阅读说明：本技术 医疗数据处理设备、系统和方法 (Medical data processing apparatus, system and method ) 是由 张鹏 于 2020-05-19 设计创作，主要内容包括：本发明提供了一种医疗数据处理设备、系统和方法。所述设备包括获取模块,其被配置成获取参考医疗数据集和医嘱界面图像,所述参考医疗数据集包括包含多个参考患者标识的参考标识集和包含多个参考医嘱名称的参考医嘱名称集；处理模块,其被配置成确定所述多个参考医嘱名称中的每一个的权重配置；从所述医嘱界面图像中识别出患者标识和医嘱名称；在确定为识别出的患者标识与所述多个参考患者标识中的一个相匹配时,基于最接近的参考医嘱名称的权重配置计算识别出的医嘱名称的各个字的权重之和；并且基于所述权重之和确定对医嘱名称的识别结果是否有效；以及输出模块,其被配置成输出的医嘱名称的有效识别结果以及相应的患者标识。(The invention provides a medical data processing device, a system and a method. The device includes an acquisition module configured to acquire a reference medical data set and an order interface image, the reference medical data set including a reference identifier set including a plurality of reference patient identifiers and a reference order name set including a plurality of reference order names; a processing module configured to determine a weight configuration for each of the plurality of reference order names; identifying a patient identification and an order name from the order interface image; upon determining that the identified patient identification matches one of the plurality of reference patient identifications, calculating a sum of weights for each word of the identified order name based on a weight configuration of the closest reference order name; and determining whether the identification result of the order name is valid or not based on the sum of the weights; and an output module configured to output a valid identification of the order name and a corresponding patient identification.)

1. A medical data processing apparatus comprising:

an acquisition module configured to acquire a reference medical data set and an order interface image of a medical terminal, the reference medical data set including a reference identifier set including a plurality of reference patient identifiers and a reference order name set including a plurality of reference order names;

a processing module configured to determine a weight configuration for each of the plurality of reference order names, the weight configuration including a weight for each word in a reference order name; identifying a patient identification and an order name from the order interface image; upon determining that the identified patient identification matches one of the plurality of reference patient identifications, calculating a sum of weights for words of the identified order name based on a weight configuration of a reference order name of the plurality of reference order names that is closest to the identified order name; and determining whether the identification result of the order name is valid or not based on the sum of the weights; and

an output module configured to output a valid recognition result of the order name and a corresponding patient identification.

2. The medical data processing apparatus according to claim 1, wherein the weight configuration contains a weight for each word in a reference order name indicating its information content proportion in the reference order name, and the sum of the weights of all words of a reference order is 1.

3. The medical data processing device of claim 1 or 2, wherein the processing module is further configured to determine the closest reference order name by:

calculating a plurality of edit distances between the identified order name and each of the plurality of reference order names; and

determining the reference order name corresponding to the smallest edit distance of the plurality of edit distances as the closest reference order name.

4. The medical data processing device of claim 3, wherein the processing module calculates a sum of the weights of the words of the identified order name by:

determining edit types of respective words of the identified order name relative to corresponding words of the closest reference order name, the edit types including equal and unequal;

setting the weight of the words with the editing types being equal to a positive value of the weight of the corresponding word, and setting the weight of the words with the editing types being unequal to a negative value of the weight of the corresponding word; and

and adding the weight values of the words set according to the edit type to obtain the sum of the weights.

5. The medical data processing device of any of claims 1-4, wherein the processing module is configured to:

determining that the identification result of the order name is valid when the sum of the weights is greater than zero, and determining that the identification result of the order name is invalid when the sum of the weights is less than or equal to zero; or

When the sum of the weights is greater than or equal to zero, determining that the recognition result of the order name is valid, and when the sum of the weights is less than zero, determining that the recognition result of the order name is invalid.

6. The medical data processing device of any of claims 1-5, wherein the processing module is configured to identify an order name by:

extracting a unit table in the medical advice interface image;

determining the most-listed unit table in the unit tables as an order table containing an order name; and

performing OCR recognition on the order table to obtain an order name therein.

7. The medical data processing device of any of claims 1-6, wherein the processing module is configured to identify the patient identification by:

extracting a region of interest containing patient identification in the order interface image; and

performing OCR recognition on the region of interest to obtain patient identification.

8. The medical data processing device of claim 7, wherein the processing module is further configured to:

calculating a plurality of edit distances of the identified patient identifier from each of the plurality of reference patient identifiers;

comparing a minimum edit distance of the calculated plurality of edit distances with a predetermined edit distance threshold;

determining that the identified patient identifier matches one of the plurality of reference patient identifiers when the minimum edit distance is not greater than the predetermined edit distance threshold; and is

And when the minimum editing distance is larger than the preset editing distance threshold value, determining that the identified patient identifier is not matched with any one of the plurality of reference patient identifiers, and ending the medical data processing.

9. The medical data processing device of any of claims 1-8, wherein the patient identification comprises a patient name, optionally a patient visit number and/or an out-patient number; and is

The order name includes at least one of: drug name, injection name, inspection/examination item name, operation name.

10. A medical data processing system comprising:

the system comprises one or more capturing modules, a processing module and a display module, wherein each capturing module is arranged on a medical terminal and is used for capturing a doctor's advice interface image of a medical information system in the medical terminal;

the database is arranged in a medical terminal server which is in communication connection with the medical terminal and is used for storing a reference medical data set; and

the medical data processing device of any one of claims 1-9, configured to identify an order name from an order interface image based on the reference medical data set and to output a valid identification of the order name.

11. A method of medical data processing, optionally performed by a medical data processing apparatus according to any of claims 1-9 and/or a medical data processing system according to claim 10, the method comprising:

acquiring a reference medical data set and an order interface image of a medical terminal, wherein the reference medical data set comprises a reference identification set containing a plurality of reference patient identifications and a reference order name set containing a plurality of reference order names;

determining a weight configuration for each of the plurality of reference order names, the weight configuration including a weight for each word in a reference order name;

identifying a patient identification and an order name from the order interface image;

upon determining that the identified patient identification matches one of the plurality of reference patient identifications, calculating a sum of weights for words of the identified order name based on a weight configuration of a reference order name of the plurality of reference order names that is closest to the identified order name;

determining whether the identification result of the medical advice name is valid or not based on the sum of the weights; and

and outputting a valid identification result of the name of the order and a corresponding patient identification.

12. A machine-readable storage medium having stored thereon executable instructions, wherein the executable instructions, when executed, cause a machine to perform the method of claim 11.

Technical Field

The invention relates to a medical data processing device, system and method.

Background

As various medical information systems are increasingly used, there is a need to interact with medical data between the medical information systems. For this reason, it is necessary to obtain truly accurate medical data.

One existing solution is to manually enter medical data in one medical information system (e.g., HIS) into another medical information system, which is problematic in that manual entry is erroneous and labor intensive.

Another existing solution is to employ an electronic device to analyze and identify documents containing medical data to obtain the medical data. Such a scheme, although not requiring manual entry, has a problem of low accuracy because such a recognition scheme has a high recognition rate for common languages of printed matter, but has a low recognition rate for medical data containing, for example, medical terms, "characters + punctuation".

Yet another conventional solution is to perform model training or transfer learning on medical data such as medical terms, "words and punctuations", but such a solution has a high computational power requirement on computing equipment, a long training and debugging period, and may still not achieve a satisfactory effect.

Therefore, it is desirable to provide a solution to the above-mentioned problems in the prior art.

Disclosure of Invention

In view of the above problems in the prior art, the present invention aims to provide an improved medical data processing technical solution, which can automatically obtain medical data with high accuracy.

To this end, according to one aspect of the invention, a medical data processing device is proposed, comprising: an acquisition module configured to acquire a reference medical data set and an order interface image of a medical terminal, the reference medical data set including a reference identifier set including a plurality of reference patient identifiers and a reference order name set including a plurality of reference order names; a processing module configured to determine a weight configuration for each of the plurality of reference order names, the weight configuration including a weight for each word in a reference order name; identifying a patient identification and an order name from the order interface image; upon determining that the identified patient identification matches one of the plurality of reference patient identifications, calculating a sum of weights for words of the identified order name based on a weight configuration of a reference order name of the plurality of reference order names that is closest to the identified order name; and determining whether the identification result of the order name is valid or not based on the sum of the weights; and an output module configured to output a valid identification of the order name and a corresponding patient identification.

According to a possible embodiment, the weight arrangement contains a weight for each word in a reference order name, which weight represents the information content thereof in the reference order name, and the sum of the weights of all words of a reference order is 1.

According to a possible embodiment, the processing module is further configured to determine the closest reference order name by: calculating a plurality of edit distances between the identified order name and each of the plurality of reference order names; and determining the reference order name corresponding to the minimum editing distance in the plurality of editing distances as the closest reference order name.

According to one possible embodiment, the processing module calculates the sum of the weights of the words of the identified order name by: determining edit types of respective words of the identified order name relative to corresponding words of the closest reference order name, the edit types including equal and unequal; setting the weight of the words with the editing types being equal to a positive value of the weight of the corresponding word, and setting the weight of the words with the editing types being unequal to a negative value of the weight of the corresponding word; and adding the weight values of the words set according to the edit type to obtain the sum of the weights.

According to a possible embodiment, the processing module is configured to: determining that the identification result of the order name is valid when the sum of the weights is greater than zero, and determining that the identification result of the order name is invalid when the sum of the weights is less than or equal to zero; or when the sum of the weights is greater than or equal to zero, determining that the identification result of the order name is valid, and when the sum of the weights is less than zero, determining that the identification result of the order name is invalid.

According to a possible embodiment, the processing module is configured to identify the order name by: extracting a unit table in the medical advice interface image; determining the most-listed unit table in the unit tables as an order table containing an order name; and performing OCR recognition on the order table to obtain the name of the order in the order table.

According to one possible embodiment, the processing module is configured to identify the patient identification by: extracting a region of interest containing patient identification in the order interface image; and performing OCR recognition on the region of interest to obtain the patient identification.

According to a possible embodiment the processing module is further configured to: calculating a plurality of edit distances of the identified patient identifier from each of the plurality of reference patient identifiers; comparing a minimum edit distance of the calculated plurality of edit distances with a predetermined edit distance threshold; determining that the identified patient identifier matches one of the plurality of reference patient identifiers when the minimum edit distance is not greater than the predetermined edit distance threshold; and when the minimum edit distance is greater than the predetermined edit distance threshold, determining that the identified patient identifier does not match any of the plurality of reference patient identifiers, and ending the medical data processing.

According to a possible embodiment, the patient identification comprises a patient name and, optionally, a patient visit serial number and/or an outpatient number; and the order name comprises at least one of: drug name, injection name, inspection/examination item name, operation name.

According to another aspect of the present invention, there is provided a medical data processing system comprising: the system comprises one or more capturing modules, a processing module and a display module, wherein each capturing module is arranged on a medical terminal and is used for capturing a doctor's advice interface image of a medical information system in the medical terminal; the database is arranged in a medical terminal server which is in communication connection with the medical terminal and is used for storing a reference medical data set; and a medical data processing apparatus as described above for recognizing an order name from the order interface image based on the reference medical data set and outputting a valid recognition result of the order name.

According to a further aspect of the invention, a method of medical data processing is proposed, optionally performed by a medical data processing apparatus as described above and/or a medical data processing system as described above, the method comprising: acquiring a reference medical data set and an order interface image of a medical terminal, wherein the reference medical data set comprises a reference identification set containing a plurality of reference patient identifications and a reference order name set containing a plurality of reference order names; determining a weight configuration for each of the plurality of reference order names, the weight configuration including a weight for each word in a reference order name; identifying a patient identification and an order name from the order interface image; upon determining that the identified patient identification matches one of the plurality of reference patient identifications, calculating a sum of weights for words of the identified order name based on a weight configuration of a reference order name of the plurality of reference order names that is closest to the identified order name; determining whether the identification result of the medical advice name is valid or not based on the sum of the weights; and outputting a valid recognition result of the name of the order and a corresponding patient identifier.

According to yet another aspect of the invention, a machine-readable storage medium is provided, having executable instructions stored thereon, wherein the executable instructions, when executed, cause a machine to perform the method as described above.

According to the technical scheme of the invention, under the condition of no need of special training and transfer learning, the high-efficiency processing of the medical data is realized by simple operation, so that the medical data with high accuracy can be obtained from the interface image of the medical information system, and the operation cost and the time cost are saved. Moreover, according to the technical scheme of the invention, the original data used for medical data processing is from the real medical behavior data directly captured at the medical terminal, so that the medical data processing result of the invention has objectivity.

Drawings

Fig. 1 is a schematic block diagram of a medical data processing system according to an embodiment of the present invention.

Fig. 2 is a schematic block diagram of a medical data processing apparatus according to an embodiment of the present invention.

Fig. 3 is a schematic illustration of a medical data processing technique according to the present invention.

FIG. 4 is a schematic illustration of an order interface image from which medical data is extracted.

Fig. 5 is a schematic diagram of a patient identification recognition process according to one embodiment of the present invention.

FIG. 6 is a schematic diagram of an order identification and weight calculation process according to one embodiment of the invention.

Fig. 7 is a flowchart of a medical data processing method according to an embodiment of the present invention.

Detailed Description

The invention relates to a medical data processing scheme under the condition of computer application.

In the present invention, "medical data" refers to medical data based on computer applications, that is, medical data that can be captured by a medical information system operation. The medical data may contain information that uniquely identifies the patient identification and the name of the order.

In the present invention, the "reference order name set" refers to those accurate (standard) sets of order names. The set of reference order names may be stored in a database of the medical terminal server. The reference order name in the reference order name set may be regarded as reference data for identification and determination of the order name.

In the present invention, a "reference patient identification set" refers to a set of those accurate (standard) patient identifications. The set of reference patient identifications may be stored in a database of the medical terminal server. The reference patient identifier in the set of reference patient identifiers may be considered as baseline data for the identification and judgment of the patient identifier.

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Fig. 1 schematically illustrates a medical data processing system 100 according to an embodiment of the invention, which mainly comprises one or more capture modules 10, 20, 30; a database 40; and a medical data processing device 50.

Each of the capturing modules 10, 20, 30 is disposed in a medical terminal. For example, each of the capture modules 10, 20, 30 is provided in one of the medical terminals 1, 2, 3, respectively. Although 3 capture modules 10, 20, 30 are illustrated in FIG. 1, the number of capture modules is not limited by the present invention. Medical data processing system 100 may include one capture module, or may include other numbers of capture modules. The capturing module is used for capturing an order interface image on the medical terminal, namely the order interface image is an image of an order interface of a medical information system comprising the medical terminal. The manner in which the interface image is captured is not limiting.

The medical terminals 1, 2, 3 can be understood as client computers, for example computers used by doctors of hospitals, medical institutions, health centers, medical testing or diagnosis centers when attending to patients. One or more medical information systems are provided in the medical terminal, such as a Hospital Information System (HIS), a computerized physician order entry system (CPOE), a Clinical Decision Support System (CDSS), and so forth. The capturing module can be applied to various medical information systems in the medical terminal and captures medical advice interface images of the medical information systems arranged in the medical terminal.

The database 40 is provided in the medical terminal server 4. The database 40 stores therein a reference medical data set used as a reference, for example, a reference patient identification set and a reference order name set. The medical terminal server is, for example, a server of a hospital. The server (i.e., medical terminal server) is connected to one or more computers for doctors (i.e., medical terminals), and can communicate with the computers for the respective doctors.

The medical data processing apparatus 50 is provided in the server computer 5. The medical data processing apparatus 50 is used to execute a strategy of processing medical data, that is, calculating a weight sum of the identified order name by means of the reference medical data set and based on the weight configuration for the reference order name, and determining whether the identification result for the order name is valid based on the weight sum, and outputting only the valid identification result.

The "valid recognition result of the order name" may be understood as a recognition accuracy reaching a predetermined criterion, i.e. the recognized order name is available, for example, for subsequent operation of the medical information system or for a medical evaluation system by means of the medical information system.

The medical data processing device 50 may be implemented in software or hardware or a combination of software and hardware. As shown in fig. 2, the medical data processing apparatus 50 mainly includes an acquisition module 51, a processing module 52, and an output module 53. These modules are, for example, software modules, the principle of which will be described in more detail below.

It is to be understood that the naming of the various modules 51-53 of the medical data processing device 50 should be understood as a logical description, not a limitation of physical form or arrangement. In other words, a medical data processing device should be understood to comprise a module as long as the medical data processing device has the functionality of the module.

It will be appreciated that the various modules 51-53 of the medical data processing device 50 may be implemented in a variety of ways. These modules may be implemented as hardware, software, or a combination thereof. In addition, any of these modules may be further divided functionally into sub-modules or combined together.

As can be seen, the medical data processing system 100 according to the present invention captures an order interface image at a client (i.e., a medical terminal), extracts medical data in the order interface image at a computer server, and outputs medical data with high accuracy. The high accuracy medical data can be used by a medical information system (e.g., a clinical information analysis system). The medical data processing system of the invention realizes the capture of the medical advice interface image and the medical data extraction in independent devices respectively, and has extremely high flexibility.

The operation and principle of the medical data processing device 50 and its various modules will be described below with reference to fig. 2 and 3.

The acquisition module 51 may acquire a reference medical data set. The acquisition module 51 may acquire the reference medical data set stored in the database 40 of the medical terminal server 4 periodically (e.g., every day). The obtaining module 51 may obtain the updated reference medical data set after the reference medical data set in the database 40 is updated.

The reference medical data set may include a reference patient identification set and a reference order name set.

The reference patient identification set contains a plurality of reference patient identifications for uniquely identifying the patient. A reference patient identification may include the name of the patient and may further include a serial number for the patient's visit and/or an out-patient number.

The reference order name set includes a plurality of reference order names for representing order names in medical orders placed by physicians during a medical activity. These reference order names may be periodically stored in the medical terminal server by the administrator of the medical terminal. A reference order name may include names such as medication name, infusion name, etc. in the items of the order, see the order name below the "item name" icon in fig. 4.

The acquisition module 51 also acquires an order interface image. After the capturing module 10 captures an order interface image (i.e., an image containing an order interface) in the medical information system of the medical terminal 1, the order interface image is transmitted to the medical data processing apparatus 50, for example, the acquisition module 51 of the medical data processing apparatus 50.

Therefore, according to the invention, the raw data of the medical data processing is from the real medical behavior data directly captured by the capture module arranged at the medical terminal. Therefore, the original data is data which is not processed by transmission and derivation, and has objectivity and timeliness.

The processing module 52 performs the following processing on the order interface image based on the reference medical data set: weight configuration, patient identification, medical advice identification, weight calculation and determination of an effective identification result. These processes are specifically described below.

The processing module 52 weights each of the acquired reference order names in the set of reference order names. In a weight arrangement of a reference order name, weights indicating the information amount weight thereof in the order name are set for respective words in the order name, and the sum of the weights of the respective words of an order name is 1. The word in the name of the order should be understood to include Chinese characters, letters, numbers, symbols, and the like.

The "amount of information" of a word in an order name may be understood as a measure of the information conveyed by the word. For example, in the order name "cinobufagin capsule", the information amount of "capsule" is lower than "cinobufagin", and accordingly the weight of each word in "cinobufagin" is higher than the weight of each word in "capsule".

In one embodiment, the processing module 52 may calculate a weight matrix of each word in an order name by using a Term Frequency-Inverse text Frequency index (TF-IDF) manner, so as to obtain the above weight configuration.

Referring to fig. 4, order interface image 400 may include an area 410 that presents patient identification information and an area 420 that presents order information. The information in both areas 410 and 420 may be presented in the form of a table. It is understood that interface elements (not shown) representing other information may also be included on the order interface.

The processing module 52 identifies the order interface image 400 to extract the patient identification therein, i.e., the identified patient identification, and determines whether the identified patient identification can match one of the plurality of reference patient identifications in the reference patient identification set. In other words, it is possible by this process to determine whether the patient represented by the identified patient identification is the patient being treated at the medical terminal.

Fig. 5 illustrates one exemplary process 500 for performing patient identification recognition.

Referring to fig. 5, in block 502, the processing module 52 extracts a region of interest (ROI region) in the order interface image 400. The region of interest may be understood as a region for displaying patient identification according to the specifications (habits) of the medical information system at the medical terminal, for example, the upper left quadrant region of the overall order interface.

In block 504, the processing module 52 identifies an image of the ROI region for identified patient identification.

In one embodiment, the processing module 52 may convert the image of the ROI region to a gray map and extract edge information of the gray map (e.g., using the canny operator). Then, the edge information is subjected to an opening and closing operation, for example, expansion (for example, expansion by 9 × 9 kernel), erosion (for example, erosion by 12 × 12 kernel), and re-expansion (for example, re-expansion by 4 × 4 kernel) are sequentially performed to remove "lines" in the image and leave only "characters" and "icons". Then, rectangular approximation is performed on the image after the opening and closing operation to obtain a text region set containing "noise". Then, the set of text regions is subjected to Optical Character Recognition (OCR) to obtain an OCR Recognition result, thereby obtaining the recognized patient identifier.

In block 506, the processing module 52 determines the reference patient identifier in the reference patient identifier set that is closest to the identified patient identifier.

In one embodiment, the processing module 52 calculates the edit distance between the identified patient identifier and each reference patient identifier in the reference patient identifier set, determines the smallest edit distance among the edit distances, and uses the reference patient identifier corresponding to the smallest edit distance as the closest reference patient identifier.

In block 508, the processing module 52 determines whether the identified patient identifier can match one of the plurality of reference patient identifiers in the set of reference patient identifiers by determining whether the minimum edit distance satisfies a predetermined edit distance threshold.

In one embodiment, the edit distance threshold may be preset, for example set to 2, i.e. the identified patient identifier may be converted to a reference patient identifier by 2 transformations. If the minimum edit distance is less than or equal to 2, it is determined that the identified patient identifier matches a reference patient identifier in the set of reference patient identifiers. If the minimum edit distance is greater than 2, it is determined that the identified patient identifier does not match any reference patient identifier in the reference patient identifier set, that is, the patient identifier may be considered as an error, and the medical data processing is ended.

In this way, by setting a reasonable edit distance threshold, it can be ensured that there is only a limited difference between the identified patient identification (e.g., name, etc.) and the accurate patient identification, e.g., only one or two words are different. This can improve the accuracy of medical data processing.

The processing module 52 identifies the order interface image 400 to extract the order name therein, that is, obtain the identified order name, and calculates the sum of the weights of the words of the order name. The order name may be understood as the name of the order in the medical order given by the physician during the medical activity, for example, the name of a medication, the name of an injection, the name of an examination/examination, the name of a surgery, etc. The name of the order illustrated in fig. 4 includes a medicine name and an injection name under the "item name".

Fig. 6 illustrates one exemplary process 600 for performing order name identification and weight calculation.

Referring to FIG. 6, in block 602, the processing module 52 obtains a cell table in the order interface image 400.

In one embodiment, the processing module 52 converts the order interface image 400 to a gray scale map and extracts edge information in the gray scale map (e.g., using the canny operator). Then, the edge information is opened and closed, for example, expansion (for example, expansion by 3 × 3 nucleus) and corrosion (for example, corrosion by 2 × 2 nucleus) are sequentially performed to merge and close the table edges to form a closed rectangle. Next, a data structure of a closed rectangle is generated, a quadrilateral fit is performed on the data structure, and the area of the quadrilateral is filtered, for example, an area threshold is set to filter, so as to remove quadrilaterals below the area threshold (for example, a part with too small area, which may be "noise", can be filtered), and a quadrilateral set is obtained. Then, in the quadrangle set, the center points of the quadrangles are grouped into one or more clusters according to the standard of whether the quadrangles are adjacent and form rows as well as columns, each cluster represents a table, and the points represent the centers of the cells, so that the cell table is obtained.

In block 604, an order table is determined from the cell table.

In one embodiment, the processing module 52 counts the row-column relationship in the unit table, and determines the most-column table as the order table (in the order interface, the table for inputting the order information is the most-column table), that is, the table including the information for displaying the order. Other tables may be tables for other purposes and the invention is not limited.

In block 606, the processing module 52 identifies an order name in the order table.

In one embodiment, the processing module 52 segments the order table and performs OCR recognition on the segmented cells to obtain the identified order name. The identified order name may be one order name or may comprise a plurality of order names, for example, three order names below the "item name" column in fig. 4. In the case where a plurality of order names are identified, the weight and the judgment may be calculated separately for each order name.

In block 608, for an identified order name, the processing module 52 determines a reference order name in the reference order name set that is closest to the identified order name.

In one embodiment, the processing module 52 calculates an edit distance between the identified order name and each reference order name in the reference order name set, determines a minimum edit distance among the edit distances and an edit type corresponding to the minimum edit distance, and takes the reference order name corresponding to the minimum edit distance as the closest reference order name.

Edit types may include equal and unequal (e.g., move, delete, add).

The edit type "equal" can be understood as two words being compared (i.e., the word in the closest reference order name is identical to the word in the identified order name), which "identical" includes both words being identical and the same location in an order name.

The edit type "unequal" may be understood as the two words compared (i.e., the word in the closest reference order name and the word in the identified order name) not being exactly the same, e.g., the words are not the same and/or the locations in the order name are not the same. The term "move" is understood to mean that the two words to be compared are not located at the same position in the name of the order, and need to be moved and converted to be consistent. "delete" may be understood as an out-word addition of the identified order name relative to the closest reference order name, requiring a delete transformation to change to consistency. "Add" may be understood as a lack of words in the identified order name relative to the nearest reference order name that would require an additional transformation to be transformed to full correspondence.

In block 610, the processing module 52 calculates a sum of the weights of the words of the identified order name.

In one embodiment, the processing module 52 calculates the sum of weights based on the configuration of weights and the edit type in the closest reference order name. For example, for one word of the identified order name, if the edit types are equal, its weight is set to the weight value (+ weight) of the corresponding word in the closest reference order name, for example, it is determined that the edit type of one word in the order name is "equal", and the weight of the corresponding word in the closest reference order name is 0.3, the weight value of the word in the identified order name is set to 0.3. If the edit types are not equal, their weights are set to the negative value (-weight) of the weight value of the corresponding word in the closest reference order name, e.g., it is determined that the edit type of a word in the order name is "not equal" and the weight of the corresponding word in the closest reference order name is 0.3, the weight value of the word in the identified order name is set to-0.3. Then, the weight values of the respective words based on such setting are added to obtain the sum of the above weights.

The processing module 52 may determine whether the identification of the order name is valid by determining whether the sum of the calculated weights meets a predetermined weight threshold.

For example, assuming that the validity probability of the recognition result of the order name is P and the number of words of the order name is n, P can be obtained by the following formula:

whether the identification result of the order name is valid may be determined by comparing the P value with zero.

In one embodiment, if P is greater than zero, the identification of the order name may be considered valid. If P is less than or equal to zero, the identification of the order name may be deemed invalid, i.e., the identification is insufficient to derive the correct order name.

In another embodiment, if P is greater than or equal to zero, the identification of the order name may be considered valid. If P is less than zero, the identification of the order name may be deemed invalid, i.e., the identification is insufficient to derive the correct order name.

The output module 53 outputs the recognition result of the order name determined to be valid together with the corresponding patient identification. For example, to a clinical information system for further medical procedures or medical behavioral assessment.

In a test using the medical data processing scheme according to the invention, the accuracy of the test data reached 98.5%. The recognition rate for words in the order name with a high amount of information and "(", ") such as full angle and half angle is extremely high. For example, for the identification of medical advice names such as "cinobufotalin tablet", "cinobufotalin capsule", and "cinobufotalin injection", the error data may be a dosage form identification error such as "tablet", "capsule", and "injection", and the success rate of identification of "cinobufotalin" with a high information content is extremely high.

The invention also relates to a medical data processing method 700. The method 700 may be performed by the medical data processing apparatus described above or by the medical data processing system described above, and the description given above applies equally. Referring now to fig. 7, the main steps of a medical data processing method 700 are described.

In step 702, a reference medical data set comprising a reference identifier set comprising a plurality of reference patient identifiers and a reference order name set comprising a plurality of reference order names and an order interface image of a medical terminal are acquired.

In step 704, a weight configuration for each of the plurality of reference order names is determined, the weight configuration including weights for words in a reference order name.

In step 706, a patient identification and an order name are identified from the order interface image.

In step 708, upon determining that the identified patient identification matches one of the plurality of reference patient identifications, a sum of weights for words of the identified order name is calculated based on a weight configuration of a reference order name of the plurality of reference order names that is closest to the identified order name.

In step 710, it is determined whether the result of identifying the order name is valid based on the sum of the weights.

In step 712, the valid recognition result of the order name and the corresponding patient identification are output.

The present invention also provides a machine-readable storage medium having stored thereon executable instructions that, when executed, cause a machine to perform the method 700 as described above.

It should be appreciated that examples of machine-readable storage media may include volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules, or other data. Storage media may include, but are not limited to: random Access Memory (RAM), Read Only Memory (ROM), Programmable Read Only Memory (PROM), Erasable Programmable Read Only Memory (EPROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, Compact Discs (CD), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium capable of being used to store information.

In some embodiments, a machine-readable storage medium may store executable computer program instructions that, when executed by one or more processing units, cause the processing units to perform the above-described methods. The executable computer program instructions may include any suitable type of code, for example, source code, compiled code, interpreted code, executable code, static code, dynamic code, and the like. The executable computer program instructions may be implemented using any suitable high-level, low-level, object-oriented, visual, compiled and/or interpreted programming language.

It should be appreciated that reference throughout this specification to "one implementation," "an example implementation," "some implementations," "various implementations," or the like, means that the implementation of the disclosure described may include a particular feature, structure, or characteristic, however, it is not necessary for every implementation to include the particular feature, structure, or characteristic. In addition, some implementations may have some, all, or none of the features described for other implementations.

It is to be understood that various operations may be described as multiple discrete actions or operations in sequence, in a manner that is most helpful in understanding the claimed subject matter. However, the order of description should not be construed as to imply that these operations are necessarily order dependent. In particular, the operations may be performed out of the order presented. In other implementations, various additional operations may be performed and/or various operations that have been described may be omitted.

What has been described above includes examples of the disclosed architecture. It is, of course, not possible to describe every conceivable combination of components and/or methodologies, but one of ordinary skill in the art may recognize that many further combinations and permutations are possible. Accordingly, the novel architecture is intended to embrace all such alterations, modifications and variations that fall within the spirit and scope of the appended claims.