㅗ ㄱ ㄱ";

String str2 = "ㅎ ㅗ ㄱ ㄱ";

Since the eighth "ㅗ" of str1 is changed to the eighth "ㅡ" of str2, the distance is 1, and the length of the entire string is 9,

1 - (1/9) = similarity score of 0.8888889

String str1 = "gdhong@naver.com";

String str2 = " gghong@naver.com ";

Since the second "d" of str1 needs to be changed to the second "g" of str2, the distance is 1 and the length of the entire string is 16,

1 - (1/16) = similarity score of 0.9375

String str1 = "01047515661";

String str2 = "01047515231";

the distance needs to be changed from the 9th "6" of str1 to the 9th "2" of str2 and the 10th "6" of str1 to the 10th "3" of str2, Since the length of the entire string is 11 at its maximum,

1 - (2/11) = similarity score of 0.8181818

Respectively.

The step of performing the probability matching process (S260) may include a step of performing an accurate matching process (S240). If the unique identification record generated by the unique identification information is not selected, the MPI identification case Is added one by one and matching is performed to increase the matching probability.

Specifically, the step of performing the probability matching process (S260) includes a step (S261) of loading the priority order of the identified case in the step S230 of performing the pattern analysis process (S261) (S262), comparing the generation date of the input data with the application date, and judging the application date record before the occurrence date as the target (S263). The matching algorithm A matching rate is calculated in the order of the identification case items in step S264, and the matching rate calculated in the order of the identification case items is compared with the matching rate threshold value to determine the unique identification information in step S265.

If it is determined in step S265 that the unique identification information is greater than or equal to the threshold value, the unique identification information is generated and stored in step S267.

The step S270 of generating and outputting the unique identification code outputs the processed results through the step S240 of performing an accurate matching process and the step S260 of performing a probability matching process and if there is no corresponding information in the master table , And a new unique identification code is generated.

Specifically, as shown in FIG. 7, the step of generating and outputting the unique identification code includes a step S240 of performing an accurate matching process and a step S260 of performing a probability matching process, A step of loading the results (step S271), a step of judging whether the loaded matching result is divided into three cases, that is, a case where the matching result is a unique matching value, a case where the matching result is two or more matching values, (S272).

At this time, when the matching result loaded in the step S272 of discriminating the three cases is the unique matching value in the master table, the searched patient data is displayed as MRN number and outputted (S274-1).

If the number of the matching results loaded in the step S272 is three or more, the step S274-2 of outputting all the matching results to the screen (step S274-2) And a step S275-2 of receiving the user's selection of the result is performed.

Finally, if there is no matching result loaded in the step S272 of discriminating the three cases, it is determined whether to generate a new MRN number according to the input of the user (S274-3) A step S275-3 of generating a new MRN number according to the input, and a step S276-3 of additionally storing and outputting the generated MRN number and information related thereto in the master table.

In the step of performing the learning process (S280), the identification case information finally selected for the patient identification and the information of the master table in which the patient information is newly generated and stored are learned using the adaptive algorithm for cumulatively analyzing the input patterns And updating a weight for the priority determination rule of the identification case.

Specifically, the step of performing the learning process (S280) includes the steps of loading (S281) a matching result finally selected for patient identification, connecting to a master table (S282) (S283) of counting the identification case priorities by counting successful cases of identification through S231, and updating the identification case priority DB according to the re-calculated identification case priority (S284).

In step S280, the initial weight value is given for each priority, the weight value is increased for the pattern used at the time of outputting the result, and a new priority value including the increased weight value is used for the next pattern analysis. To be applied.

Accordingly, the step of performing the learning process (S280) may perform the pattern analysis process (S230) more quickly and accurately by updating the identification case priority used in the step S230 of performing the pattern analysis process Accordingly, an accurate matching result can be calculated.

Experiment Comparative Example

Tests were conducted to compare with other known matching algorithms for identification accuracy, sensitivity, specificity, and processing speed according to the patient personal information identification method according to an embodiment of the present invention.

Accordingly, the other matching algorithm as the comparative example selects the Levenshtein algorithm, the JaroDistance algorithm, and the NeedlemanWunsch algorithm described in [Table 3], applies it to 45 identical sample cases together with the patient personal information identification method according to the embodiment of the present invention The test was conducted.

Comparative Example Way Characteristic One Levenshtein
(Edit Distance) How to measure how many changes are needed when changing one string to another Relatively simple and universally usable 2 JaroDistance Measure accuracy with matching and number of potential characters Suitable for short string comparison 3 NeedlemanWunsch Using penalty gaps to measure similarity Suitable for comparison of sentence unit similarity

At this time, an error matrix (Confusion Matrix) was used to quantitatively evaluate the test results. In order to compare the relative performance between the patient personal information identification method and the comparative examples according to the embodiment of the present invention by using the corresponding error matrix through accuracy, sensitivity, and specificity, TP (True Positive ), FP (False Positive), FN (False Negative), and TN (True Negative) can be defined as follows.

TP: When the same result is matched with the same result

FP: When the same person is matched by a result that is not the same

TN: if they are not the same person, they are matched by the same result

FN: when not identical, matching with the same result

The accuracy, sensitivity, and specificity using the error matrix elements are defined as shown in Equation (2) below.

Accuracy test result

The accuracy test results were detected with an accuracy of 64.44% for Levenshtein, 67.78% for Jaro, 73.33% for Needleman, and 94.44% for the patient personal information identification method according to the embodiment of the present invention, as shown in the graph shown in FIG. , It can be seen that the patient personal information identification method according to the embodiment of the present invention has higher accuracy than the algorithm of the comparative example.

This result is obtained with a higher accuracy in the patient personal information identification method according to the embodiment of the present invention in which collation is performed and matching is performed as compared with the comparative example algorithm in which matching is performed without separating Hangul.

Sensitivity test results

Sensitivity is a reference measure that can be judged to be the same in the case of the same person, and it is intended to measure how accurately the same person can be discriminated in the case of the same person.

As a result of the sensitivity test, as shown in FIG. 10, sensitivity of 62.26% for Levenshtein, 64.81% for Jaro, 68.42% for Needleman, and 95.45% for patient identification method according to the embodiment of the present invention It can be seen that the patient personal information identification method according to the embodiment of the present invention leads to a higher sensitivity than the methods of the comparative example.

Such a sensitivity result is regarded as a part that is not matched even if only one character of one letter of the comparative example is mistaken, while the method of identifying the patient personal information according to the embodiment of the present invention, . ≪ / RTI >

The difference is significant when the normalization expression can not be processed through preprocessing. In particular, unlike the methods of the comparative examples, it can be confirmed that the method of identifying the patient personal information according to the embodiment of the present invention, in which probability of typos is reflected in the matching rate after normalization in the part such as the Korean alphabet, is applied.

Specificity test result

The specificity is a reference measure to judge that it is not the same when not the same person, so as to accurately measure the case where the person is not the same person.

As a result of the specificity test, as shown in FIG. 11, 67.57% of Levenshtein, 72.22% of Jaro, 81.82% of Needleman, and 93.48% of the patient personal information identification method according to the embodiment of the present invention Thus, it can be seen that the method of identifying patient personal information according to the embodiment of the present invention has higher specificity than the methods of the comparative example.

The results of this specificity test show that, in comparison with the methods of the comparative examples, the patient personal information identification method according to the embodiment of the present invention provides a normalization priority on the preprocessing through a learning process, And a high specificity can be derived by recommending a case where the probability is high or low.

This specificity is a part that can be closely related to the processing operation time, which is different from the methods of the comparative examples.

Throughput test results

The test on the processing speed was performed by matching 10,000 records and measuring the time until the final result was obtained.

As shown in FIG. 12, the processing speed result is obtained by processing the patient personal information identification method according to the embodiment of the present invention in a shorter time than in the methods of the comparative example, and thus the processing speed is high.

Although the technical idea of the present invention has been specifically described according to the above preferred embodiments, it is to be noted that the above-described embodiments are intended to be illustrative and not restrictive.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit and scope of the invention.

100: a device for performing a patient personal information identification method
110: module section 120: database section

Claims (12)

A step (S210) of receiving a patient identification item through an input function unit as an input API capable of responding to various types of input by the module unit;
The patient identification item to which the module is input is normalized to a predetermined regular expression format according to a preprocessing rule loaded in the normalization rule DB of the database unit, and the preprocess, neutrality, and trailing are separated using Unicode representing the Korean language A step S220 of calculating an index by using the following Equation 1 and pre-processing to normalize the index;
[Equation 1]

A module analyzing step (S230) of analyzing and analyzing the input type of the patient identification item to select a group of MPI identification cases matching the identification case with the highest matching rate in the master table of the database unit;
Performing the exact matching on the MPI identification case group (S240);
Determining whether the patient identification information for identifying the patient individual is unique (S250);
In step S250, it is determined that there is no unique patient identification information. In step S240, the module adds data not corresponding to the MPI identification case group, A probability matching step (S260) for performing re-matching to calculate a matching rate;
If the matching result information of step S240 is greater than or equal to the probability matching rate threshold in step S260, the matching result information is output as matching result information. Generating and outputting a unique identification code when the probability matching rate threshold is not greater than or equal to the threshold (S270); And
Performing a learning process of updating the weight of the priority determination rule of the MPI identification case by learning the identification case information and the patient information newly selected and accumulated in the final patient identification and the information of the accumulated master table S280);
The method comprising the steps of:
The method according to claim 1,
The patient identification item input in the step S210 of receiving the patient identification item includes a name, a date of birth, a sex, an address, a mobile phone number, a wired phone number, an e-mail address, a type of insurance, How to Identify a Patient's Personal Information.
The method according to claim 1,
In the pre-processing step S220,
A step (S221) of loading the data input in the step S210 of receiving the patient identification item;
A step (S222) of loading a pre-processing rule corresponding to the loaded input data through a pre-processing rule DB;
Confirming (S223) the item-specific data type and length of the loaded input data; And
Storing the normalized data as preprocess result information (S225);
The method comprising the steps < RTI ID = 0.0 > of: < / RTI >

delete The method according to claim 1,
The pattern analysis step (S230)
A step S231 of confirming whether there is data in the normalized input item through the normalizing step S224;
Checking the input type of data existing in the normalized input item (S232);
Accessing an identification case priority DB of the database unit and loading an identification case priority (S233);
Selecting an identification case having the highest priority corresponding to the confirmed input type (S234); And
Storing a corresponding identification case sequence number and a matching rate threshold value (S235);
The method comprising the steps < RTI ID = 0.0 > of: < / RTI >
The method according to claim 1,
The step S240 of performing the accurate matching
(S241) of connecting the module unit to the master table of the database unit to start accurate matching;
A step (S242) of connecting the module to the master table and loading a threshold value corresponding to the selected identification case in the pattern analysis step (S230);
A step (S243) of loading the preprocessed input data in the preprocessing step (S220);
Loading a matching algorithm (S244);
Comparing the date of occurrence of the input data of the patient identification item with the application date of the target record, and selecting the data before the occurrence date as the target (S245);
Calculating a matching rate in the order of the identification case items using the matching algorithm (S246); And
Comparing the matching rate calculated in the order of the identification case items with a preset matching rate threshold value to determine unique identification information (S247);
The method comprising the steps < RTI ID = 0.0 > of: < / RTI >
The method according to claim 1,
The step of performing the probability matching (S260)
A step (S261) of loading the identified case priority in the pattern analysis step (S230);
Accessing the identification case priority DB of the database unit and obtaining an identification case of the next priority used in the unidentified after the accurate matching (S262);
Comparing the generation date of the input data inputted in the step S210 of receiving the patient identification item with the application date, and judging the application date record before the generation date as the target (S263);
A step (S264) of calculating a matching rate in the order of the identification case items using the matching algorithm; And
Comparing the matching rate calculated in the order of the identification case items with the set matching rate threshold value to determine the unique identification information (S265);
The method comprising the steps < RTI ID = 0.0 > of: < / RTI >
The method according to claim 1,
The step (S270) of generating and outputting the unique identification code
A step (S271) of performing the precise matching (S240) and a matching result processed through the performing of the probability matching (S260); And
(S272) if the loaded matching result is a unique matching value, if the matching result is two or more matching values, and if there is no matching value;
The method comprising the steps < RTI ID = 0.0 > of: < / RTI >
9. The method of claim 8,
If the loaded matching result is a unique matching value in the master table, the searched patient data is displayed and output as MRN number. Identification method.
9. The method of claim 8,
(S274-2) of outputting the two or more matching results to the screen when the number of the loaded matching results is two or more in the step S272 of discriminating the three cases; And
Receiving (S275-2) a selection of a user for the two or more matching results;
Further comprising the steps of:
9. The method of claim 8,
If it is determined in step S272 that there is no matching result, step S274-3 is performed to determine whether to generate a new MRN number according to the input of the user.
A step (S275-3) of generating a new MRN number according to a user's input; And
Storing and outputting information related to the generated MRN number to the master table (S276-3);
Further comprising the steps of:
The method according to claim 1,
The step of performing the learning process (S280)
Loading the selected matching result to identify the patient (S281);
Counting the successful cases of identification by connecting to the master table and restoring the identification case priority (S283); And
Updating the identification case priority DB of the database unit according to the re-calculated identification case priority (S284);
The method comprising the steps < RTI ID = 0.0 > of: < / RTI >

Images