JP2005301600A - Case retrieval system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a case retrieval system that assists physicians in analyzing data. <P>SOLUTION: The case retrieval system includes a time-series medical data storage part 121; a representative time point defining part 120 for defining processing requirements for extracting a certain point of time and the date corresponding to the certain point of time; an input part 100 for inputting predetermined retrieval items and the predetermined certain point of time defined in advance by the representative time point defining part 120; a representative time point retrieval part 110 for extracting the corresponding date by implementing the corresponding processing requirements of the representative time point defining part 120 in accordance with the certain point of time; a time-series data retrieval part 111 for extracting the value corresponding to the date with regard to the retrieval items; and an output part 101 for outputting the extraction results. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a case retrieval system that efficiently extracts medical data as a basis for diagnosis from a database and supports data analysis of doctors and pharmaceutical companies.

  In recent years, the practice of evidence-based medicine (EBM) has become an important concept in order to provide higher-quality insurance medical services. On the other hand, as medical systems such as electronic medical record systems that electronically manage medical data are networked, large-scale medical databases are being built. As a result, it is possible to dynamically derive the diagnosis basis from the database, and in the future, each medical institution will extract the diagnosis basis and evaluate the extracted diagnosis basis and implement a high-quality service in which EBM is put into practice. Can be provided. In clinical research to extract the basis of diagnosis, doctors make hypotheses based on medical knowledge, extract necessary data from a database, and statistically verify it. Therefore, a technology for efficiently extracting and analyzing target data from a large amount of data is important.

  As an existing data analysis technique, there is a method using Online-Analytical-Processing (OLAP). OLAP is used by end users to search and aggregate the database directly, and find problems and issues from them. The OLAP tool provides various inquiry functions for aggregation along various axes (dimensions). And provide data manipulation functions. For example, the OLAP tool of Microsoft SQL Server 7.0 has the functions of (1) cube construction, search execution, and (2) data reference based on the cube. The cube here indicates a dimension that defines the starting point of analysis and a definition of an aggregation target. As an analysis procedure using the OLAP tool, an administrator who is familiar with the database in advance performs (1), and the end user refers to the aggregated result while changing the dimension value designed in (1) in (2). A well-known example of using an OLAP tool for medical data is “Development of medical support and hospital management support system using OLAP technology” (Medical Informatics 23 (2), 2003: 159-164). In the above known example, data necessary for analysis is extracted from various business systems, an analysis database is created, and the OLAP tool is used to calculate the number of patients and the average medical remuneration score based on the clinical practice, disease name, drug, etc. We are building a system that can be easily referenced.

Medical Informatics 23 (2), 2003: 159-164

  The data used in clinical studies includes a lot of time-series data such as patient test results. A cross-sectional study that extracts and compares data at a certain point for each patient, and extracts data at a certain point and a certain point in time. Or for longitudinal studies such as comparing changes in data over time between time points. In the above-described prior art, in order to perform the above-described analysis in order to perform aggregation for all records stored in the analysis database, the analysis database is stored in a form in which data at a necessary time is previously extracted. Need to build. However, there are many time series data of patients, and what time point is taken for each item (for example, the test value is the test value on the first test day, the value is half a year after the first test date, etc.) Since it differs depending on the research design of the doctor, it is difficult to store the data at the predicted time point in the analysis database. Therefore, in the past, engineers who are familiar with the database extract and provide the necessary data according to the research design according to the research design without using the OLAP tool. There is a problem of lowering.

  An object of the present invention is to provide a case search system in which a user can easily extract data at a certain time point from time-series data composed of a plurality of time points in order to support analysis of time-series data.

  In order to solve the above problems, a case search system of the present invention includes a time-series medical data storage unit that stores time-series medical data including a plurality of time points, representative time point identification information that identifies a certain time point, and representative time points Representative time point defining unit for storing processing conditions for extracting the date and time corresponding to the above, a representative time point selecting unit for selecting predetermined representative time point identification information from the representative time point identification information, and an item input unit for inputting a predetermined search item And a representative time point search unit that extracts the date and time of the representative time point for each search unit from the time-series medical data storage unit by processing using the corresponding processing conditions of the representative time point definition unit according to the selected representative time point identification information; According to the input search items, a time series data search unit is provided for extracting values corresponding to the date and time extracted by the representative time point search unit from the time series medical data storage unit.

  In addition, the case search system of the present invention includes a time information input unit that inputs time information from a certain point in time, and an addition date and time obtained by adding the time information to the date and time of the representative point extracted by the representative point search unit And the date and time closest to the addition date and time are extracted for each search unit from the time-series medical data storage unit.

  In addition, the case search system of the present invention includes a start point input unit for inputting a start point to be searched for time series data from a representative time point setting unit and a time information input unit, and an end point to be searched for a representative time point setting unit and time information. In the end point input unit input from the input unit and the representative time point search unit, the start point date and time is calculated from the representative point time identification information and time information input in the start point input unit, and the representative time point identification information and time input in the end point input unit The end point date and time is calculated from the information point in time, and values corresponding to all the date and time included between the start point date and end point date and time extracted by the representative point time point search unit are extracted from the time series medical data storage unit according to the input search items. I tried to do it.

  According to the case search system of the present invention, the user can efficiently extract data necessary for time-series data analysis from time-series medical data composed of a plurality of time points, and clinical research can be accelerated.

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing a system configuration example of a case search system according to the present invention. The case retrieval system of this example includes an input / output device 10, an arithmetic processing device 11, and a storage device 12. The input / output device 10 includes an input unit 100 and an output unit 101, and the input unit 100 includes a representative time point selection unit 102, a time information input unit 103, and an item setting unit 104. The arithmetic processing unit 11 includes a representative time point search unit 110 and a time series data search unit 111. The storage device 12 includes a representative time point definition unit 120 and a time-series medical data storage unit 121.

  The time series medical data storage unit 121 includes a table storing various time series analysis medical data necessary for analysis, such as a test result table and a prescription information table. Each table is managed by patient ID, and when searching for information across a plurality of tables, each table is linked using the patient ID as a key, and the corresponding data is searched. FIG. 1 shows a test result table that is an example of time-series medical data. In the examination result table, there are a plurality of examination dates and times for each patient ID, and examination results (medical item values) such as examination values A (medical items) are stored.

  Here, the representative time point means a certain time point defined for a certain search unit such as a patient, a drug, or a case with respect to time-series data composed of a plurality of time points. For example, when searching for an examination result table on a patient basis, examples of representative time points include the first examination date and time, the latest examination date and time, the date and time when the examination value A is the maximum value, and the like.

  The representative time point search unit 110 defines the definition data stored in the representative time point definition unit 120 and the medical data stored in the time series medical data storage unit 121 according to the contents input from the representative time point selection unit 102 and the time information input unit 103. Is used to extract the date and time of the time-series medical data corresponding to the representative time point. The time series data search unit 111 extracts a corresponding value from the time and date extracted by the representative time point search unit 110 from the time series medical data storage unit 121 for the medical item input from the item setting unit 104. The extraction result is displayed on the output unit 101.

(First embodiment)
FIG. 2 is a diagram showing a table structure of the representative time point defining unit 120 in the first embodiment of the present invention. The representative time point definition unit 120 stores a search unit 20, a representative time point name 21, and a processing condition 22 for extracting the date and time corresponding to the representative time point name for each search unit. The processing condition 22 may be defined by dividing processing conditions necessary for data extraction as table items, or a corresponding database query statement may be directly defined as in the example of FIG.

  In the case of “search unit = patient” and “representative point name = initial examination”, “processing condition = SELECT patient ID, MIN (examination date) FROM examination result table WHERE Group by patient ID”. However, the processing conditions in this example are written in SQL (Structured Query Language), which is generally known as a database language for realizing database definitions and operations. Indicates a database query statement that extracts the examination date and time as a value. In the representative time point selection unit 100, a representative time point name 21 corresponding to each search unit 20 is preset as an option. Thus, for example, when “representative point name = first examination” is input in the search for each patient, the above processing conditions are executed, and the first examination date and time for each patient is extracted.

  FIG. 3 is a diagram showing a processing procedure in the first embodiment of the present invention. Here, in order to investigate the relationship between the genetic polymorphism of a patient and the test value A in clinical research, a case will be described in which data at the first test is searched for the test value A as a representative time point.

  In step 300, the medical item “examination value A” to be searched is set in the item setting unit 104. In step 301, the representative time point selection unit 102 sets the representative time point name “first inspection”.

  Next, the representative time point search unit 110 executes Step 302 and Step 303. In step 302, the representative time point definition unit 120 is searched from the value input from the representative time point selection unit 102, and the corresponding processing condition is extracted. In step 303, the processing conditions extracted in step 302 are executed, the time series medical data storage unit 121 is searched, and the examination date and time corresponding to each patient is extracted. In the examination result table of FIG. 1, the extraction result is a shaded display portion that is the date and time of the first examination of each patient, that is, (ID1,2003 / 06/05) (ID2,2003 / 04/25) (ID3,2003 / 05/06) (ID4, 2003/08/01) (ID5, 2003/03/22)

  Then, the time series data search unit 111 executes Step 304 and Step 305. In step 304, the time-series medical data storage unit 121 is searched for the medical item “examination value A” input from the item setting unit 104, and a value corresponding to the examination date and time extracted in step 303 is extracted. Extraction results are (ID1,2003 / 06/05, Value1) (ID2,2003 / 04/25, Value4) (ID3,2003 / 05/06, Value6) (ID4,2003 / 08/01, Value8) (ID5 , 2003/03/22, Value10). In step 305, the extraction result obtained in step 304 is displayed on the output unit 101.

  From the above, in this embodiment, an end user who is not a skilled database administrator can easily extract patient time-series data at a certain point in time according to the study design. In addition, by defining processing conditions for extracting representative time points in advance, for example, complicated representative time points such as occurrence of clinical events such as disease recurrence can be defined, which can be applied to various research designs.

(Second embodiment)
FIG. 4 is a diagram illustrating the input unit 100 of the input / output device 10 according to the second embodiment of the present invention. The input unit 100 includes an item setting unit 104, a representative time point selection unit 102, and a time information input unit 103. The time information input unit 103 includes a representative time point addition time setting unit 40 and a representative time point addition time margin setting unit 41. The In the item setting unit 104 and the representative time point selection unit 102, when a horizontal arrow is clicked, a plurality of medical items or a plurality of representative time points are displayed in a pull-down manner so that appropriate items can be selected.

  The representative time point addition time setting unit 40 and the representative time point addition time margin setting unit 41 are used when the time points before and after the time point extracted by the representative time point selection unit 102 are set. For example, the representative time addition time “+6 months” and the representative time addition time margin “± 14 days” are closest to the date and time 14 months before and after 6 months after the time corresponding to the representative time name. Means that. In the representative time addition time setting unit 40, “+” or “−” can be selected in addition to “±”, and “year”, “week”, and “day” can be selected in addition to “month”. The same applies to the representative time addition time margin setting unit 41. Therefore, for example, a designation method such as “first inspection date + 1 year ± 1 week” is also possible.

  FIG. 5 is a diagram showing a processing procedure in the second embodiment of the present invention. Here, in order to examine the therapeutic effect of drug administration, it is assumed that the value of the test value A after half a year is examined, and an example in which the value after half a year from the initial test date is taken out will be described.

  In step 500, the medical item “test value A” to be searched is set in the item setting unit 104. In step 501, the representative time point selection unit 102 sets the representative time point name “first inspection”. In step 502, the representative time addition time setting unit 40 sets a representative time addition time “+6 months”, and the representative time addition time margin setting unit 41 sets a representative time addition time margin “± 14 days”.

  Then, the representative time point search unit 110 executes Steps 503 to 507. In Step 503, Step 302 and Step 303 in FIG. 3 are executed from the contents input from the representative time point selection unit 102, and the examination date and time corresponding to each patient is extracted. The extraction results are (ID1,2003 / 06/05) (ID2,2003 / 04/25) (ID3,2003 / 05/06) (ID4,2003 / 08/01) (ID5,2003 / 03/22) Become. In step 504, the date and time obtained by adding “+6 months” input by the representative time addition time setting unit 40 to the result of each inspection date and time in step 503 is calculated. The calculation results are (ID1,2003 / 12/05) (ID2,2003 / 10/25) (ID3,2003 / 11/06) (ID4,2004 / 02/01) (ID5,2003 / 09/22) Become. In step 505, it is determined whether a representative time addition time margin is set. If it is set, step 506 is executed, and if it is not set, step 507 is executed. In this example, since “± 14 days” is set, the process proceeds to step 506. In step 506, the time-series medical data storage unit 121 is searched based on the result of the addition date / time in step 504, and the date / time closest to the addition date / time is 14 days before and after the addition date / time for each patient's examination date / time. Extract. As a result of extraction, ID4 and ID5 have no corresponding date and time, and are (ID1, 2003/12/10) (ID2, 2003/10/22) (ID3, 2003/11/05).

  In step 507, the analysis medical data storage unit 121 is searched based on the result of the addition date and time in step 504, and the date and time closest to the addition date and time is extracted for the examination date and time of each patient.

  Next, the time-series data search unit 111 executes Step 508 and Step 509. In step 508, the time-series medical data storage unit 121 is searched for the medical item “examination value A” input from the item setting unit 104, and a value corresponding to the examination date and time extracted in step 506 or step 507 is set. Extract. The extraction result in this example is (ID1, 2003/12/10, Value3) (ID2, 2003/10/22, Value5) (ID3, 2003/11/05, Value7). In step 509, the extraction result obtained in step 508 is displayed on the output unit 101.

  As described above, in this embodiment, it is possible to easily extract data at an arbitrary observation time point before and after a certain time point as a starting point, thereby improving the efficiency of analysis in the forward / backward investigation.

(Third embodiment)
FIG. 6 is a diagram illustrating the input unit 100 of the input / output device 10 according to the third embodiment of the present invention. The input unit 100 includes an item setting unit 104, a start point input unit 60, and an end point input unit 61. The start point input unit 60 and the end point input unit 61 include the representative time point selection unit 102 and the time information input unit 103 shown in FIG. The Here, the start point indicates the time point that is the start point of the search period in the time series data, and the end point indicates the time point that is the search end point. The start point input unit 60 and the end point input unit 61 may include only the representative time point selection unit 102 in FIG.

  FIG. 7 is a diagram showing a processing procedure in the third embodiment of the present invention. Here, in order to examine the therapeutic effect of drug administration, it is assumed that the change over time of the test value A for six months is examined, and an example in which a value included in a period half a year after the first test date is extracted from the first test date is described.

  In step 700, the medical item “examination value A” to be searched is set in the item setting unit 104. In step 701, the representative time point name “initial examination” is set in the representative time point selection unit 102 in the start point input unit 60. The representative point addition time setting unit 40 and the representative point addition time margin setting unit 41 are not set. In step 702, the representative point name “initial examination” in the representative point selection unit 102 in the end point input unit 61, the representative point addition time “+6 months” in the representative point addition time setting unit 40, and the representative point addition time margin setting unit 41. The representative time addition time margin “± 14 days” is set in FIG.

  Then, the representative time point search unit 110 executes steps 703 and 704. In Step 703, Steps 503 to 507 in FIG. 5 are executed from the contents input from the starting point input unit 60, and the examination date and time corresponding to each patient is extracted. The extraction results are (ID1,2003 / 06/05) (ID2,2003 / 04/25) (ID3,2003 / 05/06) (ID4,2003 / 08/01) (ID5,2003 / 03/22) Become. In Step 704, Steps 503 to 507 in FIG. 5 are executed from the contents input by the end point input unit 60, and the examination date and time corresponding to each patient is extracted. The extraction result is (ID1, 2003/12/10) (ID2, 2003/10/22) (ID3, 2003/11/05).

Next, the time-series data search unit 111 executes Steps 705 and 706.
In step 705, the time-series medical data storage unit 121 is searched for the medical item “examination value A” input from the item setting unit 104, and the starting examination date and time extracted in step 703 and extracted in step 704. A value corresponding to the inspection date and time included between the end point inspection date and time is extracted. Extraction results are (ID1,2003 / 06/05, Value1) (ID1,2003 / 08/06, Value2) (ID1,2003 / 12/10, Value3) (ID2,2003 / 04/25, Value4) (ID2 , 2003/10/22, Value5) (ID3, 2003/05/06, Value6) (ID3, 2003/11/05, Value7). In step 706, the extraction result obtained in step 705 is displayed on the output unit 101.

  In addition, although the said Example demonstrated the case where it searched with a patient unit, the same process sequence is performed also with respect to another search unit. For example, in order to see a tendency such as a side effect for a specific drug, it is necessary to search by drug unit. In that case, a prescription table storing patient prescription information is searched as time-series medical data shown in FIG. In the prescription table, since each drug is administered to a plurality of patients, there are a plurality of administration start dates for each drug ID. Further, the representative time point selection unit 102 displays the representative time point name 21 corresponding to “search unit = drug” in the representative time point definition unit 120 of FIG. Among these, for example, when “first side effect date and time” is selected as the representative time point of each drug, the above-mentioned prescription table is searched from the corresponding inquiry sentence 22, and the administration start with the first side effect for each drug ID is started. Extract date and time.

  From the above, in this embodiment, by defining the representative time point and any observation time points before and after it as the start point and the end point, it is possible to easily extract the time series data between them, and to analyze the temporal variation of the time series data. Can be done efficiently.

The figure which shows the system configuration example of the case search system of this invention. The figure explaining the table structure of the representative time point definition part of 1st Example of this invention. The flowchart which shows the procedure of the process in 1st Example of this invention. The figure which shows the input part of the input / output device in 2nd Example of this invention. The flowchart which shows the procedure of the process in 2nd Example of this invention. The figure which shows the input part of the input / output device in the 3rd Example of this invention. The flowchart which shows the procedure of the process in the 3rd Example of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Input / output device, 11 ... Arithmetic processing device, 12 ... Database, 100 ... Input part, 101 ... Output part, 102 ... Representative time point selection part, 103 ... Time information input part, 104 ... Item setting part, 110 ... Representative time point Retrieval part 111 ... Time series data retrieval part 120 ... Representative time point definition part 121 ... Time series medical data storage part 40 ... Representative time point addition time setting part 41 ... Representative time point addition time margin setting part 60 ... Start point input 61, an end point input unit.

Claims (4)

A medical data storage unit storing a plurality of records including a patient ID, a date and time, and a value of a medical item;
A representative time point defining unit that stores representative time point identifying information representing a representative time point for each search unit, and a processing condition for extracting a date and time corresponding to the representative time point from the medical data storage unit,
A representative time point selection unit for selecting the representative time point according to the representative time point identification information;
An item input unit for inputting medical items to be searched;
Using a processing condition corresponding to the representative time point identification information selected by the representative time point selection unit, a representative time point search unit that extracts the date and time of the representative time point for each search unit from the medical data storage unit;
A case search system comprising: a data search unit for extracting a value of the search medical item of a record corresponding to a date and time extracted by the representative time point search unit from the medical data storage unit. The case search system according to claim 1,
A time information input unit for inputting time information is provided.
The representative time point search unit calculates a date and time by adding the time information input from the time information input unit to the extracted date and time of the representative time point, and from the medical data storage unit, for each search unit, A case search system characterized by extracting the date and time closest to the addition date and time. A medical data storage unit storing a plurality of records including a patient ID, a date and time, and a value of a medical item;
A representative time point defining unit that stores representative time point identifying information representing a representative time point for each search unit, and a processing condition for extracting a date and time corresponding to the representative time point from the medical data storage unit,
An item input unit for inputting medical items to be searched;
A representative time point selection unit for selecting the representative time point based on the representative time point identification information, and a start point input unit for inputting a search start point;
A representative time point selection unit for selecting the representative time point according to the representative time point identification information, and an end point input unit for inputting a search end point;
Using the processing condition corresponding to the representative time point identification information selected by the representative time point selection unit of the start point input unit, the date and time of the representative time point is extracted for each search unit from the medical data storage unit, and the end point input Using a processing condition corresponding to the representative time point identification information selected by the representative time point selection unit, a representative time point search unit that extracts the date and time of the representative time point for each search unit from the medical data storage unit;
The search of records corresponding to all the dates and times included between the start date and time extracted by the representative time point search unit and the end point date and time extracted by the representative time point search unit from the medical data storage unit for each search unit. A case search system comprising a data search unit for extracting a value of a medical item. The case search system according to claim 3,
Each of the start point input unit and the end point input unit has a time information input unit for inputting time information,
The representative time point search unit uses the processing conditions corresponding to the representative time point identification information selected by the representative time point selection unit of the start point input unit, and the date and time of the representative time point for each search unit from the medical data storage unit. And the date and time calculated by adding the time information input from the time information input unit to the extracted date and time of the representative time point is calculated, and the addition date and time is calculated for each search unit from the medical data storage unit. The nearest date and time is extracted, and the representative time point for each search unit is retrieved from the medical data storage unit using processing conditions corresponding to the representative time point identification information selected by the representative time point selection unit of the end point input unit. The date and time of the representative point in time is extracted, and the date and time calculated by adding the time information input from the time information input unit is calculated, and the detection is performed from the medical data storage unit. Case search system and extracts the nearest date to the added time for each unit.

Images