An intelligent input method and system for patients' vital signs based on speech recognition

Abstract

The present invention discloses an intelligent input method and system for a patient's vital signs based on speech recognition, which relates to the technical field of intelligent input, and includes the following steps: calculating a pronunciation matrix, correcting the first information to generate second information, extracting keywords of the second information and verifying and calibrating the keywords, calculating a health factor and a collection time period, updating a nursing record database and marking the update time and the name of the updating personnel. The present invention analyzes the medical staff's accent to obtain pronunciation features, corrects the first information based on the pronunciation features for keyword pairs, extracts the keywords in the second information, and verifies and calibrates the keywords, effectively avoiding the probability of misrecognition, facilitating the elimination of incorrect data, improving the accuracy of the nursing record database, calculating the collection time period based on the patient's basic information and historical disease information, and improving the stability of the nursing record database.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent input, and particularly relates to a method and system for intelligent input of patient vital signs based on speech recognition. Background Art

[0002] In recent years, the development of intelligent information processing has been continuous, and the nursing industry has been developing towards digitalization, intelligence, networking, and personalization. The continuous innovation and progress of intelligent input technology have promoted the rapid development of the nursing industry, realized the upgrading and transformation of nursing services, and met the growing nursing needs of the people.

[0003] Currently, a Chinese invention patent with the application publication number CN 110931089 A discloses a system and method for EMR vital sign recording. Patient vital signs are obtained by converting voice information into text. The voice-to-text algorithm is activated by keywords spoken by the caregiver. The positioning system consists of a caregiver positioning tag worn by the caregiver, which communicates with a positioning receiver within the vicinity to enable ward positioning. However, in the related art, there is no correction of word errors for the pronunciation characteristics of the medical staff, which easily leads to missed recognition or misrecognition of keywords. There is no setting of the collection frequency based on the patient's health information, resulting in untimely data collection, which has an adverse impact on the patient's health, or overly dense collection, which is not conducive to the stability of the nursing record database. Summary of the Invention

[0004] The technical problem solved by the present invention is that in the related art, there is no correction of word errors for the pronunciation characteristics of the medical staff, which easily leads to missed recognition or misrecognition of keywords. There is no setting of the collection frequency based on the patient's health information, resulting in untimely data collection, which has an adverse impact on the patient's health, or overly dense collection, which is not conducive to the stability of the nursing record database.

[0005] To achieve the above object, the present invention is implemented through the following technical solutions: In the first aspect, the present invention provides a method for intelligent input of patient vital signs based on speech recognition, including the following steps:

[0006] Step S1, obtain voice data, preprocess the voice data to generate first information;

[0007] Step S2, extract the names of medical staff in the first information, match the medical staff accent information corresponding to the names of medical staff, analyze the medical staff accent information to obtain pronunciation characteristics, calculate the weights of the elements in the pronunciation characteristics respectively, generate a pronunciation matrix of the medical staff based on the weights of the elements in the pronunciation characteristics, correct the first information based on the pronunciation matrix of the medical staff to generate second information, store the pronunciation matrix of the medical staff, and generate a pronunciation database;

[0008] Step S3: Obtain the second piece of information, extract the keywords of the second piece of information, where the keywords include the patient's name and the vital sign information corresponding to the patient's name. Verify and calibrate the vital sign information corresponding to the patient's name based on historical keywords, store the verified and calibrated vital sign information corresponding to the patient's name, and update the nursing record database.

[0009] Step S4: Obtain the patient's basic information and the patient's medical history information, analyze the patient's basic information and the medical history information to obtain health factors, calculate the acquisition time period based on the health factors, and loop through steps S1 to S3 according to the acquisition time period.

[0010] As a preferred solution of the intelligent voice recognition-based patient vital sign input method of the present invention: The preprocessing includes pre-emphasis processing, frame addition and windowing processing, and endpoint detection processing. The pre-emphasis processing is performed by a high-pass digital filter.

[0011] As a preferred solution of the intelligent voice recognition-based patient vital sign input method of the present invention: Step S2 includes the following sub-steps:

[0012] Step S21: Obtain the first piece of information, extract the medical staff's name, call the medical staff information database, and obtain the medical staff's accent information corresponding to the medical staff's name.

[0013] Step S22: Analyze the medical staff's accent information corresponding to the medical staff's name to obtain the pronunciation features, where the pronunciation features include liaison features, elision features, and pronunciation deformation features.

[0014] Step S23: Calculate the weights of the liaison features, elision features, and pronunciation deformation features in the pronunciation features respectively, and generate a pronunciation matrix of the medical staff according to the weights of the liaison features, elision features, and pronunciation deformation features in the pronunciation features.

[0015] Step S24: Correct the first piece of information according to the pronunciation matrix of the medical staff to generate the second piece of information, store the pronunciation matrix of the medical staff, and generate a pronunciation database.

[0016] As a preferred solution of the intelligent voice recognition-based patient vital sign input method of the present invention: The calculation expression of the weight is:

[0017]

[0018] The expression of the pronunciation matrix of the medical staff is:

[0019]

[0020] The corrected calculation expression is as follows:

[0021]

[0022] Wherein, is the pronunciation matrix of the medical staff numbered ; is the weight of the th pronunciation feature; is the standard deviation of the formant frequency of the th pronunciation feature; is the average value of the formant frequency of the th pronunciation feature; is the weight of the liaison feature in the pronunciation features; is the weight of the elision feature in the pronunciation features; is the weight of the pronunciation deformation feature in the pronunciation features; is the second information of the medical staff numbered ; the is the first information of the medical staff numbered .

[0023] As a preferred solution of the intelligent patient vital sign input method based on speech recognition according to the present invention, wherein: the step S3 includes the following sub-steps:

[0024] Step S31: Prior to call the nursing record database, obtain the historical vital sign information corresponding to the patient's name, and generate a change curve for the historical same vital sign information corresponding to the patient's name respectively;

[0025] Step S32: Obtain the second information, and extract the keywords of the second information, where the keywords include the patient's name, body temperature, pulse, blood oxygen, blood pressure, blood sugar, and collection time;

[0026] Step S33: Call the nursing record database, obtain the historical vital sign information corresponding to the patient's name, verify and calibrate the vital sign information corresponding to the patient's name in the second information according to the first threshold. When the offset of the vital sign information corresponding to the patient's name in the second information in the change curve is greater than the first threshold, calibrate the vital sign information corresponding to the patient's name in the second information to make its offset in the change curve less than the first threshold;

[0027] Step S34: Store the body temperature, pulse, blood oxygen, blood pressure, blood sugar, and collection time corresponding to the patient's name after verification and calibration, and mark the update time and the name of the update person, and update the nursing record database.

[0028] As a preferred solution of the intelligent input method for patient vital signs based on speech recognition according to the present invention, wherein: the step S4 includes the following sub-steps:

[0029] Step S41, obtain the patient's name, call the patient database, obtain the patient's basic information and the patient's medical history information. The patient's basic information includes the patient's height information, the patient's weight information, the patient's gender information, and the patient's age information. Analyze the patient's height information, the patient's weight information, the patient's gender information, the patient's age information, and the patient's medical history information to obtain a health factor;

[0030] Step S42, calculate the acquisition time period based on the health factor.

[0031] As a preferred solution of the intelligent input method for patient vital signs based on speech recognition according to the present invention, wherein: the calculation expression of the health factor is:

[0032]

[0033] The calculation expression of the acquisition time period is:

[0034]

[0035] Wherein, is the health factor, the is the compensation constant, is the scale constant, is the weight of the th information of the patient, is the th information of the patient, is a constant, is the acquisition time period;

[0036] Retrieve the nursing record database and update the body temperature, pulse, blood oxygen, blood pressure, blood sugar, and acquisition time.

[0037] In a second aspect, the present invention provides an intelligent input system for patient vital signs based on speech recognition. This system is used to execute the intelligent input method for patient vital signs based on speech recognition described in claim 1, and includes a collection module, a processing module, a control module, and an update module;

[0038] The collection module is used to collect speech data and transmit the speech data to the processing module;

[0039] The processing module is used to perform emphasis processing, frame windowing processing, and endpoint detection processing on the speech data, generate the first information, and transmit the first information to the control module;

[0040] The control module is used to calculate the pronunciation matrix of the medical staff, perform deviation correction processing on the first based on the pronunciation matrix to generate second information, calculate a health factor based on the second information, and calculate a collection time period according to the health factor;

[0041] The update module updates the nursing record database according to the collection time period.

[0042] In a third aspect, the present invention provides an electronic device, including a storage, a processor, and a computer-readable instruction stored in the storage, and when the computer-readable instruction is executed by the processor, the steps in any one of the above methods are run.

[0043] In a fourth aspect, the present invention provides a storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps in any one of the above methods are run.

[0044] The beneficial effects of the present invention: Analyze the medical staff's accent to obtain pronunciation features, correct the first information based on the pronunciation features to generate second information, effectively avoid the probability of misrecognition, extract keywords in the second information, and verify and calibrate the keywords, which is beneficial to eliminating incorrect data and improving the accuracy of the nursing record database. Calculate the collection time period based on the patient's basic information and historical disease information, improving the stability of the nursing record database. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the basic process of a method and system for intelligent input of patient vital signs based on speech recognition provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific embodiments of the present invention is made in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments.

[0047] Example 1, referring to Figure 1 , which is an embodiment of the present invention, provides a method for intelligent input of patient vital signs based on speech recognition, including the following steps:

[0048] Step S1, obtain speech data, preprocess the speech data to generate first information;

[0049] Step S2: Extract the names of medical staff in the first piece of information, match the corresponding medical staff accent information for the names of medical staff, analyze the medical staff accent information to obtain pronunciation features, calculate the weights of elements in the pronunciation features respectively, generate a pronunciation matrix for the medical staff based on the weights of elements in the pronunciation features, correct the first piece of information based on the pronunciation matrix of the medical staff to generate a second piece of information, store the pronunciation matrix of the medical staff, and generate a pronunciation database;

[0050] Step S3: Obtain the second piece of information, extract the keywords of the second piece of information, where the keywords include the patient's name and the vital sign information corresponding to the patient's name, verify and calibrate the vital sign information corresponding to the patient's name based on historical keywords, store the verified and calibrated vital sign information corresponding to the patient's name, and update the nursing record database;

[0051] Step S4: Obtain the patient's basic information and the patient's medical history information, analyze the patient's basic information and the medical history information to obtain health factors, calculate the collection time period based on the health factors, and loop steps S1 - S3 according to the collection time period;

[0052] Step S5: Retrieve the nursing record database, update the vital sign information corresponding to the patient's name, and mark the update time and the name of the updater.

[0053] The present invention corrects the word errors in the pronunciation features of the medical staff, avoids missed recognition and misrecognition of keywords, sets the collection frequency based on the patient's health information, is conducive to timely data collection and the patient's health, or avoids overly dense collection, which is conducive to the stability of the nursing record database.

[0054] The preprocessing includes pre - emphasis processing, frame - adding window processing, and endpoint detection processing, and the pre - emphasis processing is performed by a high - pass digital filter.

[0055] In specific implementation, the pre - emphasis coefficient is 0.97, the frame length of the frame - adding window processing is 20 ms, and the frame shift is 10 ms.

[0056] Step S2 includes the following sub - steps:

[0057] Step S21: Obtain the first piece of information, extract the names of medical staff, call the medical staff information database, and obtain the corresponding medical staff accent information for the names of medical staff;

[0058] Step S22: Analyze the medical staff accent information corresponding to the names of medical staff to obtain the pronunciation features, where the pronunciation features include liaison features, elision features, and pronunciation deformation features;

[0059] Step S23: Calculate the weights of the liaison feature, elision feature, and pronunciation deformation feature in the pronunciation feature respectively, and generate a pronunciation matrix of medical staff based on the weights of the liaison feature, elision feature, and pronunciation deformation feature in the pronunciation feature;

[0060] Step S24: Correct the first information according to the pronunciation matrix of medical staff to generate second information, store the pronunciation matrix of medical staff, and generate a pronunciation database.

[0061] In specific implementation, the medical staff name corresponds to a number in the medical staff system. The weight is the proportion of the liaison feature, elision feature, and pronunciation deformation feature in the pronunciation feature. Based on the weight, compensate for the pronunciation of medical staff to make the pronunciation of medical staff return to the standard pronunciation formant frequency.

[0062] The calculation expression of the weight is:

[0063]

[0064] The expression of the pronunciation matrix of medical staff is:

[0065]

[0066] The calculation expression of the correction is:

[0067]

[0068] Among them, is the pronunciation matrix of the medical staff numbered ; is the weight of the th pronunciation feature; is the standard deviation of the formant frequency of the th pronunciation feature; is the average value of the formant frequency of the th pronunciation feature; is the weight of the liaison feature in the pronunciation feature; is the weight of the elision feature in the pronunciation feature; is the weight of the pronunciation deformation feature in the pronunciation feature; is the second information of the medical staff numbered ; the is the first information of the medical staff numbered .

[0069] In specific implementation, estimate the formant frequency based on the cepstrum method.

[0070] Step S3 includes the following sub-steps:

[0071] Step S31: First, call the nursing record database to obtain the historical vital sign information corresponding to the patient's name, and generate a change curve for the same historical vital sign information corresponding to the patient's name respectively;

[0072] Step S32: Obtain the second piece of information, and extract the keywords of the second piece of information, where the keywords include the patient's name, body temperature, pulse, blood oxygen, blood pressure, blood sugar, and collection time;

[0073] Step S33: Call the nursing record database to obtain the historical vital sign information corresponding to the patient's name, verify and calibrate the vital sign information corresponding to the patient's name in the second piece of information according to the first threshold. When the offset of the vital sign information corresponding to the patient's name in the second piece of information in the change curve is greater than the first threshold, calibrate the vital sign information corresponding to the patient's name in the second piece of information so that its offset in the change curve is less than the first threshold;

[0074] Step S34: Store the body temperature, pulse, blood oxygen, blood pressure, blood sugar, and collection time corresponding to the patient's name after verification and calibration, and mark the update time and the name of the updater, and update the nursing record database.

[0075] In the specific implementation, the first threshold is 0.3, and the offset of the vital sign information corresponding to the patient's name in the second piece of information in the change curve is the change rate of the difference between the same vital sign information and the previous vital sign information.

[0076] The step S4 includes the following sub-steps:

[0077] Step S41: Obtain the patient's name, call the patient database to obtain the patient's basic information and the patient's disease history information. The patient's basic information includes the patient's height information, the patient's weight information, the patient's gender information, and the patient's age information. Analyze the patient's height information, the patient's weight information, the patient's gender information, the patient's age information, and the patient's disease history information to obtain a health factor;

[0078] Step S42: Calculate the collection time period based on the health factor.

[0079] In the specific implementation, the patient's name corresponds to a number in the patient database, and the patient's disease history information is directly retrieved from the electronic medical record system.

[0080] The calculation expression of the health factor is:

[0081]

[0082] The calculation expression of the collection time period is:

[0083]

[0084] Among them, is a health factor, and the is a compensation constant, is a scale constant, is the weight of the th piece of information of the patient, is the th piece of information of the patient, is a constant, is the acquisition time period.

[0085] The present invention analyzes the medical staff's accent to obtain pronunciation features, corrects the first information based on the pronunciation features for keywords, and generates the second information, effectively avoiding the probability of misrecognition. Extracts the keywords in the second information, and verifies and calibrates the keywords, which is beneficial to eliminating incorrect data and improving the accuracy of the nursing record database. Calculates the acquisition time period based on the patient's basic information and historical disease information, improving the stability of the nursing record database.

[0086] Embodiment 2, In a second aspect, the present application provides an electronic device, including a processor and a memory. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the steps in any one of the above methods are run. Through the above technical solution, the processor and the memory are interconnected and communicate with each other through a communication bus and / or other forms of connection mechanisms. The memory stores a computer program executable by the processor. When the electronic device runs, the processor executes the computer program to execute the method in any optional implementation manner of the above embodiment to achieve the following functions: obtaining voice data, preprocessing the voice data to generate the first information, extracting the medical staff name in the first information, matching the medical staff accent information corresponding to the medical staff name, analyzing the medical staff accent information to obtain pronunciation features, respectively calculating the weights of the elements in the pronunciation features, generating a pronunciation matrix of the medical staff based on the weights of the elements in the pronunciation features, correcting the first information based on the pronunciation matrix of the medical staff to generate the second information, storing the pronunciation matrix of the medical staff, and generating a pronunciation database;

[0087] Obtaining the second information, extracting the keywords of the second information, the keywords including the patient name and the vital sign information corresponding to the patient name, verifying and calibrating the vital sign information corresponding to the patient name based on historical keywords, and storing the verified and calibrated vital sign information corresponding to the patient name, updating the nursing record database, obtaining the patient's basic information and the patient's disease history information, analyzing the patient's basic information and the disease history information to obtain a health factor, and calculating the acquisition time period based on the health factor;

[0088] Retrieve the nursing record database, update the vital sign information corresponding to the patient's name, and mark the update time and the name of the updater.

[0089] It should be recognized that the embodiments of the present invention can be implemented or carried out by a combination of computer hardware and software, or by computer instructions stored in a non-transitory computer-readable memory. The method can be implemented in a computer program using standard programming techniques - including a non-transitory computer-readable storage medium configured with the computer program, wherein the storage medium so configured causes the computer to operate in a specific and predefined manner - according to the methods and drawings described in the specific embodiments. Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system. However, if desired, the program can be implemented in assembly or machine language. In any case, the language can be a compiled or interpreted language. In addition, for this purpose the program is capable of running on a special integrated circuit programmed for this purpose.

[0090] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. An intelligent input method for a patient's vital signs based on speech recognition, characterized in that, It includes the following steps: Step S1: Obtain voice data, preprocess the voice data, and generate first information; Step S2: Extract the names of medical staff in the first information, match the corresponding medical staff accent information for the names of medical staff, analyze the medical staff accent information to obtain pronunciation features, calculate the weights of elements in the pronunciation features respectively, generate a pronunciation matrix for the medical staff based on the weights of elements in the pronunciation features, correct the first information based on the pronunciation matrix of the medical staff to generate second information, store the pronunciation matrix of the medical staff, and generate a pronunciation database; Step S3: Obtain the second information, extract the keywords of the second information, where the keywords include the patient's name and the vital sign information corresponding to the patient's name, verify and calibrate the vital sign information corresponding to the patient's name based on historical keywords, store the verified and calibrated vital sign information corresponding to the patient's name, and update the nursing record database; Step S4: Obtain the patient's basic information and the patient's medical history information, analyze the patient's basic information and the medical history information to obtain health factors, calculate the collection time period based on the health factors, and loop steps S1 to S3 according to the collection time period; Step S5: Retrieve the nursing record database, update the vital sign information corresponding to the patient's name, and mark the update time and the name of the updating personnel; Step S2 includes the following sub-steps: Step S21: Obtain the first information, extract the names of medical staff, call the medical staff information database, and obtain the corresponding medical staff accent information for the names of medical staff; Step S22: Analyze the medical staff accent information corresponding to the names of medical staff to obtain the pronunciation features, where the pronunciation features include elision features, aspiration features, and pronunciation deformation features; Step S23: Calculate the weights of the elision features, aspiration features, and pronunciation deformation features in the pronunciation features respectively, and generate a pronunciation matrix for the medical staff based on the weights of the elision features, aspiration features, and pronunciation deformation features in the pronunciation features; Step S24: Correct the first information based on the pronunciation matrix of the medical staff to generate second information, store the pronunciation matrix of the medical staff, and generate a pronunciation database; The names of medical staff correspond to numbers in the medical staff system. The weights are the proportions of the elision features, aspiration features, and pronunciation deformation features in the pronunciation features. Based on the weights, compensate for the pronunciation of the medical staff to make the pronunciation of the medical staff return to the standard pronunciation formant frequency; Step S4 includes the following sub-steps: Step S41: Obtain the patient's name, call the patient database, and obtain the patient's basic information and the patient's medical history information. The patient's basic information includes the patient's height information, the patient's weight information, the patient's gender information, and the patient's age information. Analyze the patient's height information, the patient's weight information, the patient's gender information, the patient's age information, and the patient's medical history information to obtain health factors; Step S42: Calculate the collection time period based on the health factors; The calculation expression of the health factor is as follows: ; The calculation expression of the acquisition time period is as follows: ; Among them, is a health factor, and the is a compensation constant, is a scale constant, is the weight of the -th information of the patient, is the -th information of the patient, is a constant, is the acquisition time period.

2. The intelligent input method for the patient's vital signs based on speech recognition according to claim 1, characterized in that: The preprocessing includes pre-emphasis processing, framing and windowing processing, and endpoint detection processing. The pre-emphasis processing is performed by a high-pass digital filter.

3. The intelligent input method for the patient's vital signs based on speech recognition according to claim 1, wherein: The calculation expression of the weight is as follows: ; The expression of the pronunciation matrix of the medical staff is as follows: ; The calculation expression of the correction is as follows: ; Among them, is the pronunciation matrix of the medical staff numbered . is the weight of the th pronunciation feature, is the standard deviation of the formant frequency of the th pronunciation feature, is the average value of the formant frequency of the th pronunciation feature, is the weight of the liaison feature among the pronunciation features, is the weight of the elision feature among the pronunciation features, is the weight of the pronunciation deformation feature among the pronunciation features, is the second information of the medical staff numbered , and the is the first information of the medical staff numbered .

4. The intelligent input method for a patient's vital signs based on speech recognition according to claim 1, characterized in that: Step S3 includes the following sub-steps: Step S31: First, call the nursing record database to obtain the historical vital sign information corresponding to the patient's name, and generate change curves for the historical vital sign information of the same type corresponding to the patient's name respectively. Step S32: Obtain the second information and extract the keywords of the second information. The keywords include the patient's name, body temperature, pulse, blood oxygen, blood pressure, blood glucose, and collection time. Step S33: Call the nursing record database to obtain the historical vital sign information corresponding to the patient's name, verify and calibrate the vital sign information corresponding to the patient's name in the second information according to the first threshold. When the offset of the vital sign information corresponding to the patient's name in the second information in the change curve is greater than the first threshold, calibrate the vital sign information corresponding to the patient's name in the second information to make its offset in the change curve less than the first threshold. Step S34: Store the body temperature, pulse, blood oxygen, blood pressure, blood glucose, and collection time corresponding to the patient's name after verification and calibration, and mark the update time and the name of the updater, and update the nursing record database.

5. An intelligent patient vital sign input system based on speech recognition, which is used to execute the intelligent patient vital sign input method based on speech recognition described in claim 1, and is characterized in that, It includes a collection module, a processing module, a control module, and an update module; The collection module is used to collect voice data and transmit the voice data to the processing module; The processing module is used to perform pre-emphasis processing, framing and windowing processing, and endpoint detection processing on the voice data to generate the first information, and transmit the first information to the control module; The control module is used to calculate the pronunciation matrix of the medical staff, perform deviation correction processing on the first information based on the pronunciation matrix to generate the second information, calculate the health factor based on the second information, and calculate the collection time period according to the health factor; The update module updates the nursing record database according to the collection time period.

Images