CN109102844A - A kind of clinical test source data automatic Verification method - Google Patents
A kind of clinical test source data automatic Verification method Download PDFInfo
- Publication number
- CN109102844A CN109102844A CN201810970722.8A CN201810970722A CN109102844A CN 109102844 A CN109102844 A CN 109102844A CN 201810970722 A CN201810970722 A CN 201810970722A CN 109102844 A CN109102844 A CN 109102844A
- Authority
- CN
- China
- Prior art keywords
- error correction
- result
- source data
- correction result
- text filed
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H70/00—ICT specially adapted for the handling or processing of medical references
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Epidemiology (AREA)
- Artificial Intelligence (AREA)
- Public Health (AREA)
- Computational Linguistics (AREA)
- Primary Health Care (AREA)
- General Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Character Discrimination (AREA)
Abstract
The present invention discloses a kind of clinical test source data automatic Verification method, comprising steps of using CTPN network model to the source data image recognition of the clinical test of acquisition, determines text filed, then carries out text filed cutting, cut out each style of writing this;To each style of writing this progress upright projection column cutting is cut out, each the effective text filed of this of composing a piece of writing is obtained;Effectively text filed set is sequentially input into housebroken CRNN network, obtains variable length recognition sequence as a result, then extracting text identification result using regular expression;Error correction is carried out to text identification result, obtains error correction result;Characteristic value is extracted from error correction result one by one according to characteristic value collection, is compared with the Standard Eigenvalue recorded in database, the characteristic value to the extraction not being inconsistent with Standard Eigenvalue, indicates alarm status, forms error prompting.The present invention carries out the identification of clinical test source data pictograph by core of CPTN and CRNN, and then realizes the data check of automation.
Description
Technical field
The present invention relates to technical field of data check, and in particular to a kind of clinical test source data automatic Verification method.
Background technique
Clinical test electronic data acquisition system (EDC, Electric Data Capture system) is suitable for drug
The purpose of core information system of clinical test, medicine randomized controlled trial and medicine cohort study, core is for recording
The information of subject forms electronics follow-up list.For clinical test, core is data accuracy, Input Process the most
In human error be to lead to a main cause of data inaccuracy.For this purpose, clinical test implement team needs to assign specially
Data auditor, carry out source data verification: to the original paper data of data source, (laboratory test report, case notes, ICU are guarded
Record/report etc.) manually verified, compare electronical record and source data consistency, referred to as source data verification (SDV,
Source Data Validation).SDV is a committed step for guaranteeing clinical testing data accuracy and the quality of data.
Current EDC system carries out source data verification due to generalling use manual type, there are two aspects:
First, time and effort consuming, auditor needs that height is kept to concentrate, and verifies critical data one by one, and workload is no less than and records again
Enter a pass evidence;Second, not can guarantee quality, test executes team and considers the factors such as cost of labor, time cycle, can not be into
The comprehensive source data verification of row, generallys use the mode of selective examination, can not the general warranty quality of data.
Summary of the invention
In view of the technical drawbacks of the prior art, it is an object of the present invention to provide a kind of clinical test source datas certainly
Dynamic method of calibration, the source data for clinical test electronic data acquisition system verify, and are adopted by carrying out image to former data
Collection, the accuracy of automatic Verification core data (characteristic value) reduce personnel's expense, promote clinical testing data quality.
The technical solution adopted to achieve the purpose of the present invention is:
A kind of clinical test source data automatic Verification method, comprising the following steps:
S1, using CTPN network model to the source data image recognition of the clinical test of acquisition, determine it is text filed, then
Text filed cutting is carried out, cuts out each style of writing originally;To each style of writing this progress upright projection column cutting is cut out, every a line is obtained
Text it is effective text filed;
Effectively text filed set is sequentially input housebroken CRNN network, obtains variable length recognition sequence knot by S2
Then fruit extracts text identification result using regular expression;
S3 carries out error correction to text identification result, obtains error correction result;
S4 extracts characteristic value according to characteristic value collection from error correction result one by one, and special with the standard that records in database
Value indicative compares, the characteristic value to the extraction not being inconsistent with Standard Eigenvalue, indicates alarm status, forms error prompting.
Step S3 to text identification knot error correction, the step of obtaining error correction result, is as follows:
Corresponding Feature Words are searched in characteristic value dictionary using editing distance algorithm, obtain preliminary error correction result;
Judge whether the preliminary error correction result is unique consequence, if so, the preliminary error correction result is determined as final
Error correction result,
Otherwise, to, to each Chinese character string, being used in the preliminary error correction result obtained by editing distance algorithm
The method of character shape coding determines final error correction result.
Using the method for character shape coding, the step of determining error correction result, is as follows:
Character shape coding first is carried out to the Chinese character in the Chinese character string in preliminary error correction result set;
The character shape coding distance for calculating each Chinese character Yu database Plays intercharacter, will be between all Chinese characters
Character shape coding distance is added and obtains the overall distance between two character strings, determines error correction result according to the string overall distance.
In step S1, it is as follows to cut out each this step of of style of writing:
Determine it is text filed after, judge two texts shared by two text filed laps in the vertical direction
The ratio of the total height in region whether be greater than certain threshold value determine two it is text filed whether in a line;If so, being considered as
Two rows, are otherwise considered as a line.
The present invention carries out the identification of clinical test source data pictograph using CPTN and CRNN algorithm as core, and then realizes
The data check of automation realizes the effective integration of artificial intelligence, deep learning and clinical trial informationization.By this hair
It is bright, automation verification can be carried out to clinical test source data, to there may be the undesirable data of problem to carry out certainly
Dynamicization alarm, auditor only needs to check fraction problem data, greatly reduces the workload of data auditor, and can be with
100% verification for guaranteeing clinical trial critical data, can effectively promote the service energy of clinical test electronic data acquisition system
Power and level of intelligence, reduce the expense of clinical test implement team, and guarantee the quality of data.
Detailed description of the invention
Fig. 1 is the workflow schematic diagram of clinical test source data automatic Verification method;
Fig. 2 is CPTN to text filed positioning schematic diagram;
Fig. 3 is the calculating schematic diagram of two text filed overlap ratios;
Fig. 4 is the text filed positioning result schematic diagram of a line;
Fig. 5 is the structural schematic diagram of character shape coding.
Specific embodiment
The present invention is described in further detail below in conjunction with the drawings and specific embodiments.It should be appreciated that described herein
Specific embodiment be only used to explain the present invention, be not intended to limit the present invention.
The present invention includes source data Image Acquisition, character area positioning, Text region, characteristics extraction and logging data ratio
It is that target is determined by character area positioning first to five parts, word content is extracted by Text region algorithm, so
The characteristic value paid close attention to during clinical test, the feature of the result for finally automatically extracting characteristic value and investigator's typing are obtained afterwards
Value mutually compares, and is identified alarm for the typing of mistake.
Referring to shown in Fig. 1-5, a kind of clinical test source data automatic Verification method, comprising:
S1, using CTPN network model to the source data image recognition of the clinical test of acquisition, determine it is text filed, then
Text filed cutting is carried out, cuts out each style of writing originally;To each style of writing this progress upright projection column cutting is cut out, every a line is obtained
Text it is effective text filed;
Wherein, determine it is text filed after, text filed according to any two that detected, first is text filed
100 and second text filed, their lap Y in the vertical directionOverlappingAccount for two text filed total height YAlwaysRatio
Whether example is greater than certain threshold value, such as 20%, come determine two it is text filed whether in a line, finally, accurately cutting out list
Each style of writing sheet in (such as laboratory test report), to realize the positioning of character area.
Specifically, can be the lap Y on vertical direction account for total height ratio be greater than 20%, then be considered as two
Row, is otherwise considered as a line.Carry out home row by this principle, may be implemented accurately to cut out each style of writing sheet in list, thus real
Existing text filed accurate positionin.
The present invention carries out the method that text filed identification is combined with upright projection column cutting by CPTN, carries out random length
Character area positioning, realizes the every a line accurately cut out in list, and obtains each the effective text filed of this of composing a piece of writing.Phase
Method than directly CTPN being used to carry out String localization to source data image, more can effectively improve the accuracy rate of String localization.
Wherein, in step S1 of the invention, the source data image of the clinical test is led to before image recognition
It crosses and takes pictures and obtain, for image obtained by synchronizing and in advance by investigator according to the key in source data input database
Whether correct numerical value matches numerical value typing for subsequent verification.
It should be noted that further including the steps that data preparation and model training before step S1 of the invention.Its
In, the data preparation determines inspection data for preparing:
First, determine the characteristic value dictionary during clinical test, by taking whole blood assay list as an example, wherein may be concerned
Characteristic value includes the crucial laboratory indexes such as red blood cell number, leukocyte count;
Second, determining the feature set of words of this automation verification;
Third, being labeled to the data set during clinical test, data set will cover usual verification certificate, common medical
Report etc. marks the lteral data in (selecting mode to mark using frame) data set.
Prepare with training CTPN and CRNN network model, wherein the CRNN using Sohu's news content training Chinese and English,
The identification of number and conventional sign;The CTPN uses the training pattern of network open source.
S2, effective text filed set that step S1 is obtained, sequentially inputs housebroken CRNN network, can be changed
Then long recognition sequence is as a result, extract text identification using regular expression as a result, to realize the identification of text;
S3 carries out error correction to above-mentioned the obtained text identification result of step, obtains error correction result;
In the present invention, since there may be certain errors for the recognition result that is formed in step S2, error correction side is used
Method carries out error correction to the recognition result that step S2 is formed.
The error correction method can be using one kind of existing error correction method such as editing distance algorithm etc. or character shape coding
Or two methods are in conjunction with realizing.
Specifically, in the present invention, the step S3 to text identification knot error correction, the step of obtaining error correction result, is specific
As follows (method combined using editing distance algorithm with the method for character shape coding):
Similar Feature Words are searched in pre-set characteristic value dictionary using editing distance algorithm, obtain preliminary error correction
As a result;
Judge whether the preliminary error correction result is unique consequence, if so, the preliminary error correction result is determined as final
Error correction result,
Otherwise, to, to each Chinese character string, being used in the preliminary error correction result obtained by editing distance algorithm
The method of character shape coding determines final error correction result.
The editing distance algorithm (or referred to as smallest edit distance algorithm) refers to that a character string is converted into another
Minimum editor (insertion is deleted, replacement) number (i.e. editing distance) of character string, for the existing algorithm for calculating text similarity,
It is no longer described in detail.
It should be noted that in the present invention, using the method for character shape coding, the step of determining error correction result, is specific as follows:
Character shape coding first is carried out to the Chinese character in the Chinese character string in preliminary error correction result;
The character shape coding distance for calculating each Chinese character Yu database Plays intercharacter, will be between all Chinese characters
Character shape coding distance is added and obtains the overall distance between two character strings, determines error correction result according to the string overall distance.
When editing distance algorithm can not uniquely determine final error correction result, a preliminary error correction result set can be obtained.
Therefore, when carrying out character shape coding, in preliminary error correction result set, each Chinese character string is continued to use
Character shape coding method calculates the distance of character shape coding after character shape coding, to determine error correction result.Wherein, character shape coding away from
From the smallest for final error correction result.
Specifically, using method shown in Fig. 5 to encoding of chinese characters, and calculating individual character font in the present invention using following formula (1) and compiling
Code distance.
Wherein, S indicates the distance between two Chinese characters, siIndicate corresponding i-th of bits of coded, i is less than or equal to 6
Natural number, Δ siIndicate the absolute value that bits of coded difference is corresponded between two Chinese characters, s6Indicate the pen of a Chinese character
Draw number, s '6Indicate the stroke number of another Chinese character.
Calculate in order each Chinese character in preliminary error correction result in Chinese character string corresponding in database
After distance between Chinese character, the distance of the Chinese character of all calculating is added summation, two Chinese character strings can be obtained
Overall distance be determined as final error correction result apart from the smallest.
The character shape coding of the Chinese character is as shown in figure 5, include structured coding, quadrangle coding and stroke number.
The cataloged procedure of Chinese character therein is as follows:
Structured coding: first is set as constructive code (S1) in character shape coding, indicates the structure of Chinese character, such as left and right knot
Structure, up-down structure etc., specific coding situation are as shown in table 1;
Quadrangle coding: the second of character shape coding saves the quadrangle coding of the Chinese character to the 5th (S2-S5), this is encoded in
Nineteen twenty-five is invented by Wang Yunwu, is encoded using number to basic stroke and multiple pen, is successively extracted the Chinese character upper left corner, upper right
Angle, the lower left corner, the lower right corner feature form final coding.The coding can be inquired by four-corner system dictionary;
Stroke number.Last of character shape coding is the stroke number (S6) of the Chinese character, may be one or two.
Word is encoded, so that character shape coding dictionary can be obtained, can be obtained in corresponding by consulting character shape coding dictionary
The character shape coding of Chinese character.
By taking " blood " word as an example, graphemic code be " 027106 ", totally 6.
Table 1
After carrying out recognition result error correction, recognition accuracy is further increased, and table 2 show Text region result and error correction knot
Fruit example prepares for subsequent characteristics extraction with comparison.
Table 2
In the present invention, by combining two methods of editing distance and character shape coding to carry out error correction to recognition result, and then seek
Best match characteristic value is looked for, more accurate carry out error correction may be implemented, the problem of single error correction method inaccuracy is overcome, mentions
The accuracy rate of error correction is risen.
S4 extracts characteristic value according to pre-set characteristic value collection from error correction result one by one, and with remember in database
The Standard Eigenvalue comparing of the preparatory typing of record, the characteristic value to the extraction not being inconsistent with standard feature Value Data, mark are accused
Alert state, forms error prompting, afterwards by manually being verified.
It should be noted that being according to predetermined characteristic value collection, one by one from Text region in characteristics extraction
As a result feature is extracted in, if entering the step of comparing characteristic value with the presence of characteristic value in Text region result, otherwise determining should
The recognition failures of characteristic value.
Wherein, in the step of comparing characteristic value, be using the characteristic value data of preparatory institute typing in database, and it is automatic
The characteristic value extracted from Text region result, if extraction value indicative and the characteristic value data entered be not identical, is transferred to life compared to
The step of at error prompting, otherwise terminate.
Experimental comparison:
Whole blood assay list, ICU report and liver function test list etc. are had chosen respectively amounts to 300 test samples, every class 100
A, 10 characteristic values of each class declaration amount to 3000 characteristic values, and the characteristic value of 300 samples correctly identifies that number is respectively
999,996 and 997, average recognition accuracy is up to 99.73%.
The above is only a preferred embodiment of the present invention, it is noted that for the common skill of the art
For art personnel, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications
Also it should be regarded as protection scope of the present invention.
Claims (4)
1. a kind of clinical test source data automatic Verification method, which comprises the following steps:
S1 determines text filed using CTPN network model to the source data image recognition of the clinical test of acquisition, then carries out
Text filed cutting cuts out each style of writing originally;To each style of writing this progress upright projection column cutting is cut out, each style of writing is obtained originally
It is effective text filed;
Effectively text filed set is sequentially input housebroken CRNN network by S2, obtains variable length recognition sequence as a result, so
Text identification result is extracted using regular expression afterwards;
S3 carries out error correction to text identification result, obtains error correction result;
S4 extracts characteristic value according to characteristic value collection from error correction result one by one, and with the Standard Eigenvalue that is recorded in database
It compares, the characteristic value to the extraction not being inconsistent with Standard Eigenvalue, indicates alarm status, form error prompting.
2. clinical test source data automatic Verification method as described in claim 1, which is characterized in that step S3 to text identification
The step of tying error correction, obtaining error correction result is as follows:
Corresponding Feature Words are searched in characteristic value dictionary using editing distance algorithm, obtain preliminary error correction result;
Judge whether the preliminary error correction result is unique consequence, if so, the preliminary error correction result is determined as final error correction
As a result,
Otherwise, to, to each Chinese character string, using font in the preliminary error correction result obtained by editing distance algorithm
The method of coding determines final error correction result.
3. clinical test source data automatic Verification method as claimed in claim 2, which is characterized in that use the side of character shape coding
Method, the step of determining error correction result, are as follows:
Character shape coding first is carried out to the Chinese character in the Chinese character string in preliminary error correction result set;
The character shape coding distance for calculating each Chinese character Yu database Plays intercharacter, by the font between all Chinese characters
Coding distance is added and obtains the overall distance between two character strings, determines error correction result according to the string overall distance.
4. clinical test source data automatic Verification method as described in claim 1, which is characterized in that in step S1, cut out each
This step of of composing a piece of writing, is as follows:
Determine it is text filed after, judge shared by two text filed laps in the vertical direction two it is text filed
Total height ratio whether be greater than certain threshold value determine two it is text filed whether in a line;If so, it is considered as two rows,
Otherwise it is considered as a line.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810970722.8A CN109102844B (en) | 2018-08-24 | 2018-08-24 | Automatic calibration method for clinical test source data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810970722.8A CN109102844B (en) | 2018-08-24 | 2018-08-24 | Automatic calibration method for clinical test source data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109102844A true CN109102844A (en) | 2018-12-28 |
CN109102844B CN109102844B (en) | 2022-02-15 |
Family
ID=64851220
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810970722.8A Active CN109102844B (en) | 2018-08-24 | 2018-08-24 | Automatic calibration method for clinical test source data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109102844B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109977942A (en) * | 2019-02-02 | 2019-07-05 | 浙江工业大学 | A kind of scene character recognition method based on scene classification and super-resolution |
CN110196917A (en) * | 2019-05-30 | 2019-09-03 | 厦门一品威客网络科技股份有限公司 | Personalized LOGO format method for customizing, system and storage medium |
CN110298343A (en) * | 2019-07-02 | 2019-10-01 | 哈尔滨理工大学 | A kind of hand-written blackboard writing on the blackboard recognition methods |
CN111368693A (en) * | 2020-02-28 | 2020-07-03 | 中国建设银行股份有限公司 | Identification method and device for identity card information |
CN111368697A (en) * | 2020-02-28 | 2020-07-03 | 中国建设银行股份有限公司 | Information identification method and device |
CN111681724A (en) * | 2020-05-07 | 2020-09-18 | 浙江大学医学院附属第四医院(浙江省义乌医院、浙江大学医学院附属第四医院医共体) | Electronic medical record key entity standardized identification method and identification system |
WO2020208632A1 (en) * | 2019-04-10 | 2020-10-15 | Beacon Cure Ltd. | System and method for validating tabular summary reports |
CN112397159A (en) * | 2019-08-19 | 2021-02-23 | 金色熊猫有限公司 | Automatic clinical test report input method and device, electronic equipment and storage medium |
CN116682519A (en) * | 2023-08-03 | 2023-09-01 | 广东杰纳医药科技有限公司 | Clinical experiment data unit analysis method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103425975A (en) * | 2013-07-17 | 2013-12-04 | 中国中医科学院 | System and method for acquiring clinical case data |
CN107346420A (en) * | 2017-06-19 | 2017-11-14 | 中国科学院信息工程研究所 | Text detection localization method under a kind of natural scene based on deep learning |
CN107633250A (en) * | 2017-09-11 | 2018-01-26 | 畅捷通信息技术股份有限公司 | A kind of Text region error correction method, error correction system and computer installation |
CN107862303A (en) * | 2017-11-30 | 2018-03-30 | 平安科技(深圳)有限公司 | Information identifying method, electronic installation and the readable storage medium storing program for executing of form class diagram picture |
-
2018
- 2018-08-24 CN CN201810970722.8A patent/CN109102844B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103425975A (en) * | 2013-07-17 | 2013-12-04 | 中国中医科学院 | System and method for acquiring clinical case data |
CN107346420A (en) * | 2017-06-19 | 2017-11-14 | 中国科学院信息工程研究所 | Text detection localization method under a kind of natural scene based on deep learning |
CN107633250A (en) * | 2017-09-11 | 2018-01-26 | 畅捷通信息技术股份有限公司 | A kind of Text region error correction method, error correction system and computer installation |
CN107862303A (en) * | 2017-11-30 | 2018-03-30 | 平安科技(深圳)有限公司 | Information identifying method, electronic installation and the readable storage medium storing program for executing of form class diagram picture |
Non-Patent Citations (1)
Title |
---|
浦世亮等: "低分辨率自然场景文本识别", 《中国安防》 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109977942A (en) * | 2019-02-02 | 2019-07-05 | 浙江工业大学 | A kind of scene character recognition method based on scene classification and super-resolution |
WO2020208632A1 (en) * | 2019-04-10 | 2020-10-15 | Beacon Cure Ltd. | System and method for validating tabular summary reports |
CN110196917A (en) * | 2019-05-30 | 2019-09-03 | 厦门一品威客网络科技股份有限公司 | Personalized LOGO format method for customizing, system and storage medium |
CN110298343A (en) * | 2019-07-02 | 2019-10-01 | 哈尔滨理工大学 | A kind of hand-written blackboard writing on the blackboard recognition methods |
CN112397159A (en) * | 2019-08-19 | 2021-02-23 | 金色熊猫有限公司 | Automatic clinical test report input method and device, electronic equipment and storage medium |
CN112397159B (en) * | 2019-08-19 | 2024-03-22 | 金色熊猫有限公司 | Automatic entry method and device for clinical test report, electronic equipment and storage medium |
CN111368693A (en) * | 2020-02-28 | 2020-07-03 | 中国建设银行股份有限公司 | Identification method and device for identity card information |
CN111368697A (en) * | 2020-02-28 | 2020-07-03 | 中国建设银行股份有限公司 | Information identification method and device |
CN111681724A (en) * | 2020-05-07 | 2020-09-18 | 浙江大学医学院附属第四医院(浙江省义乌医院、浙江大学医学院附属第四医院医共体) | Electronic medical record key entity standardized identification method and identification system |
CN116682519A (en) * | 2023-08-03 | 2023-09-01 | 广东杰纳医药科技有限公司 | Clinical experiment data unit analysis method |
CN116682519B (en) * | 2023-08-03 | 2024-03-19 | 广东杰纳医药科技有限公司 | Clinical experiment data unit analysis method |
Also Published As
Publication number | Publication date |
---|---|
CN109102844B (en) | 2022-02-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109102844A (en) | A kind of clinical test source data automatic Verification method | |
WO2017075957A1 (en) | Recognition rate determining method and device | |
WO2021043085A1 (en) | Method and apparatus for recognizing named entity, computer device, and storage medium | |
CN105468468B (en) | Data error-correcting method towards question answering system and device | |
CN103631966B (en) | A kind of method of configurable parsing multivalue matching field | |
CN109344250A (en) | Single diseases diagnostic message rapid structure method based on medical insurance data | |
CN106682397A (en) | Knowledge-based electronic medical record quality control method | |
CN106844351B (en) | Medical institution organization entity identification method and device oriented to multiple data sources | |
CN112257613B (en) | Physical examination report information structured extraction method and device and computer equipment | |
CN111144191A (en) | Font identification method and device, electronic equipment and storage medium | |
CN109993709B (en) | Image registration error correction method based on deep learning | |
CN109063713A (en) | A kind of timber discrimination method and system based on the study of construction feature picture depth | |
CN108170468A (en) | The method and its system of a kind of automatic detection annotation and code consistency | |
CN109992770A (en) | A kind of Laotian name entity recognition method based on combination neural net | |
CN110472209A (en) | Table generation method, device and computer equipment based on deep learning | |
CN115510330B (en) | Intelligent information processing method and system based on data mining | |
CN110634546A (en) | Electronic medical record text standardization detection method | |
Jiang et al. | Reo-relevance, extraness, omission: A fine-grained evaluation for image captioning | |
CN107590119A (en) | Character attribute information extraction method and device | |
CN107908601A (en) | Participle model construction method, equipment, readable storage medium storing program for executing and the segmenting method of medical text | |
CN116206319B (en) | Data processing system for clinical trials | |
CN110442729A (en) | A kind of Corpus System construction method based on rough collection | |
CN112015871B (en) | Automatic character relation labeling method based on event set remote supervision | |
CN113642563A (en) | Drug use rechecking method, device, equipment and storage medium | |
CN109960707B (en) | College recruitment data acquisition method and system based on artificial intelligence |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |