Health big data intelligent management method and system

Abstract

The invention relates to the technical field of health informatics, in particular to an intelligent management method and system for health big data, and the method comprises the steps: recognizing sentences from a handwritten character region of a prescription image to be recognized; determining candidate characters of each character in the sentence; combining the candidate characters of each character to obtain a candidate character group; according to the similarity between the candidate characters in the candidate character group and the corresponding characters, determining the possible degree of the candidate character group as the recognition result of the sentence; according to the occurrence frequency of vocabularies formed by the candidate characters in the candidate character group in the database, determining the comprehensive reasonability degree of the recognition result of taking the candidate characters in the candidate character group as the sentences; determining the recognition accuracy of the recognition result of taking the candidate characters in the candidate character group as the sentences by utilizing the possible degree and the comprehensive reasonability degree; and selecting the candidate character in the candidate character group with the highest recognition accuracy as a sentence recognition result. Therefore, the recognition accuracy of the prescription content in the prescription image is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of health informatics, and in particular to a method and system for intelligent management of health big data. Background Art

[0002] With the deep integration of big data and medical health, online health services have attracted the attention of users due to their high efficiency in medical services. By integrating various types of medical information, health big data can not only provide convenient services such as remote consultation and online drug purchase, but also provide patients with personalized health management plans through the analysis of massive medical data. Among them, the prescription photo upload function is a key link in online health services. It allows users to take a photo of the doctor's handwritten prescription and upload it to the online health service platform. The prescription content in the prescription image is automatically analyzed through text recognition technology to provide medication guidance and health advice.

[0003] In some scenarios, Optical Character Recognition (OCR) technology is widely used in online health services, using OCR technology to identify characters in prescription images one by one. Because doctors' writing habits may make the shape of characters close to other characters, it is difficult to distinguish similar characters in prescription images using OCR technology, resulting in misjudgment of the recognition of characters in prescription images, that is, recognizing a character in the prescription image as another text similar to the character. Therefore, the accuracy of OCR technology in recognizing prescription content in prescription images is low. Summary of the invention

[0004] In order to solve the technical problem that the accuracy of OCR technology in identifying prescription content in prescription images is low, the purpose of the present invention is to provide a health big data intelligent management method and system, and the technical solutions adopted are as follows: In a first aspect, an embodiment of the present application provides a method for intelligent management of health big data, comprising: identifying a sentence from a handwritten text area of ​​a prescription image to be identified, the sentence comprising at least one character; determining candidate text for each character in the sentence; combining the candidate text for each character to obtain a candidate text group for the sentence, each character in the candidate text group corresponding to a candidate text; determining the degree of possibility of the candidate text group as a recognition result of the sentence based on the similarity between the candidate text in the candidate text group and the corresponding characters; determining the comprehensive rationality of the recognition result of the candidate text in the candidate text group as a sentence based on the number of times the vocabulary composed of the candidate text in the candidate text group appears in the database of the health big data system; determining the recognition accuracy of the candidate text in the candidate text group as a recognition result of the sentence using the degree of possibility and the comprehensive rationality; selecting the candidate text in the candidate text group with the highest recognition accuracy as the recognition result of the sentence.

[0005] Optionally, combining the candidate characters of each character to obtain a candidate character group for the sentence includes: for each character, arbitrarily selecting a candidate character from its candidate characters; and combining the candidate characters selected for each character to obtain a candidate character group for the sentence.

[0006] Optionally, determining the possibility of the candidate character group being a recognition result of a sentence based on the similarity between the candidate characters in the candidate character group and the corresponding characters includes: multiplying the similarity between each candidate character in the candidate character group and the corresponding character to obtain the possibility of the candidate character group being a recognition result of a sentence.

[0007] Optionally, determining the comprehensive rationality of the recognition results of the candidate characters in the candidate character group as sentences based on the number of times the vocabulary composed of the candidate characters in the candidate character group appears in the database of the health big data system includes: respectively forming a group of candidate character sets with three adjacent candidate characters in the candidate character group; for each candidate character set, respectively forming the first two candidate characters in the candidate character set into a first vocabulary, the last two candidate characters into a second vocabulary, and the three candidate characters into a third vocabulary; determining the rationality of the candidate character set based on the number of times the first vocabulary, the second vocabulary and the third vocabulary appear in the database; determining the comprehensive rationality of the recognition results of the candidate characters in the candidate character group as sentences based on the rationality of the candidate character set.

[0008] Optionally, determining the rationality of the candidate character set based on the number of times the first word, the second word, and the third word appear in the database includes: calculating a first product of the first number of times the first word appears in the database, a second number of times the second word appears in the database, and a third number of times the third word appears in the database, a first difference between the first number and the third number, a second difference between the second number and the third number, and a second product of the first difference and the second difference; determining the rationality of the candidate character set based on the first product and the second product.

[0009] Optionally, determining the comprehensive rationality of the recognition result of the candidate characters in the candidate character group as a sentence based on the rationality of the candidate character sets includes: multiplying the rationality of each candidate character set to obtain the comprehensive rationality.

[0010] Optionally, using the possibility degree and the comprehensive rationality degree to determine the recognition accuracy of the recognition result of the candidate characters in the candidate character group as the sentence includes: determining the product of the possibility degree and the comprehensive rationality degree as the recognition accuracy.

[0011] Optionally, recognizing a sentence from a handwritten text area of ​​a prescription image to be recognized includes: preprocessing the prescription image to be recognized to obtain a grayscale prescription image, the preprocessing including grayscaling and noise reduction; identifying a connected domain in the grayscale prescription image, and grouping continuous text in the same connected domain into a sentence.

[0012] Optionally, determining candidate text for each character in the sentence includes: recognizing the characters in the sentence based on an OCR recognition algorithm; and determining text having a similarity with the characters greater than a threshold as candidate text for the characters.

[0013] In the second aspect, an embodiment of the present application provides a health big data intelligent management system, which includes: a processor and a memory; wherein the memory is used to store computer programs that can be run on the processor; the processor is used to execute the program stored in the memory to implement the steps of the health big data intelligent management method mentioned in the first aspect.

[0014] The present invention has the following beneficial effects: firstly, a sentence is identified from a handwritten text area of ​​a prescription image to be identified, the sentence includes at least one character, then the candidate text of each character in the sentence is determined, and then the candidate text of each character is combined to obtain a candidate text group of the sentence, each character in the candidate text group corresponds to a candidate text; and according to the similarity between the candidate text in the candidate text group and the corresponding character, the possibility degree of the candidate text group as the recognition result of the sentence is determined; secondly, according to the number of times the vocabulary composed of the candidate text in the candidate text group appears in the database of the health big data system, the comprehensive rationality degree of the candidate text in the candidate text group as the recognition result of the sentence is determined; then the possibility degree and the comprehensive rationality degree are used to determine the recognition accuracy of the candidate text in the candidate text group as the recognition result of the sentence; finally, the candidate text in the candidate text group with the largest recognition accuracy is selected as the recognition result of the sentence.

[0015] In this way, after determining the candidate text of the character, the embodiment of the present application combines the candidate text of each character to obtain a candidate text group and determines the possibility of the candidate text in the candidate text group as a recognition result of the sentence according to the similarity between the candidate text and the corresponding character, and then the candidate text in the candidate text group is composed of words, and the number of times these words appear in the database of the health big data system is counted, so as to further determine the rationality of the recognition result of these candidate texts as sentences. Therefore, the embodiment of the present application takes into account the rationality and semantics of the vocabulary composed of the candidate text of each character, thereby determining the recognition result of the character in the sentence, and improving the recognition accuracy of the prescription content in the prescription image. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A flowchart of a method for intelligent management of health big data provided by one embodiment of the present invention; Figure 2 A schematic diagram of the structure of a health big data intelligent management system provided by one embodiment of the present invention; Figure 3 A schematic diagram of the structure of another intelligent management system for health big data provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, characteristics and effects of a health big data intelligent management method proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0020] The specific scheme of the intelligent management method of health big data provided by the present invention is described in detail below with reference to the accompanying drawings.

[0021] Embodiment 1: See also Figure 1 , which shows a flow chart of a health big data intelligent management method provided by an embodiment of the present invention, including: Step S101 : recognizing a sentence from a handwritten text area in a prescription image to be recognized.

[0022] Wherein, a sentence includes at least one character.

[0023] Specifically, the prescription image to be identified is obtained by preprocessing the prescription image taken by the user. The prescription image is a doctor's handwritten prescription, which includes sentences written by the doctor, and each sentence includes a vocabulary consisting of at least one character or a continuous short sentence or long sentence with certain semantics. The sentence can be identified from the handwritten text area of ​​the prescription image to be identified by using a regional growing algorithm, so as to identify the sentence in the handwritten text area in the prescription image.

[0024] As an optional embodiment of the present application, recognizing a sentence from a handwritten text area of ​​a prescription image to be recognized includes: The prescription image to be identified is preprocessed to obtain a prescription image, wherein the preprocessing includes graying and noise reduction; the connected domain in the prescription image is identified, and continuous text in the same connected domain is grouped into sentences.

[0025] Specifically, preprocessing the prescription image can be graying, denoising, and other processing of the prescription image, and the obtained prescription image is a grayed image, and then the handwritten text area in the grayed prescription image is identified through the region growing algorithm, and the handwritten text area contains the doctor's handwritten text. Then the connected domain in the handwritten text area in the prescription image is identified, and the text in the same connected domain represents that it belongs to the same sentence. In this way, the sentences in the prescription image are identified. The number of sentences can be one or more, and the characters in the sentence can be one or more, which is not limited in the embodiments of the present application. Furthermore, the text type in the sentence can be the name of the drug, symptom information, doctor's medical advice and guidance, etc. It is worth noting that in the prior art, the specific implementation scheme of identifying sentences through the region growing algorithm also belongs to the technical content that needs to be protected in the embodiments of the present application.

[0026] Step S102, determining candidate words for each character in the sentence.

[0027] Specifically, for each sentence obtained in the above steps, each character in each sentence is recognized. Character recognition can be specifically performed using OCR technology, so as to find a series of characters of the character as candidate characters of a certain character. For example, a number of sentences in a prescription are recognized by a region growing algorithm. For example, this sentence has characters, for which For characters, there are Candidate text.

[0028] Furthermore, the similarity between a character and a candidate text can directly indicate the possibility that the character belongs to a certain text, and therefore, similarity can be used as the primary criterion for screening candidate texts. Each character in each sentence is recognized, and OCR technology can be used to recognize characters, so as to find a series of texts with a high similarity to the character as candidate texts for a certain character. As an optional embodiment of the present application, determining the candidate texts for each character in a sentence includes: recognizing the characters in the sentence based on an OCR recognition algorithm; and determining texts with a similarity to the characters greater than a threshold as candidate texts for the characters.

[0029] Specifically, the threshold value can be determined according to the actual scenario, and the embodiment of the present application is not limited here. For example, assuming that for a certain character, it is recognized by OCR technology, and a character with a similarity greater than The text is taken as the candidate text of the character. Among them, when calculating the similarity between characters and texts by OCR technology, the template matching method can be used. The specific process is as follows: a standard template character library is established, and then the characters to be recognized are matched with the characters in the template character library for matching similarity calculation, and the text corresponding to the characters with matching similarity greater than the threshold is obtained as the candidate text. The advantage of using the template matching method is that the recognition process directly compares the similarity matching of the two images, and the accuracy and practicality are very high. Of course, the similarity between characters and texts can also be calculated in other ways, and the embodiments of the present application are not limited here.

[0030] Step S103, combining the candidate words of each character to obtain a candidate word group of the sentence.

[0031] Each character in the candidate character group corresponds to a candidate character.

[0032] Specifically, the text in the prescription image is usually some medical professional terms, such as drug names, disease names, etc. There is a certain correlation and coherence between each word and its adjacent words. Therefore, the candidate words can be further screened according to the contextual characteristics of each word combination. For example, this sentence has characters, for which For characters, there are Candidate text.

[0033] Furthermore, as an optional embodiment of the present application, the candidate characters of each character are combined to obtain a candidate character group for the sentence, including: for each character, arbitrarily selecting a candidate character from its candidate characters; and combining the candidate characters selected for each character to obtain a candidate character group for the sentence.

[0034] Specifically, for the first Sentences, of which there are characters, Characters have candidate words, then the paragraph The candidate word combinations for each character in the sentence have a total of For example, The The candidate text group corresponding to the possible situation indicates that the first character of a sentence selects the first character candidate text, the second character selects the second character Candidate words, Character selection The character Candidate text.

[0035] Step S104, determining the likelihood of the candidate character group being a recognition result of a sentence based on the similarity between the candidate characters in the candidate character group and the corresponding characters.

[0036] Specifically, according to the similarity between each character in the candidate character group and its corresponding candidate character, the possibility of the combination of candidate characters in the candidate character group being used as the recognition result of the sentence is determined. Among them, the product of the similarity between each character and its corresponding candidate character can be used as the accuracy of all characters in the sentence as the recognition result of the sentence. Therefore, the similarity between each character and its corresponding candidate character can be multiplied together to determine the possibility of the combination of each candidate character in the candidate character group being used as the recognition result of the sentence.

[0037] As an optional embodiment of the present application, determining the possibility of the candidate character group being a recognition result of a sentence based on the similarity between the candidate characters in the candidate character group and the corresponding characters includes: multiplying the similarity between each candidate character in the candidate character group and the corresponding character to obtain the possibility of the candidate character group being a recognition result of a sentence.

[0038] Specifically, assuming that Character and its The similarity of candidate words is recorded as . Then for In the case of For the candidate character group corresponding to the possible situation, the possibility of this situation is recorded as , The calculation formula is as follows: ; In the above formula, Indicates the first Characters similar to The similarity of the candidate words directly reflects the character selects the first character among all its candidate characters The candidate characters are used as the accuracy of the character recognition result.

[0039] Furthermore, in a sentence there are characters, is to include all The similarity of a similar candidate character is multiplied by the first character in the candidate character group. The second character selects its first candidate words, ..., Select the character When there are candidate characters, the accuracy of the recognition result of all characters in the sentence in this case represents the possibility of using the candidate character group as the recognition result of the sentence. The larger the value of , the higher the accuracy of the recognition result of all the characters in the sentence, that is, the greater the possibility that the candidate character group is recognized as a sentence.

[0040] Step S105, determining the comprehensive rationality of the recognition results of the candidate characters in the candidate character group as sentences according to the number of times the vocabulary composed of the candidate characters in the candidate character group appears in the database of the health big data system.

[0041] Specifically, because doctors often write prescriptions in a hurry and their handwriting is rather sloppy, the similarity between characters and candidate characters is not very reliable as a screening criterion. Therefore, the candidate characters can be further screened through other information to further improve the recognition accuracy. Among them, prescriptions are often issued by diagnosing the patient's physical condition and providing the patient with information such as the name of the medicine required for the treatment. The information of the characters themselves and the information of several characters connected are very likely to be the name information of common drugs included in the database of the health big data system. Therefore, the candidate characters can be further screened based on the degree of rationality of the different combinations of the characters themselves and the candidate characters in the context.

[0042] Furthermore, the comprehensive rationality refers to the rationality of the combination selection of candidate characters in the candidate character group. If the comprehensive rationality is greater, it means that the selection of candidate characters in the candidate character group is more reasonable. The number of times the vocabulary composed of candidate characters in the candidate character group appears in the database of the health big data system refers to the number of times the vocabulary composed of candidate characters appears in the drug use record of the database of the health big data system. The drug use record can be a record of drugs used by historical users. For example, the vocabulary composed of candidate characters is "something aspirin", and "something aspirin" appears 5 times in the drug use record of the database of the health big data system. Then the number of times the vocabulary composed of candidate characters "something aspirin" appears in the database of the health big data system is 5 times.

[0043] As an optional embodiment of the present application, determining the comprehensive rationality of the recognition results of the candidate characters in the candidate character group as sentences according to the number of times the vocabulary composed of the candidate characters in the candidate character group appears in the database of the health big data system includes: respectively forming a group of candidate character sets with three adjacent candidate characters in the candidate character group; for each candidate character set, respectively forming the first two candidate characters in the candidate character set into a first vocabulary, the last two candidate characters into a second vocabulary, and the three candidate characters into a third vocabulary; determining the rationality of the candidate character set according to the number of times the first vocabulary, the second vocabulary and the third vocabulary appear in the database; determining the comprehensive rationality of the recognition results of the candidate characters in the candidate character group as sentences according to the rationality of the candidate character set.

[0044] Specifically, for the There are many different possible permutations and combinations of candidate characters in a sentence. The combined meaning of the candidate characters in the entire candidate character group may not only be the name of the drug, but may also contain other information. Therefore, local continuous characters (such as two to three adjacent candidate characters) are more likely to express the name, and their recognition accuracy is higher. Therefore, two to three adjacent candidate characters in a candidate character group are analyzed first.

[0045] Further, for the candidate text group possible cases, the three adjacent candidate characters are grouped into a candidate character set, and the candidate character set has a total of Group, each candidate text set is recorded as , represents three adjacent candidate characters in a candidate character group; for a candidate character set, the number of occurrences of the first two candidate characters, the second two candidate characters, and the vocabulary consisting of the three candidate characters in the drug use records recorded in the database of the health big data system are observed respectively, and recorded as .in, Indicates its Candidate word set The number of times the word consisting of the first two candidate characters appears in the drug use records recorded in the database of the health big data system, Indicates its Candidate word set The number of times the word consisting of the two candidate characters appears in the drug use records recorded in the database of the health big data system, Indicates its Candidate word set The number of times the word consisting of the three candidate characters appears in the drug usage records recorded in the database of the health big data system.

[0046] As an optional embodiment of the present application, determining the rationality of a candidate character set based on the number of times a first word, a second word, and a third word appear in a database includes: calculating a first product of the first number of times the first word appears in the database, a second number of times the second word appears in the database, and a third number of times the third word appears in the database, a first difference between the first number and the third number, a second difference between the second number and the third number, and a second product of the first difference and the second difference; determining the rationality of the candidate character set based on the first product and the second product.

[0047] Specifically, the rationality of the candidate text set can be calculated using the following formula: ; In the above formula, Indicates The first sentence Candidate word set reasonable degree, Indicates that for The first Candidate word set The number of times the word consisting of the first two candidate characters appears in the drug use records recorded in the database of the health big data system, Indicates that for The first sentence Candidate word set The number of times the word consisting of the two candidate characters appears in the drug use records recorded in the database of the health big data system, Indicates that for The first Candidate word set The number of times the word consisting of the three candidate characters appears in the drug usage records recorded in the database of the health big data system. They can respectively represent the rationality of the vocabulary they constitute. The larger the three values ​​are, the more reasonable the selection of the three candidate characters in the candidate character set is. represents the normalization function, which can The calculation results are normalized to the range of [0,1].

[0048] Furthermore, It represents the product of the difference between the number of times the combination of two candidate characters appears and the number of times the combination of three candidate characters appears. Because the combination of two candidate characters is included in the combination of three candidate characters, the number of times the combination of two candidate characters appears must be greater than the number of times the combination of three candidate characters appears, and or The smaller the difference, the lower the probability of ambiguity between the two candidate word combinations. The smaller the value of is, the more reasonable the selection of the three candidate characters in the candidate character set is.

[0049] As an optional embodiment of the present application, determining the comprehensive rationality of the recognition result of the candidate characters in the candidate character group as a sentence according to the rationality of the candidate character sets includes: multiplying the rationality of each candidate character set to obtain the comprehensive rationality.

[0050] Specifically, for the The first character selected in each sentence The candidate text groups corresponding to the possible situations are calculated by comprehensively considering the rationality of each candidate text set. The overall rationality of the sentence is set as a quantitative value in this embodiment of the application: To express the overall rationality, The calculation formula is as follows: ; In the above formula, Indicates The first character selected in each sentence The comprehensive rationality of the candidate text groups corresponding to the possible situations, Indicates multiplying the rationality of each candidate text set to obtain the The candidate words for each character in the sentence are selected separately In the case of . The larger the value, the The first character selected in each sentence The candidate text group corresponding to the possible situation is taken as the first The higher the reasonableness of the recognition result of a sentence.

[0051] Step S106, using the possibility degree and the comprehensive rationality degree, determining the recognition accuracy of the candidate characters in the candidate character group as the recognition result of the sentence.

[0052] Specifically, in getting The first character selected in each sentence The candidate text group corresponding to the possible situation is taken as the first After determining the probability degree and comprehensive rationality degree of the recognition result of a sentence, the comprehensive rationality degree can be used to correct the probability degree, that is, the product of the probability degree and the comprehensive rationality degree is determined as the recognition accuracy.

[0053] In the present application embodiment, the It represents the recognition accuracy of the candidate characters in the candidate character group as the recognition result of the sentence, which is specifically calculated using the following formula: ; In the above formula, This value comprehensively represents the Each character in the sentence is selected In the case of a candidate character, The candidate character combination corresponding to the situation is taken as the first The recognition accuracy of the recognition results of the sentences. The larger the value, the The candidate character combination corresponding to the situation is taken as the first The higher the accuracy of the recognition result of a sentence, the more it can represent the The candidate character combination corresponding to the situation is taken as the first The authenticity of the recognition results of the sentences.

[0054] Step S107, selecting the candidate character in the candidate character group with the highest recognition accuracy as the recognition result of the sentence.

[0055] Specifically, calculate the The recognition accuracy of the candidate characters in each character selection case of the sentence is calculated, and the case with the highest recognition accuracy is selected as the final screening result.

[0056] It is worth noting that all sentences in the prescription image are identified and screened using the same method, and the final screened results are calculated and the final screened candidate texts obtained by the above calculations are output as the final recognized texts.

[0057] After determining the candidate texts of the characters, the embodiment of the present application combines the candidate texts of each character to obtain a candidate text group and determines the possibility of the candidate text in the candidate text group as a recognition result of the sentence according to the similarity between the candidate text and the corresponding character, and then the candidate text in the candidate text group is composed of words, and the number of times these words appear in the database of the health big data system is counted, so as to further determine the rationality of the recognition results of these candidate texts as sentences. Therefore, the embodiment of the present application takes into account the rationality and semantics of the vocabulary composed of the candidate texts of each character, so as to determine the recognition results of the characters in the sentence, and improve the recognition accuracy of the prescription content in the prescription image.

[0058] Embodiment 2: Corresponding to the health big data intelligent management method provided in the above embodiment, based on the same technical concept, the embodiment of the present application also provides a health big data intelligent management system, Figure 2 A schematic diagram of the module composition of the health big data intelligent management system provided in the embodiment of the present application, which is used to execute Figure 1 The intelligent management method of health big data described in this paper is as follows: Figure 2 As shown, the health big data intelligent management system 200 includes: a recognition module 201, which is used to recognize a sentence from a handwritten text area of ​​a prescription image to be recognized, wherein the sentence includes at least one character; a determination module 202, which is used to determine the candidate text of each character in the sentence; a combination module 203, which is used to combine the candidate text of each character to obtain a candidate text group of the sentence, wherein each character in the candidate text group corresponds to a candidate text; the determination module 202 is also used to determine the possibility of the candidate text group as a recognition result of the sentence according to the similarity between the candidate text in the candidate text group and the corresponding character; the determination module 202 is also used to determine the comprehensive rationality of the recognition result of the candidate text in the candidate text group as a sentence according to the number of times the vocabulary composed of the candidate text in the candidate text group appears in the database of the health big data system; the determination module 202 is also used to determine the recognition accuracy of the candidate text in the candidate text group as a recognition result of the sentence by using the possibility degree and the comprehensive rationality degree; the selection module 204 is used to select the candidate text in the candidate text group with the largest recognition accuracy as the recognition result of the sentence.

[0059] Optionally, the combination module 203 is further used to select, for each character, any candidate character from its candidate characters; and combine the candidate characters selected for each character to obtain a candidate character group for the sentence.

[0060] Optionally, the determination module 202 is further configured to multiply the similarity between each candidate character in the candidate character group and the corresponding character to obtain a probability degree of the candidate character group being a recognition result of a sentence.

[0061] Optionally, the determination module 202 is further used to respectively group three adjacent candidate characters in the candidate character group into a group of candidate character sets; for each candidate character set, respectively group the first two candidate characters in the candidate character set into a first vocabulary, the last two candidate characters into a second vocabulary, and the three candidate characters into a third vocabulary; determine the rationality of the candidate character set according to the number of times the first vocabulary, the second vocabulary, and the third vocabulary appear in the database; and determine the comprehensive rationality of the recognition results of the candidate characters in the candidate character group as sentences according to the rationality of the candidate character set.

[0062] Optionally, the determination module 202 is further used to calculate a first product of the first number of times the first word appears in the database, a second number of times the second word appears in the database, and a third number of times the third word appears in the database, a first difference between the first number and the third number, a second difference between the second number and the third number, and a second product of the first difference and the second difference; and determine the rationality of the candidate character set based on the first product and the second product.

[0063] Optionally, the determination module 202 is further configured to multiply the rationality of each candidate character set to obtain a comprehensive rationality.

[0064] Optionally, the determination module 202 is further configured to determine the product of the possibility degree and the comprehensive rationality degree as the recognition accuracy.

[0065] Optionally, the recognition module 201 is further used to preprocess the prescription image to be recognized to obtain a grayscale prescription image, wherein the preprocessing includes grayscale conversion and noise reduction; recognize connected domains in the grayscale prescription image, and group continuous text in the same connected domain into sentences.

[0066] Optionally, the determination module 202 is further configured to recognize characters in the sentence based on an OCR recognition algorithm; and determine characters having a similarity with the characters greater than a threshold as candidate characters of the characters.

[0067] The health big data intelligent management system provided in the embodiment of the present application can implement each process in the embodiment corresponding to the above-mentioned health big data intelligent management method, and has the same or similar beneficial effects. To avoid repetition, it will not be repeated here.

[0068] Embodiment three: Corresponding to the health big data intelligent management method provided in the above embodiment, based on the same technical concept, the embodiment of the present application also provides a health big data intelligent management system, which is used to execute the above health big data intelligent management method. Figure 3 A schematic diagram of a health big data intelligent management system for implementing each embodiment of the present application is shown below: Figure 3The intelligent management system for health big data may have relatively large differences due to different configurations or performances, and may include one or more processors 301 and memory 302, wherein the memory 302 is used to store computer programs that can be run on the processor, and the processor is used to execute the programs stored in the memory to achieve the above Figure 1 The various steps in the method embodiment. Among them, the memory 302 can be a temporary storage or a permanent storage. The application stored in the memory 302 may include one or more modules (not shown in the figure), and each module may include a series of computer executable instructions in the health big data intelligent management system.

[0069] Furthermore, the processor 301 can be configured to communicate with the memory 302, and execute a series of computer executable instructions in the memory 302 on the health big data intelligent management system. The health big data intelligent management system can also include one or more power supplies 303, one or more wired or wireless network interfaces 304, one or more input and output interfaces 305, and one or more keyboards 306.

[0070] Specifically in this embodiment, the health big data intelligent management system includes a processor, a communication interface, a memory and a communication bus; wherein the processor, the communication interface and the memory communicate with each other through the bus; the memory is used to store computer programs; the processor is used to execute the programs stored in the memory to achieve the above Figure 1 The various steps in the method embodiment are similar to those in the method embodiment, and have the beneficial effects of the above method embodiments. To avoid repetition, the embodiments of the present application will not be described again here.

[0071] It should be noted that the health big data intelligent management system provided in the embodiment of the present application and the health big data intelligent management method provided in the embodiment of the present application are based on the same application concept. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned health big data intelligent management method, and has the same or similar beneficial effects, and the repetitions will not be repeated.

[0072] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0073] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for intelligent management of health big data, characterized in that: The health big data intelligent management method comprises: Recognize a sentence from a handwritten text area of ​​the prescription image to be recognized, wherein the sentence includes at least one character; Determining candidate words for each of the characters in the sentence; Combining the candidate words of each of the characters to obtain a candidate word group of the sentence, wherein each character in the candidate word group corresponds to a candidate word; Determining the likelihood of the candidate character group being a recognition result of the sentence based on the similarity between the candidate characters in the candidate character group and the corresponding characters; Determining the comprehensive rationality of the candidate characters in the candidate character group as the recognition result of the sentence according to the number of times the vocabulary composed of the candidate characters in the candidate character group appears in the database of the health big data system; Determine the recognition accuracy of the candidate characters in the candidate character group as the recognition result of the sentence by using the possibility degree and the comprehensive rationality degree; The candidate character in the candidate character group with the highest recognition accuracy is selected as the recognition result of the sentence.

2. The method for intelligent management of health big data according to claim 1, characterized in that: The candidate words of each of the characters are combined to obtain the candidate word group of the sentence, which includes: For each character, select a candidate character from its candidate characters at random; The candidate characters selected for each of the characters are combined to obtain a candidate character group for the sentence.

3. The intelligent management method for health big data according to claim 1, characterized in that: Determining the possibility of the candidate character group being the recognition result of the sentence according to the similarity between the candidate characters in the candidate character group and the corresponding characters includes: The similarity between each candidate character in the candidate character group and the corresponding character is multiplied to obtain the possibility of the candidate character group being the recognition result of the sentence.

4. The method for intelligent management of health big data according to claim 1, characterized in that: The step of determining the comprehensive rationality of the candidate characters in the candidate character group as the recognition result of the sentence according to the number of times the vocabulary composed of the candidate characters in the candidate character group appears in the database of the health big data system includes: respectively forming a candidate character set from three adjacent candidate characters in the candidate character group; For each candidate character set, the first two candidate characters in the candidate character set are respectively formed into a first vocabulary, the second two candidate characters are formed into a second vocabulary, and the three candidate characters are formed into a third vocabulary; Determining the rationality of the candidate word set according to the number of times the first word, the second word, and the third word appear in the database; According to the rationality of the candidate character set, the comprehensive rationality of the candidate characters in the candidate character group as the recognition result of the sentence is determined.

5. The intelligent management method for health big data according to claim 4 is characterized in that: The determining the rationality of the candidate word set according to the number of times the first word, the second word, and the third word appear in the database includes: Calculate a first product of a first number of occurrences of the first word in the database, a second number of occurrences of the second word in the database, and a third number of occurrences of the third word in the database, a first difference between the first number and the third number, a second difference between the second number and the third number, and a second product of the first difference and the second difference; The rationality of the candidate character set is determined according to the first product and the second product.

6. The intelligent management method for health big data according to claim 4 is characterized in that: Determining the comprehensive rationality of the candidate characters in the candidate character group as the recognition result of the sentence according to the rationality of the candidate character set includes: The rationality of each candidate character set is multiplied to obtain the comprehensive rationality.

7. The method for intelligent management of health big data according to claim 1, characterized in that: The determining, by using the possibility degree and the comprehensive rationality degree, the recognition accuracy of the candidate characters in the candidate character group as the recognition result of the sentence comprises: The product of the possibility degree and the comprehensive rationality degree is determined as the recognition accuracy.

8. The method for intelligent management of health big data according to claim 1, characterized in that: The step of recognizing a sentence from a handwritten text region of the prescription image to be recognized comprises: Preprocessing the prescription picture to be identified to obtain the prescription image, wherein the preprocessing includes grayscale conversion and noise reduction; Connected domains in the prescription image are identified, and continuous characters in the same connected domain are grouped into the sentence.

9. The intelligent management method for health big data according to claim 1, characterized in that: The determining of candidate words for each of the characters in the sentence comprises: Recognize characters in the sentence based on an OCR recognition algorithm; Determine the characters whose similarity to the character is greater than a threshold as candidate characters for the character.

10. A health big data intelligent management system, characterized in that: The health big data intelligent management system includes: a processor and a memory; wherein the memory is used to store computer programs that can be run on the processor; the processor is used to execute the program stored in the memory to implement the steps of the health big data intelligent management method as described in any one of claims 1-9.

Images