Nursing care certification support system, nursing care certification support method, and nursing care certification support program

Abstract

Provided are a care-required certification support system, a care-required certification support method, and a care-required certification support program that can reduce the workload in the certification survey for care-required certification and improve the accuracy of determination. 【Solution means】A care-required certification support system including a terminal that inputs a certification survey form consisting of a "basic survey" and "special notes" and notifies the status of the subject, an artificial intelligence unit that analyzes the status of the subject, and a server that estimates the status of the subject. The artificial intelligence unit includes a morphological analysis unit that performs morphological analysis on the analysis target in the "special notes", an estimated basis sentence extraction unit that extracts a sentence serving as an estimated basis from the sentences in the "special notes", a normalization unit that normalizes the expression of the sentence extracted by the estimated basis sentence extraction unit, a classification estimation unit that estimates a classification number based on the sentence normalized by the normalization unit, and an answer generation unit that generates an answer corresponding to the criteria of the certification survey form based on the classification number estimated by the classification estimation unit.

Description

Technical Field

[0001] The present invention relates to a long-term care certification support system, a long-term care certification support method, and a long-term care certification support program for assisting in the review and determination of long-term care certification.

Background Art

[0002] In order for the elderly to receive long-term care services under the long-term care insurance system, it is necessary for the insured of the long-term care insurance to undergo an authentication survey by the local government and obtain long-term care certification. In recent years, with the progress of the declining birthrate and aging population, the aging rate has been rising and the working-age population has been decreasing. The number of applicants for long-term care certification has been on the rise, and in recent years, while the pace of increase has been accelerating, the number of local government employees conducting long-term care certification surveys has been on the decline.

Prior Art Documents

Patent Documents

[0003]

Non-Patent Document 1

Non-Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] Non-Patent Document 1 describes the application method for the certification of needing care for receiving care services under the Long-Term Care Insurance System for service users. Specifically, the process from application to receiving care services, such as "Application for Certification of Needing Care", "Certification Survey and Attending Physician's Opinion Letter", "Review and Judgment", "Creation of a Certification and Care (Care Prevention) Service Plan", and "Start of Using Care Services", is described.

[0005] Non-Patent Document 2 discloses the description methods of "Basic Survey" and "Special Notes" in the "Certification Survey Form" used for the certification of needing care. For the "Basic Survey", the applicant's status is selected and described based on the definitions of each survey item, etc. For the "Special Notes", when the basis for the selection of each survey item, or when the selected item in each survey item does not fit well with the definition of the basic survey item, or when considering whether the actual assistance method is appropriate, the relevant details are described and it is created as the "Certification Survey Form". Note that the "Basic Survey" and "Special Notes" are associated with the survey items by survey numbers respectively.

[0006] Fig. 19 shows the process from when a local government receives an application for the certification of needing care, investigates and confirms the application content, makes a determination of needing care, and outputs a notice of the certification result of the certification of needing care.

[0007] Referring to Fig. 19, the conventional business of the certification of needing care will be described. First, the subject (a long-term care insurance insured person who wishes to receive care services or a person who acts as an agent for the procedures of such an insured person, etc.) makes an application for the certification of needing care to the local government (Flow F90 in Fig. 19). Based on the application for the certification of needing care, the local government dispatches an investigator (a local government employee, a staff member of a long-term care certification survey center, etc.) to the location of the subject (home, facility, etc.). The investigator interviews and investigates the physical and mental state of the subject and creates a "Certification Survey Form" consisting of "Basic Survey" and "Special Notes" in paper form (Flow F91 in Fig. 19).

[0008] The local government staff performs the digitization process (OCR (Optical Character Recognition / Reader) capture) of the created "Recognition Survey Form" (Flow F92 in Figure 19). The computer inputs some items of the digitized survey results (Recognition Survey Form) and the attending physician's opinion letter received at the time of application or up to that stage, and the computer determines the degree of care need using a uniform national determination method and outputs it as the result of the primary determination (Flow F93 in Figure 19). Based on the result of the primary determination and the attending physician's opinion letter, the Care Certification Review Committee determines the degree of care need and outputs the result of the secondary determination (Flow F94 in Figure 19). The local government conducts the care need certification based on the result of the secondary determination and notifies the subject of the certification result (Flow F95 in Figure 19).

[0009] As described above, in response to the increasing trend in the number of applicants for care need certification, the number of local government staff conducting the care need recognition surveys is on a decreasing trend, increasing the burden on each local government. What particularly requires a lot of manpower in the process of the care need recognition survey is the confirmation work of the "Recognition Survey Form" created in paper media by local government staff shown in Flow F92 of Figure 19. The local government mobilizes not only the staff of the department in charge of care need certification but also the staff of non-in charge departments to deal with it. Due to the shortage of manpower causing the normal work to be crowded, there is an urgent issue to improve the work of the confirmation work of the "Recognition Survey Form".

[0010] In contrast, using artificial intelligence for the work of the confirmation work of the "Recognition Survey Form" is being considered to reduce the number of personnel required for the work. However, there are fluctuations in the description of the "Special Notes" by each investigator, and there is a risk that it cannot be accurately analyzed even using artificial intelligence. In such a case, it becomes difficult to guarantee the accuracy of the determination result of care need certification.

[0011] Therefore, an object of the present invention is to provide a long-term care certification support system, a long-term care certification support method, and a long-term care certification support program that can reduce the workload in the long-term care certification survey and improve the accuracy of determination by unifying the description content by the examiner in the "Special Notes" of the "Certification Survey Form" into a uniform one nationwide through digitization and classification using artificial intelligence, and estimating it to be an accurate and objective expression regardless of the form of the question.

Means for Solving the Problems

[0012] The long-term care certification support system according to the present invention Basic survey and special notes A certification survey form consisting of Input A terminal, Basic survey and special notes An artificial intelligence unit that analyzes Receive a request from the terminal, request the artificial intelligence unit to analyze the basic survey and special notes, and return the judgment result obtained from the analysis to the terminal A long-term care certification support system including a server, wherein the artificial intelligence unit Special notes A morphological analysis unit that performs morphological analysis on the analysis target in A non-essential sentence removal unit that removes non-essential sentences from the analysis result by the morphological analysis unit, and from the analysis target from which non-essential sentences have been removed by the non-essential sentence removal unit An extraction unit for extraction basis sentences that extracts sentences serving as extraction bases, a normalization unit that normalizes the expressions of the sentences extracted by the extraction unit for extraction basis sentences, and the sentences normalized by the normalization unit Estimate which option among the options for the questions in the basic survey applies, and based on this result, generate a classification number for the corresponding question A classification and estimation unit that Generate Based on the classification numbers classified by the classification and estimation unit, corresponding to the criteria of the certification survey form Determine the judgment result of the need for care certification An answer generation unit that, characterized by having.

[0013] The long-term care certification support system according to the present invention In The unnecessary sentence removal unit is characterized by splitting the sentence as the analysis result by the morphological analysis unit into sentences and removing a predetermined character string and a predetermined keyword from the sentences.

[0014] In the long-term care certification support system according to the present invention, the morphological analysis unit is characterized by splitting the analysis target into tokens and converting the split tokens into their original forms.

[0015] The long-term care certification support method according to the present invention Basic survey and special notes A certification survey form consisting of Input A terminal, Basic survey and special notesAn artificial intelligence unit that analyzes Receive a request from the terminal, request the artificial intelligence unit to analyze the basic survey and special notes, and return the judgment result obtained from the analysis to the terminal A server, and a method for supporting certification of nursing care requirements, using The morphological analysis unit is for special notes A step of performing morphological analysis on the analysis target in A step in which the non-essential sentence removal unit removes non-essential sentences from the analysis result by the morphological analysis unit, and a step in which the basis sentence extraction unit extracts from the analysis target from which non-essential sentences have been removed by the non-essential sentence removal unit A step of extracting sentences serving as the basis for estimation The normalization unit A step of normalizing the expression of the sentences extracted by the estimation basis sentence extraction unit For the sentence normalized by the normalization unit by the classification estimation unit, estimate which option among the options for the questions in the basic survey applies, and based on this result, generate a classification number for the corresponding question Step The answer generation unit In the classification estimation unit Generate Based on the classified number obtained, corresponding to the criteria of the certification survey form Determine the judgment result of the need for care certification A step of doing

[0016] The nursing care certification support program according to the present invention Basic survey and special notes Consisting of a certification survey form To Input Do Terminal, and Basic survey and special notes An artificial intelligence unit that analyzes Receive a request from the terminal, request the artificial intelligence unit to analyze the basic survey and special notes, and return the judgment result obtained from the analysis to the terminal A server, and a nursing care certification support program using To The artificial intelligence unit A morphological analysis step of morphologically analyzing the analysis target in the special notes, a non-essential sentence removal step of removing non-essential sentences from the analysis result by the morphological analysis step, an extraction step of extracting the sentence serving as the basis from the analysis target from which non-essential sentences have been removed by the non-essential sentence removal step, a normalization step of normalizing the expression of the sentence extracted in the extraction step, a generation step of estimating which option among the options for the questions in the basic survey applies to the sentence normalized by the normalization step, and based on this result, generating a classification number for the corresponding question, and a step of determining the judgment result of the need for care certification corresponding to the criteria of the certification survey form based on the classification number generated in the generation step Cooperates with the terminal and the server to Cause to Execute

Effect of the Invention

[0017] According to the present invention, by using artificial intelligence, the content described by the investigator in the "basic survey" and "special remarks" of the "certification survey form" is unified into a uniform one nationwide, and it is possible to estimate so as to be an accurate and objective expression regardless of the form of the question. As a result, the workload in the certification survey for nursing care requirements can be reduced and the accuracy of the determination can be improved.

Brief Description of the Drawings

[0018]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Embodiments for Carrying Out the Invention

[0019] Hereinafter, a nursing care certification support system, a nursing care certification support method, and a nursing care certification support program according to an embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a nursing care certification support system 10 according to the present embodiment. The nursing care certification support method is executed according to the procedure shown in FIGS. 3 to 6 by the nursing care certification support system 10 shown in FIG. 1. The nursing care certification support program is also executed according to the procedure shown in FIGS. 3 to 6 by the operation and control of each component included in the nursing care certification support system 10 shown in FIG. 1.

[0020] First, the configuration of the nursing care certification support system 10 will be described. In the following embodiments, an example will be described in which the "basic survey" described in FIGS. 11 to 15 and the "special notes" shown in FIGS. 16 to 18 among the "certification questionnaires" shown in FIGS. 10 to 18 are targeted.

[0021] As partially illustrated in FIG. 16, the "special notes" include questions and options for the questions. Examples of the questions include (Group 1) "Special notes on items related to physical functions and daily living activities", (Group 2) "Special notes on items related to living functions", (Group 3) "Special notes on items related to cognitive functions", (Group 4) "Special notes on items related to mental / behavioral disorders", (Group 5) "Special notes on items related to adaptation to social life", (Group 6) "Special notes on special medical care", and (Group 7) "Special notes on items related to the degree of independence in daily life" can be mentioned.

[0022] The above (Group 1) corresponds to questions 1-1 to 1-13 of the "basic survey", (Group 2) corresponds to questions 2-1 to 2-12, (Group 3) corresponds to questions 3-1 to 3-9, (Group 4) corresponds to questions 4-1 to 4-15, (Group 5) corresponds to questions 5-1 to 5-6, (Group 6) corresponds to question 6, and (Group 7) corresponds to question 7, respectively.

[0023] As options for the questions, there are matters corresponding to the questions described in the "Basic Survey". For example, for the above question (1) "Special matters regarding items related to physical functions and daily living activities", matters corresponding to each question in the "Basic Survey" include "Presence or absence of paralysis, etc." (Question 1-1), "Presence or absence of contracture" (Question 1-2), "Turning over in bed" (Question 1-3), "Getting up" (Question 1-4), "Seat holding" (Question 1-5), "Standing position on both feet" (Question 1-6), "Walking" (Question 1-7), "Standing up" (Question 1-8), "Standing position on one foot" (Question 1-9), "Body washing" (Question 1-10), "Nail cutting" (Question 1-11), "Vision" (Question 1-12), and "Hearing" (Question 1-13).

[0024] As shown in FIG. 1, the care-need certification support system 10 includes a terminal 20, a server 30, and an artificial intelligence unit 40. The terminal 20 and the server 30 are connected via a communication network 11, and the server 30 and the artificial intelligence unit 40 are connected via a closed network 12. The closed network 12 refers to a wide-area communication network within an organization that is not directly connected to the Internet or the like and connects only limited users and bases.

[0025] The terminal 20 is, for example, a smartphone, a mobile phone, a PC (Personal Computer), or a tablet, and any of them may be used as long as it can implement the functions according to this embodiment. The terminal 20 receives an input from the user, sends a request to the server 30, and performs processes such as displaying the result returned from the server 30. The terminal 20 includes a communication control unit 21, a display unit 22, an input unit 23, and a storage unit 24.

[0026] The communication control unit 21 controls the communication with the server 30 via the communication network 11. The display unit 22 can display various information in addition to the answer as the determination result, and not only display but also notify by voice or the like. The input unit 23 inputs various information by the operation of the investigator having the terminal 20, and has a keyboard function, a touch panel function, an image operation function, a camera function, etc. The storage unit 24 stores the information input from the input unit 23, the information received from the server 30, the information regarding the "basic survey" and "special notes" of the "certification survey form", etc.

[0027] Also, the terminal 20 may be composed of one or two or more (also referred to as a plurality) of terminals 20. In this case, it is advisable to provide a terminal 20 for each of one or two or more persons in charge, one or two or more business offices, etc. Further, computer peripheral devices such as a printer and a mouse (not shown) may be provided for each of these terminals 20.

[0028] The server 30 is composed of a computer having a server function, and any one may be used as long as it can exhibit the functions according to the present embodiment. The server 30 receives a request from the terminal 20, analyzes the "basic survey" and "special notes" of the "certification survey form" via the artificial intelligence unit 40, and performs a process of returning the determination result to the terminal 20. The server 30 has a communication control unit 31, a survey item registration unit 32, an answer extraction unit 33, and a storage unit 34.

[0029] The communication control unit 31 controls the communication with the terminal 20 via the communication network 11 and the communication with the artificial intelligence unit 40 via the closed network 12. The survey item registration unit 32 registers the survey items such as the "certification survey form" received from the terminal 20 in the storage unit 34. The survey item data registered by the survey item registration unit 32 is transmitted to the artificial intelligence unit 40 and stored in the storage unit 47.

[0030] When a request for displaying an answer is made to the server 30 by operating the input unit 23 of the terminal 20, the answer extraction unit 33 accesses the storage unit 47 of the artificial intelligence unit 40, calls an answer corresponding to the request, and sends the called answer to the terminal 20. The storage unit 34 stores information regarding the extraction by the answer extraction unit 33 and the like.

[0031] The artificial intelligence unit 40 is composed of a computer having an artificial intelligence (AI (Artificial Intelligence)) function, but is not limited to artificial intelligence, and any device can be used as long as it can implement the functions according to the present embodiment. The artificial intelligence unit 40 receives a request from the server 30, analyzes the "basic survey" and "special notes" of the "certification survey form", and performs a process of returning the analysis result to the server 30. The artificial intelligence unit 40 includes a communication control unit 41, a data conversion unit 42, an evidentiary statement extraction unit 43, an AI normalization unit 44 as a normalization unit, an AI estimation unit 45 as a classification estimation unit, an answer generation unit 46, and a storage unit 47.

[0032] The communication control unit 41 controls communication with the server 30 via the closed network 12. The data conversion unit 42 includes an investigation item extraction unit 421, a morphological analysis unit 422, and an unnecessary sentence removal unit 423. For the sentences in the "special notes" received from the server 30, the investigation item extraction unit 421 separates the "certification survey form" into "basic survey" and "special notes" for extraction by the evidentiary statement extraction unit 43, and further extracts data item by item such as questions and options in each of the "basic survey" and "special notes". Further, morphological analysis by the morphological analysis unit 422 and removal of unnecessary sentences by the unnecessary sentence removal unit 423 are performed on the sentences subjected to morphological analysis. The sentences subjected to these processes are output to the evidentiary statement extraction unit 43 and stored in the storage unit 47. The extraction conditions by the investigation item extraction unit 421, the conditions for morphological analysis and unnecessary sentence removal, etc. are stored in the storage unit 47, and the deletion keywords and question keywords used for unnecessary sentence removal are also stored in the storage unit 47 in advance.

[0033] The inference basis sentence extraction unit 43 extracts sentences containing the above question keywords from the data received from the data conversion unit 42 for AI inference sentence normalization. The extracted sentences are output to the AI normalization unit 44 and stored in the storage unit 47. The control conditions and the like of the inference basis sentence extraction process are stored in the storage unit 47.

[0034] The AI normalization unit 44 as a normalization unit normalizes Chinese numeral notations, Japanese fraction notations, range notations, unnecessary modifiers, and frequency notations for the sentences extracted by the inference basis sentence extraction unit 43 for classification by the AI inference unit 45. The normalized sentences are output to the AI inference unit 45 and stored in the storage unit 47. The conditions for normalization and the like are stored in the storage unit 47 in advance.

[0035] The AI inference unit 45 as a classification inference unit is configured by, for example, a language model, and infers the classification corresponding to the question in the "special note items". The inference result is output to the answer generation unit 46 and stored in the storage unit 47.

[0036] The answer generation unit 46 determines an AI inference option, that is, a care-required determination result based on a national uniform standard, based on the inference result by the AI inference unit 45, and stores this determination result in the storage unit 47 as an answer.

[0037] The communication network 11 and the closed network 12 are each configured as a communication network connecting the server 30 and the terminal 20, and the server 30 and the artificial intelligence unit 40, and are, for example, a public switched telephone network (PSTN), a mobile phone network, an IP phone network, a closed network, a wireless LAN (WiFi), and any network or other communication network that functions as such may be used.

[0038] Note that although the artificial intelligence unit 40 and the server 30 are configured separately as shown in FIG. 1, they may be configured integrally.

[0039] Next, referring to FIGS. 2 and 3, the process from when the local government receives an application for long-term care certification, analyzes and examines the application content using the long-term care certification support system 10 based on the content of the investigated application, makes a determination of long-term care, and outputs a notification of the certification result of long-term care certification will be described.

[0040] As shown in FIG. 2, first, the subject submits an application for long-term care certification to the local government (flow F10 in FIG. 2). Based on the application for long-term care certification, the local government dispatches an investigator to the location of the subject. The investigator conducts an interview survey on the physical and mental state of the subject, inputs the items of the "certification survey form" consisting of "basic survey" and "special remarks" from the input unit 23 of the terminal 20, creates the "certification survey form" as electronic data, and stores it in the storage unit 24 of the terminal 20 (flow F11 in FIG. 2, step S1 in FIG. 3).

[0041] The "certification survey form" as the above-mentioned electronic data is, for example, created by the investigator bringing a portable terminal 20 such as a tablet when going to the location of the subject and inputting it into the tablet terminal at any time while conducting the interview survey. Also, the investigator may create the "certification survey form" on paper during the interview survey and input the "certification survey form" (electronic data) from the input unit 23 of the terminal 20 installed in the local government based on the paper "certification survey form". Furthermore, the paper "certification survey form" written by the investigator may be captured from the input unit 23 of the terminal 20 by OCR (Optical Character Recognition) or the like. These are just examples, and as long as the "certification survey form" consisting of "basic survey" and "special remarks" can be captured as electronic data via the terminal 20.

[0042] Next, the content confirmation process by the long-term care certification support system 10 shown in flow F12 of FIG. 2 will be described below with reference to FIG. 3. Note that since the flows F13 to F15 in FIG. 2 are the same processes as the flows F93 to F95 in FIG. 19, the description thereof will be omitted.

[0043] First, the terminal 20 transmits, via its communication control unit 21, the electronic data of the created "certification survey form" (step S1 in FIG. 3) to the server 30 (step S2 in FIG. 3). The server 30 receives the electronic data of the "certification survey form" via its communication control unit 31 and registers each survey item in the storage unit 34 (step S3 in FIG. 3). Regarding the electronic data of the "certification survey form", the survey item extraction unit 421 divides it into "basic survey" and "special notes", and extracts it item by item in each of the "basic survey" and "special notes", and registers it in the storage unit 47 (step S4 in FIG. 3).

[0044] The artificial intelligence unit 40 receives, via its communication control unit 41, the electronic data of the "certification survey form" extracted item by item by the survey item extraction unit 421, and the data conversion unit 42 performs conversion of the data to be analyzed. Here, the data to be analyzed in this specification refers to the item-by-item data in the "basic survey" and "special notes" extracted item by item by the survey item extraction unit 421.

[0045] The data conversion unit 42 includes a morphological analysis unit 422 and an unnecessary sentence removal unit 423. The morphological analysis unit 422 performs morphological analysis on each item of the "special notes" in the received data to be analyzed (step S5 in FIG. 3), and the unnecessary sentence removal unit 423 removes unnecessary sentences from the analysis result of the morphological analysis (step S6 in FIG. 3). The morphological analysis and unnecessary part removal will be described in detail later.

[0046] For the data to be analyzed from which unnecessary sentences have been removed in the unnecessary sentence removal unit 423, the presumptive basis sentence extraction unit 43 extracts the presumptive basis sentence (step S7 in FIG. 3), the AI normalization unit 44 performs predetermined normalization (step S8 in FIG. 3), and further, classification is estimated in the AI estimation unit 45 (step S9 in FIG. 3).

[0047] The AI estimation unit 45 sends the estimated classification to the answer generation unit 46, and the answer generation unit 46 generates an answer as an option (estimated option) selected in the "basic investigation" and "special notes" based on the received data (step S10 in FIG. 3). The generated answer is stored in the storage unit 47 of the server 30. These steps (steps S7 to S10 in FIG. 3) will be described in detail later.

[0048] In response to a request from the terminal 20, the answer extraction unit 33 of the server 30 extracts the answer generated by the answer generation unit 46 and stored in the storage unit 47 (step S11 in FIG. 3), and transmits it to the terminal 20 via the communication control unit 31 and returns it (step S12 in FIG. 3). The terminal 20 receives the transmitted answer via the communication control unit 21 (step S13 in FIG. 3), and displays the content corresponding to the instruction information input from the input unit 23 on the display unit 22 (step S14 in FIG. 3).

[0049] <Unnecessary sentence removal> (step S6 in FIG. 3, FIG. 4) Referring to FIG. 4, the flow of the unnecessary sentence removal process shown in step S6 of FIG. 3 will be described. FIG. 4 also shows the morphological analysis before the unnecessary sentence removal process and the estimated basis sentence extraction process after the unnecessary sentence removal process. Here, an example of question 1-1 "presence or absence of paralysis, etc." will be given for explanation. Assume that the sentence of the special note taken out in step S4 of FIG. 3 is "Both lower limbs cannot be lifted to the horizontal. An operation was performed on the right femur 5 years ago and an artificial bone is inserted. "Both lower limbs" was selected." In the morphological analysis in the stage before text removal (step S5 in FIGS. 3 and 4), first, the morphological analysis unit 422 reads each sentence described in the "Special Note" among the data retrieved by the survey item retrieval unit 421 and stored in the storage unit 47. The morphological analysis unit 422 divides the read sentence into tokens, discriminates the part of speech of each token, and normalizes each token. A token is a sequence of characters representing one unit of meaning in a sentence or text, including punctuation marks, numbers, and other symbols that convey meaning. As normalization, for example, the conversion of a verb in hiragana notation to kanji notation and the conversion to the base form of the verb are performed. Examples of the conversion to the base form of the verb include the conversion of "not able to lift up" to "lift up + be able to + not" and the conversion of "is inside" to "enter + te + be inside". For the former, it is converted to the base form, "lift up" + "be able to" + "not", and for the latter, it is converted to the base form "enter" + "te" + "be inside". In the example of Question 1-1, the sentence "Both lower limbs cannot be lifted up to the horizontal. An operation on the right femur was performed 5 years ago, and an artificial bone is inside. Select 'both lower limbs'." is output after morphological analysis. The tokens normalized in this way and the sentence obtained by applying the normalized tokens to the sentence before division are stored in the storage unit 47.

[0050] Next, the unnecessary sentence removal unit 423 divides the sentences of the "Special Notes" after the morphological analysis by the morphological analysis unit 422 is completed (step S61 in FIG. 4), and for each of the divided sentences, removes the character strings of the options of the "Basic Survey" (step S62 in FIG. 4). Since the "Special Notes" are sentences for the current situation survey, when the options are directly described, the option character strings are removed. The character strings to be removed are the character strings shown as the options set for each question of the "Basic Survey" (FIGS. 11 to 15), and the symbols in parentheses are also removed. For example, as the options for question 1-1 "Presence or absence of paralysis, etc.", "None", "Left upper limb", "Right upper limb", "Left lower limb", "Right lower limb", "Others (limb defect)" are set. In the example of question 1-1, by sentence division in step S61 of FIG. 4, "Both lower limbs cannot be lifted up to the horizontal. Surgery was performed on the right femur 5 years ago. There is an artificial bone in it. For the selection of "both lower limbs"." is output, and by removing the option character strings in step S62 of FIG. 4, "Both lower limbs cannot be lifted up to the horizontal. Surgery was performed on the right femur 5 years ago. There is an artificial bone in it. For the selection of." is output.

[0051] Subsequently, for the sentences for which the unnecessary sentence removal unit 423 has executed the option character string removal (step S62 in FIG. 4), if they contain the preset "deletion keyword", this is removed (step S63 in FIG. 4). The "deletion keyword" is stored in the storage unit 47 of the server 30. Examples of the deletion keyword include words that are assumed to be described in the special notes of question 1-1, such as "surgery", "interview", etc., and verbs are described in the original form of the token. In the example of question 1-1, "Both lower limbs cannot be lifted up to the horizontal. 5 years ago, the right femur was performed, and there is an artificial bone in it. For the selection of." is output.

[0052] Next, the unnecessary sentence removal unit 423 determines whether a preset "question keyword" is included in the sentence for which the deletion keyword removal (step S63 in FIG. 4) has been executed (step S64 in FIG. 4). In this determination, for each sentence, if the question keyword is included, it is determined as "True" as the state expression, and if not, it is determined as "False". The "question keyword" is a word included in the questions of the "basic survey" (FIGS. 11 to 15) and the options set for each question, which has a certain meaning and is stored in the storage unit 47 in advance. As the "question keyword" for question 1-1, for example, "limb", "leg", "hand", "finger", "paralysis", "contracture", etc. can be cited, and verbs are described in the original form of the token. In the example of question 1-1, "Both lower limbs cannot be lifted to the horizontal." is determined as "True", "Performed on the right femur 5 years ago." is determined as "False", "There is an artificial bone in it." is determined as "False", and "For selection." is determined as "False".

[0053] Subsequently, the estimated basis sentence extraction unit 43 extracts the estimated basis sentence from the original text based on the sentence in which the "True" determination that the question keyword is included is made in the keyword inclusion determination (step S64 in FIG. 4) among the sentences for which the deletion keyword removal (step S63 in FIG. 4) process has been executed (step S7 in FIGS. 3 and 4). In the example of question 1-1, the original text "Both lower limbs cannot be lifted to the horizontal." of "Both lower limbs cannot be lifted to the horizontal." for which the keyword inclusion determination result in step S64 of FIG. 4 is "True" is extracted as the estimated basis sentence.

[0054] <AI Estimated Sentence Normalization> (step S8 in FIG. 3, FIG. 5) The estimated basis sentence extracted by the estimated basis sentence extraction unit 43 is output to the AI normalization unit 44 as the normalization unit. The AI normalization unit 44 executes the AI estimated sentence normalization process (see S8 in FIG. 3) shown in FIG. 5 described below to create an AI estimated sentence for subsequent AI estimation (step S9 in FIG. 3) for the sentence input from the estimated basis sentence extraction unit 43.

[0055] In the AI ​​estimated sentence normalization process, first, as described above, the estimated basis sentence extracted by the estimated basis sentence extraction unit 43 is input to the AI ​​normalization unit 44 (step S81 in FIG. 5). The AI ​​normalization unit 44 performs morphological analysis on each sentence of the input sentence, and converts the parts of speech and original forms into sequences on a token-by-token basis (step S82 in FIG. 5).

[0056] Next, the AI ​​normalization unit 44 normalizes the Chinese numerals for each token into Arabic numerals (step S83 in FIG. 5). For example, normalization is performed to convert the description "一二三四五六七八九0" into Arabic numerals "1234567890".

[0057] Next, the AI ​​normalization unit 44 normalizes the Japanese fraction notation containing a mixture of kanji and hiragana for each token into a notation containing only Arabic numerals and symbols (step S84 in FIG. 5). For example, normalization is performed to convert the notation "two-thirds" into the Arabic numeral notation "2 / 3."

[0058] Furthermore, the AI ​​normalization unit 44 performs a normalization process for each token to unify the range notation into a predetermined notation (step S85 in FIG. 5). For example, normalization is performed to convert notations such as "1, 2" or "1·2" into notation such as "1-2," which means the range from 1 to 2.

[0059] Next, the AI ​​normalization unit 44 performs a normalization process to remove predetermined unnecessary modifiers from each token (step S86 in FIG. 5). The modifiers to be removed here are modifiers that may cause estimation errors during later AI estimation (step S9 in FIG. 3), and are stored in advance in the form of a table in the storage unit 47. The AI ​​normalization unit 44 refers to this table to remove unnecessary modifiers from tokens. For example, the modifier "here" is removed from expressions such as "past week" and "past month", which are words with ambiguous meanings or words that may have multiple meanings, and normalizes the expressions to "1 week" and "1 month".

[0060] Subsequently, the AI normalization unit 44 performs normalization processing to unify the frequency expressions for each token into a predetermined format (step S87 in FIG. 5). The normalization of the frequency expressions is a unification process to the expressions defined in the "basic survey", for example, "once a week", "once a month", or expressions similar thereto used in "going out frequency" of 2-12, which are preset and stored in the storage unit 47. When there are multiple such stored expressions, rules for which expression to apply according to the frequency are also preset and stored in the storage unit 47, and based on these rules, they are unified into the optimal frequency expression. For example, expressions such as "1 to 9 times a week", "2 to 9 times every two weeks", "3 to 9 times every three weeks", "4 to 9 times every four weeks", "1 to 9 times / week", "4 to 9 times / month", "2 to 9 times every 10 days", "4 to 9 times a month", "once every 1 to 7 days" are all unified to "once a week", and expressions such as "1 to 3 times every 2 to 4 weeks", "1 to 3 times / month", "once every 10 days", "1 to 3 times a month", "once every 2 to 4 weeks", "once a month", "once a month" are all unified to "once a month", and expressions such as "1 to 9 times a day", "1 to 9 times / day" are all unified to "once a day".

[0061] In the above description, steps S83 to S87 were executed in sequence, but the order of these processes is not limited to this example.

[0062] <AI Estimation, Answer Generation> (Steps S9·S10 in FIG. 3, FIGS. 6 to 8) The text of the "Special Notes" for which the processing of AI estimation unit normalization (step S8 in FIG. 3) has ended is input to the AI estimation unit 45. The "basic survey" data retrieved in the retrieval process of the survey items is also sent to the AI estimation unit 45. The AI estimation unit 45 performs AI estimation processing on the "basic survey" and the "Special Notes" that have been processed from morphological analysis (step S5 in FIG. 3) to AI estimation sentence normalization (step S8 in FIG. 3), and based on this estimation result, the answer generation unit 46 generates an answer as an estimation option.

[0063] Figure 6 shows an example of the flow of AI estimation processing (steps S91 to S92) for the sentences described in the "Special Notes" and the subsequent answer generation processing (steps S93 to S97). In this example, first, the sentence normalized by the AI estimation sentence normalization processing (step S8 in FIG. 3) is input to the AI estimation unit 45 (step S91 in FIG. 6). The AI estimation unit 45 executes AI estimation processing for each sentence of the input sentence (step S92 in FIG. 6). This AI estimation processing is a natural language processing model pre-trained by associating sentences with classification numbers for each question, and is performed using a known model such as BERT (Bidirectional Encoder Representations from Transformers), for example. From the target sentence, it is estimated which option of the "basic survey" question it corresponds to, and based on this result, a classification number for the corresponding question is generated.

[0064] Specifically, as illustrated in FIGS. 7 and 8, the AI estimation unit 45 assigns a classification number to the target sentence. FIG. 7 shows the case where the answer to the question is single, and FIG. 8 shows the case where the answer to the question is multiple.

[0065] In the example shown in FIG. 7, for the question 2-1 "Transfer" of the "basic survey", it is estimated which option the sentence resulting from the AI estimation sentence normalization processing (step S8 in FIG. 3) corresponds to. For example, for the AI estimation sentence "Transferring by oneself while holding onto the surroundings.", it is estimated that the option "Not assisted" applies, and the number "1" corresponding to this option becomes the classification number. Also, for the sentence "A nurse is watching during wheelchair transfer.", it is estimated that the option "Watching, etc." applies, and the number "2" corresponding to this option becomes the classification number. For the question 2-1 "Transfer", the options are "Not assisted", "Watching, etc.", "Partially assisted", and "Fully assisted", and a single answer selecting one of them is required, and classification numbers 1 to 4 corresponding one-to-one to the selected option are assigned. As a result, it becomes clear which option of the "basic survey" the sentence resulting from the AI estimation sentence normalization processing corresponds to.

[0066] In the example shown in FIG. 8, for question 1-1 of the "basic survey", "Presence or absence of paralysis, etc.", it is estimated which option the sentence in the result of the AI estimation sentence normalization process (step S8 in FIG. 3) corresponds to. Multiple answers are allowed for question 1-1.

[0067] For example, for the AI estimation sentence "Both upper and lower limbs can perform the specified movements", it is estimated that the option "None" applies, and the number "1" corresponding to this option becomes the classification number. Also, for the sentence "Both upper limbs cannot perform the specified movements", it is estimated that the two options "Left upper limb" and "Right upper limb" apply, and the number "2" corresponding to "Left upper limb" and the number "4" corresponding to "Right upper limb" are added together, and "6" becomes the classification number.

[0068] For question 1-1, "Presence or absence of paralysis, etc.", for the six options, they are assigned to the bits of a 6-bit binary array of 1, 2, 4, 8, 16, 32, and the sum of the coefficients of each bit is set as the classification number. Even when there are two or more multiple answers, non-overlapping classification numbers can be assigned. As a result, even in the case of multiple answers, it becomes clear which option of the "basic survey" the sentence in the result of the AI estimation sentence normalization process corresponds to. Note that in this embodiment, an example of a 6-bit array is given, but the bit array can be arbitrarily set according to the number of options.

[0069] After classification numbers are assigned to each sentence of the text that has undergone the AI estimation sentence normalization process as described above, the answer generation unit 46 executes an answer generation process (steps S93 to S97 in FIG. 6) for generating an answer. In this answer generation process, first, it is determined whether the question for the classified sentence is a single answer or a multiple answer (step S93 in FIG. 6). If it is a single answer, the classification number is determined as the AI estimation option (step S94 in FIG. 6).

[0070] On the other hand, in the case of multiple responses, the assigned classification number is converted into a set of numbers corresponding to each of the multiple selected options, and the AI estimated option is determined as these numbers (step S95 in FIG. 6). For example, in the example shown in FIG. 8, when the classification number is 30, since the options are "left upper limb", "right upper limb", "left lower limb", and "right lower limb", it is converted into a set of numbers "2", "4", "8", "16" corresponding to these options.

[0071] The AI estimated options respectively determined in steps S94 and S95 above are generated as answers to each sentence of the "Special Notes" (step S94 in FIG. 6). This answer is stored in the storage unit 47 and transmitted to the server 30 via the communication control units 31 and 41 and stored in the storage unit 34 (step S97 in FIG. 6).

[0072] Here, regarding the "basic survey", in addition to being input from the input unit 23 of the terminal 20, the answer content is recognized as text from the PDF data of the basic survey (see FIGS. 11 to 15) and stored in the storage unit 24. The AI estimation unit 45 stores in the storage unit 47, as a table, the answers generated for the "Special Notes" corresponding to each question of the "basic survey". The stored table can be displayed on the display unit 22 of the terminal 20 via the communication network 11 and the closed area network 12. Since the answers to each question of the "basic survey" and the answers in the "Special Notes" can be compared, it is possible to easily confirm the inconsistencies in the corresponding questions, easily recognize and correct the inconsistencies before the secondary determination in the flow F14 of FIG. 2, and thus contribute to improving the efficiency of the review by the review committee.

[0073] <Multi - process processing> The care - need certification support system 10 is configured to be capable of multi - process processing. The multi - process processing is controlled, for example, by the control function of the server 30, and can perform batch processing for the same question of the "Special Notes" in a plurality of "certification survey forms". Thereby, it becomes possible to efficiently perform the series of processes shown in FIG. 3 for a plurality of "certification survey forms".

[0074] Figure 9 shows an overview of the batch processing flow for the same question, for example, question 1-1, in the "Special Notes" of six "Certification Questionnaires". In Figure 9, the "Certification Questionnaire" is abbreviated as the "Questionnaire". Also, Figure 9 gives an example of the processing for question 1-1, but multi-process processing is also possible for other questions. Furthermore, the number of processes that can be batch-processed is not limited to the six illustrated in Figure 9, and can be 1 to 5, or 7 or more.

[0075] In the example shown in Figure 9, when the texts for the same question 1-1 of questionnaires 1 to 6 are input in batch from the input unit 23 or read out in batch from the storage unit 24 (step S101 in Figure 9), a series of processes shown in Figure 3 are processed in multi-process in the server 30 and the artificial intelligence unit 40. For each questionnaire, AI estimation (step S9 in Figure 3) is performed and an answer is generated (steps S102 and S103 in Figure 9). At the terminal 20, the answers generated for each questionnaire are returned all at once, and the determination result is displayed on the display unit 22 (step S104 in Figure 9).

[0076] According to the care-need certification support system, care-need certification support method, and care-need certification support program according to this embodiment, by the processing shown in Figure 3, an answer that is a uniform and objective expression nationwide corresponding to the questions and options of the "Basic Survey" can be generated from the text described in the "Special Notes". Therefore, it is possible to improve the efficiency while ensuring the quality of the confirmation work of the "Certification Questionnaire", and to enhance the care-need certification system.

[0077] Regarding the consistency between the "basic survey" and the "special notes", a method of decomposing and analyzing the response sentences described in the "special notes" through artificial intelligence and comparing them with the responses of the "basic survey" can be considered. However, since there are fluctuations in the descriptions by the investigators remaining in the analysis results of the response sentences in the "special notes", it is difficult to lead to the efficiency improvement of the comparison work with the "basic survey" in the review meeting. In contrast, in this embodiment, since the content of the "special notes" is digitized and classified according to the "basic survey", the comparison with the responses of the "basic survey" can be accurately performed in a short time, so the work efficiency of the examiner can be improved.

[0078] This invention can be embodied in many forms without departing from its essential characteristics. Therefore, it goes without saying that the above-described embodiments are for illustrative purposes only and do not limit the present invention.

Explanation of Signs

[0079] 10 Care-need Certification Support System 11 Communication Network 12 Closed Domain Network 20 Terminal 21 Communication Control Unit 22 Display Unit 23 Input Unit 24 Storage Unit 30 Server 31 Communication Control Unit 32 Survey Item Registration Unit 33 Answer Extraction Unit 34 Storage Unit 40 Artificial Intelligence Unit 41 Communication Control Unit 42 Data Conversion Unit 421 Survey Item Extraction Unit 422 Morphological Analysis Unit 423 Unnecessary Sentence Removal Unit 43 Presumption Basis Sentence Extraction Unit 44 AI Normalization Unit (Normalization Unit) 45 AI Presumption Unit (Presumption Unit) 46 Answer Generation Unit 47 Storage Unit

Claims

1. A terminal for inputting a certification survey form consisting of a basic survey and specific notes, an artificial intelligence unit for analyzing the basic survey and the specific notes, a server that receives a request from the terminal, requests the artificial intelligence unit to analyze the basic survey and the specific notes, and returns the determination result obtained from the analysis to the terminal, wherein the nursing care certification support system includes: the artificial intelligence unit includes: a morphological analysis unit that performs morphological analysis on the analysis target in the specific notes, an unnecessary sentence removal unit that removes unnecessary sentences from the analysis result by the morphological analysis unit, a presumption basis sentence extraction unit that extracts sentences serving as presumption bases from the analysis target from which unnecessary sentences have been removed by the unnecessary sentence removal unit, a normalization unit that normalizes the expression of the sentences extracted by the presumption basis sentence extraction unit, a classification presumption unit that presumes which of the options for the questions in the basic survey the sentence normalized by the normalization unit corresponds to, and generates a classification number for the corresponding question based on this result, and an answer generation unit that determines a determination result of nursing care certification corresponding to the criteria of the certification survey form based on the classification number generated by the classification presumption unit. A nursing care certification support system characterized by the above.

2. The unnecessary sentence removal unit divides the sentence as the analysis result by the morphological analysis unit into sentences, and removes a predetermined character string and a predetermined keyword from the sentences. The nursing care certification support system according to claim 1, characterized by the above.

3. The morphological analysis unit divides the analysis target into tokens and converts the divided tokens into their original forms. The nursing care certification support system according to claim 2, characterized by the above.

4. A nursing care certification support method using a terminal for inputting a certification survey form consisting of a basic survey and specific notes, an artificial intelligence unit for analyzing the basic survey and the specific notes, and a server that receives a request from the terminal, requests the artificial intelligence unit to analyze the basic survey and the specific notes, and returns the determination result obtained from the analysis to the terminal, the method including: a step in which a morphological analysis unit performs morphological analysis on the analysis target in the specific notes, a step in which an unnecessary sentence removal unit removes unnecessary sentences from the analysis result by the morphological analysis unit, a step in which a presumption basis sentence extraction unit extracts sentences serving as presumption bases from the analysis target from which unnecessary sentences have been removed by the unnecessary sentence removal unit, a step in which a normalization unit normalizes the expression of the sentences extracted by the presumption basis sentence extraction unit, A step of estimating which option among the options for the questions of the basic survey the classification estimation unit corresponds to for the sentence normalized by the normalization unit, and generating a classification number for the corresponding question based on this result; A step of the answer generation unit determining a judgment result of the need for care certification corresponding to the standard of the certification survey form based on the classification number generated by the classification estimation unit, Characterized by a method for supporting the determination of the need for care certification.

5. A terminal for inputting a certification survey form consisting of a basic survey and special notes, an artificial intelligence unit for analyzing the basic survey and the special notes, receiving a request from the terminal, requesting the artificial intelligence unit to analyze the basic survey and the special notes, and returning the judgment result by the analysis to the terminal. A program for supporting the determination of the need for care certification using a server, In the artificial intelligence unit, in cooperation with the terminal and the server, A morphological analysis step of performing morphological analysis on the analysis target in the special notes; An unnecessary sentence removal step of removing unnecessary sentences from the analysis result of the morphological analysis step; An extraction step of extracting a sentence serving as an estimation basis from the analysis target from which unnecessary sentences have been removed by the unnecessary sentence removal step; A normalization step of normalizing the expression of the sentence extracted in the extraction step; For the sentence normalized by the normalization step, estimating which option among the options for the questions of the basic survey it corresponds to, and generating a classification number for the corresponding question based on this result; A step of determining a judgment result of the need for care certification corresponding to the standard of the certification survey form based on the classification number generated in the generation step, characterized in that a program for supporting the determination of the need for care certification is executed.

Images