JP2020181568A - Information processing device, information processing method, and information processing program - Google Patents

Abstract

To make it possible to accurately acquire answers written in a medical questionnaire.SOLUTION: A medical examination support device 1 according to one embodiment of the present invention comprises: a layout storage unit 121 to store a layout of a medical questionnaire in association with identification information for identifying the medical questionnaire; a candidate word storage unit 122 to store multiple candidate words; an entry content acquisition unit 111 to acquire the identification information from an image of the medical questionnaire and, on the basis of the layout associated with the acquired identification information, acquire one or more characters included in the image; a narrowing unit 112 to narrow the multiple candidate words on the basis of positions where the one or more characters have been acquired; an answer specification unit 113 to specify answers written in the medical questionnaire by comparing the one or more characters acquired by the entry content acquisition unit 111 to the multiple candidate words narrowed by the narrowing unit 112; and an output unit 114 to output the answers specified by the answer specification unit 113.SELECTED DRAWING: Figure 4

Description

The present invention relates to an information processing device, an information processing method, and an information processing program for acquiring an answer entered in a questionnaire.

A system is known in which a scanner reads paper to acquire image data, and then performs character recognition processing such as OCR (Optical Character Recognition) processing on the image data to acquire the contents written on the paper. (See, for example, Patent Document 1). Such a system can facilitate subsequent processing by storing the contents written on the paper in a storage device or displaying them on a display device.

Japanese Unexamined Patent Publication No. 2012-9000

By the way, when a patient (examinee) receives medical treatment at a hospital, he / she fills in an answer such as his / her symptoms on a questionnaire (also called a questionnaire). When obtaining the answer entered in the questionnaire by the character recognition process, there may be an error due to misrecognition of the character recognition process, or the patient may make an error such as a typographical error or omission when entering the answer. is there. In particular, in the medical field, the characters entered may be rough because the patient has a disability, and because there are many elderly people, the patient himself remembers the drug name, so the drug name entered is incorrect. There may be. When medical professionals such as doctors and nurses refer to the answers obtained by the character recognition process, it is burdensome to manually correct such errors.

The present invention has been made in view of the above points, and an object of the present invention is to provide an information processing device, an information processing method, and an information processing program capable of obtaining answers entered in a questionnaire with high accuracy.

The information processing device according to the first aspect of the present invention includes a layout storage unit that stores the layout of the questionnaire in association with the identification information that identifies the questionnaire, a candidate word storage unit that stores a plurality of candidate words, and the like. An acquisition unit that acquires the identification information from the image of the questionnaire and acquires one or more characters included in the image based on the layout associated with the acquired identification information, and the one or more characters. By comparing the narrowing-down unit that narrows down the plurality of candidate words based on the acquired positions with the one or more characters acquired by the acquisition unit and the plurality of candidate words narrowed down by the narrowing-down unit. , A specific unit that specifies the answer entered in the questionnaire, and an output unit that outputs the answer specified by the specific unit.

The specific unit calculates the similarity between the one or more characters acquired by the acquisition unit and each of the plurality of candidate words narrowed down by the narrowing unit, and selects based on the calculated similarity. One of the plurality of candidate words may be specified as the answer.

The information processing device further has a similar information storage unit that stores a value indicating similarity between two characters, and the specific unit is included in one or more characters acquired by the acquisition unit. The value between a character or a combination of a plurality of characters and one character or a combination of a plurality of characters included in each of the plurality of candidate words is acquired, and the similarity is calculated using the acquired value. You may.

The layout storage unit stores the layout indicating the position of the answer column included in the questionnaire based on the predetermined position, and the acquisition unit is located at the position of the answer column indicated by the layout in the image. One or more characters may be acquired.

The candidate word storage unit stores the plurality of candidate words in association with the answer column, and the narrowing unit is associated with the answer column in which one or more characters are acquired from the candidate word storage unit. The plurality of candidate words may be narrowed down by acquiring a plurality of candidate words.

The candidate word storage unit stores the plurality of candidate words in association with medical terms, and the acquisition unit acquires the one or more characters in the image in addition to the one or more characters. By acquiring the medical term associated with the medical term, and acquiring the plurality of candidate words associated with the medical term acquired by the acquisition unit from the candidate word storage unit, the narrowing unit obtains the plurality of candidate words. Candidate words may be narrowed down, and the output unit may output the answer in association with the medical term.

The layout storage unit may store the question described in the questionnaire in association with the medical term, and the output unit may output the question in addition to the answer and the medical term.

The narrowing section is based on the answer specified by the specific section for the first answer column at the position where the one or more characters are acquired, and the second answer column different from the first answer column. , The plurality of candidate words may be narrowed down.

The candidate word storage unit is in at least one state of each of the plurality of candidate words, one word state, one word divided into a plurality of parts, and a state in which a plurality of words are combined. You may remember.

The acquisition unit identifies a line segment connecting two predetermined symbols included in the image, and the answer is entered in the questionnaire by multiplying the length of the line segment by a predetermined coefficient. The length of the sides that partition the potential region may be calculated and the image corrected based on the length.

In the information processing method of the second aspect of the present invention, a layout storage unit that stores the layout of the questionnaire in association with the identification information that identifies the questionnaire, a candidate word storage unit that stores a plurality of candidate words, and a candidate word storage unit. A step of acquiring the identification information from the image of the questionnaire and acquiring one or more characters included in the image based on the layout associated with the acquired identification information. Based on the position where the above characters are acquired, the step of narrowing down the plurality of candidate words is compared with the one or more characters acquired by the acquisition step and the plurality of candidate words narrowed down by the narrowing down step. Thereby, it has a step of specifying the answer entered in the questionnaire and a step of outputting the answer specified by the specifying step.

The information processing program according to the third aspect of the present invention includes a layout storage unit that stores the layout of the questionnaire in association with the identification information that identifies the questionnaire, a candidate word storage unit that stores a plurality of candidate words, and a candidate word storage unit that stores a plurality of candidate words. The step of acquiring the identification information from the image of the questionnaire and acquiring one or more characters included in the image based on the layout associated with the acquired identification information, and the above-mentioned one. Based on the position where the above characters are acquired, the step of narrowing down the plurality of candidate words is compared with the one or more characters acquired by the acquisition step and the plurality of candidate words narrowed down by the narrowing down step. By doing so, the step of specifying the answer entered in the questionnaire and the step of outputting the answer specified by the specifying step are executed.

According to the present invention, there is an effect that the answer entered in the questionnaire can be obtained with high accuracy.

It is a schematic diagram of the medical care support system which concerns on embodiment. It is a schematic diagram of the medical care support method executed by the medical care support system which concerns on embodiment. It is a schematic diagram of the method of specifying an answer based on the character acquired from an image. It is a block diagram of the medical care support system which concerns on embodiment. It is a front view of an exemplary questionnaire. It is a schematic diagram of the method in which the entry content acquisition part specifies the position of the answer area. It is a schematic diagram of the narrowing down process by a narrowing down part. It is a schematic diagram of the matrix which shows the distance between characters. It is a front view of the display part of a medical treatment terminal. It is a figure which shows the flowchart of the medical care support method which concerns on embodiment.

[Outline of medical care support system S]
FIG. 1 is a schematic diagram of a medical care support system S according to the present embodiment. The medical care support system S includes a medical care support device 1, a medical care terminal 2, and a reading device 3. The medical care support system S may include other devices such as servers and terminals.

The medical care support device 1 (information processing device) is a computer that acquires the patient's answer entered in the questionnaire 31 (questionnaire) and provides the answer to the medical care terminal 2. The questionnaire 31 is a set of papers bundled by a fixture F such as a stapler, and each paper contains a question to the patient and an answer column for the patient to enter an answer to the question.

The medical care terminal 2 is a computer that displays information to medical staff such as doctors and nurses and accepts input by the medical staff, for example, a personal computer having a display unit 21 and an operation unit 22. The display unit 21 includes a display device such as a liquid crystal display. The operation unit 22 includes an operation device such as a keyboard and a mouse. The medical care terminal 2 receives information for display from the medical care support device 1 via a network N such as the Internet or a local area network, and transmits input by a medical worker to the medical care support device 1.

The reading device 3 is a scanner that reads the surface of the paper questionnaire 31 as an image. The reading device 3 includes an image pickup device such as a CCD (Charge Coupled Device) or a CMOS (Complementary Metal Oxide Semiconductor) capable of reading the surface of paper and outputting image data. The reading device 3 transmits the image of the questionnaire 31 to the medical care support device 1 via the network N.

The reading device 3 may be a digital camera or a mobile terminal such as a smartphone having a camera. In this case, the medical staff or the patient reads the surface of the questionnaire 31 as an image by photographing the front surface of the questionnaire 31 using a digital camera or a mobile terminal.

The medical care support system S may have a patient terminal (for example, a tablet terminal) for the patient to input an answer to the questionnaire in place of or in addition to the reading device 3. In this case, the patient fills in the answer by handwriting the patient terminal displaying the electronic questionnaire using a finger or a stylus. The patient terminal transmits the information of the handwritten input performed by the patient on the electronic questionnaire 31 to the medical care support device 1 as an image of the questionnaire 31 via the network N.

FIG. 2 is a schematic diagram of a medical care support method (information processing method) executed by the medical care support system S according to the present embodiment. First, the patient fills in the answer to the question on the questionnaire 31 on the questionnaire 31. The medical staff or the patient cuts a part of the questionnaire 31 including the fixture F and divides the questionnaire 31 into sheets of paper (a). Then, the medical staff or the patient causes the reading device 3 to read the questionnaire 31 (b). The reading device 3 transmits the image of the questionnaire 31 to the medical care support device 1.

Next, the medical care support device 1 identifies the questionnaire 31 based on the image of the questionnaire 31. The questionnaire 31 includes different question and answer columns for each questionnaire 31 (type of questionnaire 31). The questionnaire 31 represents a code (for example, a two-dimensional bar code) indicating its own identification information. The medical care support device 1 can identify the questionnaire 31 by detecting the code from the image.

Next, the medical care support device 1 acquires the layout information of the specified questionnaire 31 from the storage unit. The storage unit of the medical care support device 1 stores the layout information of the questionnaire 31 in advance. The layout information indicates the position of each answer column included in the questionnaire 31. Then, the medical care support device 1 acquires characters included in the image by performing character recognition processing such as OCR processing on a predetermined range including the position of the answer column indicated by the layout information on the image. The medical care support device 1 manages one or more characters acquired for each answer column in association with the answer column as a group of characters. Next, the medical care support device 1 identifies the answer based on the characters acquired from the image by the character recognition process (c). Then, the medical care support device 1 stores the specified answer in the storage unit and displays it on the display unit.

FIG. 3 is a schematic diagram of a method of identifying an answer based on characters obtained from an image. The medical care support device 1 narrows down the candidate words stored in the storage unit based on the position where the group of characters was acquired by the character recognition process (that is, which answer column the group of characters was entered in) (c1). ). The storage unit of the medical care support device 1 stores in advance a plurality of candidate words that may be entered in the questionnaire 31.

Depending on the question written on the questionnaire 31, the candidate words that may be entered in the answer column corresponding to the question differ. For example, in the case of a question about time, a word indicating time is entered in the answer column, and in the case of a question about a drug, a word such as a drug name is entered in the answer column. Therefore, the medical care support device 1 narrows down the candidate words based on which answer column contains the group of characters. In this way, the medical care support device 1 narrows down the candidate words that may be entered in the position (answer column) where the group of characters is acquired, so that the processing time required to identify the answer can be shortened and the answer can be answered. The accuracy of identifying is improved. At this time, the medical care support device 1 narrows down the candidate words based on the learning results learned in advance for each position in the answer column. The medical care support device 1 performs machine learning using the learning data and the teacher data prepared for each answer column. The storage unit of the medical care support device 1 stores the learning result of machine learning in advance. The medical care support device 1 narrows down candidate words based on the characters read from the questionnaire, the position where the characters are acquired (answer column), and the learning result stored in the storage unit. In this way, the medical care support device 1 identifies the answer in order to narrow down the candidate words that the group of characters acquired from the image may indicate by using the learning result of the machine learning performed for each answer column. It is possible to shorten the processing time required for the process and improve the accuracy of identifying the answer.

Finally, the medical care support device 1 calculates the similarity based on the inter-character distance between each character included in the group of characters read from the questionnaire and each character contained in each of the narrowed down candidate words. The answer is identified using the calculated similarity (c2). The inter-character distance is a value indicating the appearance similarity between two characters. The storage unit of the medical care support device 1 stores in advance a matrix indicating the distance between characters for each combination of characters. It should be noted that narrowing down is not an indispensable process, and the medical care support device 1 answers the candidate words as they are without narrowing down if the similarity of the narrowing down candidates does not meet the predetermined conditions in the evaluation at the time of narrowing down. May be specified as.

The medical care support device 1 calculated and calculated the similarity based on the inter-character distance between each character included in a group of characters read from the questionnaire and each character included in each of the narrowed down candidate words. A candidate word whose similarity satisfies a predetermined condition is specified as an answer. In this way, the medical care support device 1 identifies the answer corresponding to the group of characters acquired from the image from the candidate words in consideration of the similarity between the characters, so that the answer entered in the questionnaire is highly accurate. Can be specified.

As described above, the medical care support system S according to the present embodiment has an error due to an erroneous recognition of the character recognition process for the image of the questionnaire, or even if the patient makes an error when entering the answer. , The answers entered in the questionnaire can be obtained with high accuracy and in a short processing time.

[Configuration of medical care support system S]
FIG. 4 is a block diagram of the medical care support system S according to the present embodiment. In FIG. 4, the arrows indicate the main data flows, and there may be data flows other than those shown in FIG. In FIG. 4, each block shows a functional unit configuration, not a hardware (device) unit configuration. Therefore, the block shown in FIG. 4 may be mounted in a single device, or may be mounted separately in a plurality of devices. Data transfer between blocks may be performed via an arbitrary means such as a data bus, a network, or a portable storage medium.

The medical care support device 1 (information processing device) has a control unit 11 and a storage unit 12. The control unit 11 includes an entry content acquisition unit 111, a narrowing unit 112, a response identification unit 113, and an output unit 114. The storage unit 12 includes a layout storage unit 121, a candidate word storage unit 122, a similar information storage unit 123, and an answer storage unit 124.

The storage unit 12 is a storage medium including a ROM (Read Only Memory), a RAM (Random Access Memory), a hard disk drive, and the like. The storage unit 12 stores in advance the program executed by the control unit 11. The layout storage unit 121, the candidate word storage unit 122, the similar information storage unit 123, and the answer storage unit 124 may each be a storage area on the storage unit 12, or are a database configured on the storage unit 12. May be good.

The layout storage unit 121 stores in advance layout information indicating the position of each answer column included in the questionnaire 31 in association with the identification information of the questionnaire 31. The candidate word storage unit 122 stores in advance a plurality of candidate words that may be entered in the questionnaire 31. The similar information storage unit 123 stores in advance a matrix indicating the inter-character distance of each combination of characters as similar information. The answer storage unit 124 stores the patient's answer identified by the medical care support device 1 in association with the patient's identification information.

The control unit 11 is, for example, a processor such as a CPU (Central Processing Unit), and by executing a program stored in the storage unit 12, the entry content acquisition unit 111, the narrowing unit 112, the answer identification unit 113, and the output unit Functions as 114. At least some of the functions of the control unit 11 may be performed by an electric circuit. Further, at least a part of the functions of the control unit 11 may be executed by a program executed via the network.

The medical care support system S according to the present embodiment is not limited to the specific configuration shown in FIG. For example, the medical care support device 1 and the medical care terminal 2 may be integrated into one device.

[Explanation of the process of acquiring characters from the questionnaire 31]
Among the medical treatment support methods (information processing methods) executed by the medical treatment support system S, the process of acquiring characters from the questionnaire 31 will be described below. FIG. 5 is a front view of an exemplary questionnaire 31. Questionnaire 31 represents the answer area 311. The questionnaire 31 includes a plurality of papers each representing a different answer area 311. The plurality of papers are bundled by a fixture F such as a stapler. If the questionnaire 31 is a sheet of paper, the fixture F may be omitted.

The answer area 311 contains one or more questions 312, one or more answer columns 313 for the patient to fill in the answer to the question 312, and the medical term headings 314 associated with the questions 312 and answer columns 313. Including. The answer area 311 may include one answer column 313 for one question 312, or may include a plurality of answer columns 313. The response area 311 is not limited to the patient, and may include a column for the medical staff to enter information.

Question 312 is a character string representing a question to the patient. The answer column 313 is a column for the patient to enter an answer to the question 312. The answer column 313 may be a selection column such as a check box in which the patient selects the answer content, or may be a free entry field in which the patient enters characters representing the answer content.

Heading 314 is a physician-interpreted medical term that describes the content of question 312 and answer column 313. Heading 314 is located near question 312 and answer column 313. There is often a discrepancy between the words that the patient can understand and the medical terms, and if the medical terms are listed in Question 312, the patient may not understand them. Therefore, question 312 is often written in plain language. As a result, it takes time for the medical staff to convert the simple expressions into medical terms when medically understanding the contents of the question 312 and the answer column 313.

Therefore, by displaying the medical term heading 314 associated with the question 312 and the answer column 313 on the questionnaire 31, the medical staff does not need to convert the plain expression into the medical term, and the question 312 and the answer column 313 do not need to be converted. You can shorten the time to understand the contents medically. For example, Question 312, which corresponds to the medical term "history", is written as "Have you ever been ill?", But for healthcare professionals, the string "history" is shorter. There is a merit that the contents can be understood in an instant. Further, by associating the answer column 313 with the medical terminology, the medical staff can quickly check whether the answer column 313 is entered correctly.

In addition, the question 312 corresponding to the medical term "drug history" is written as "what drug is currently being taken?", But the drug name is described in the answer column 313. The number of drugs on the market in Japan is limited to about 200,000. Therefore, as described later, the solution is solved by registering the candidate word of the drug name in association with the answer column 313 of the "drug history" associated with the answer column 313 and incorporating the frequency information into the learning. It becomes possible to narrow down, improve the accuracy for identifying the answer, and shorten the processing time required to identify the answer. At the same time, the short medical term "drug history" is included when looking at the scan results of the questionnaire and the questionnaire, so that the medical staff can instantly understand where and what kind of information is available. In particular, there are many katakana in drug names, so in character recognition processing, so (so) and n (n), shi (shi) and tsu (tsu), tsu (uppercase) and tsu (lowercase), pa (pa) Misrecognition such as ba is likely to occur. Conventionally, it is difficult to improve the accuracy of these handwritten inputs, and mistaken drug names can immediately affect the life and death of patients. Therefore, the accuracy of character recognition processing can be sufficiently improved, and medical staff can intuitively improve it. It was important to be able to understand. Since the medical care support system S according to the present embodiment describes the perception of medical words, it is possible for the medical staff to intuitively perform the check, and at the same time, it is possible to improve the accuracy of identifying the answer.

Further, the questionnaire 31 includes a plurality of symbols 315 used for positioning the answer area 311 and a code 316 for identifying the questionnaire 31. The plurality of symbols 315 are arranged at predetermined positions outside the response area 311. The plurality of symbols 315 are symbols having a predetermined shape such as a circle, a rectangle, and a polygon, and may have the same shape or different shapes. Further, the plurality of symbols 315 may be four corners such as an area surrounded by a line or intermediate points thereof. The code 316 is, for example, a barcode or a two-dimensional code in which the identification information of the questionnaire 31 (ID of the questionnaire 31 or the like) is encoded. The medical care support device 1 can specify the positions of the answer area 311 and the answer column 313 included in the questionnaire 31 based on the symbol 315 and the code 316 included in the image of the questionnaire 31 as described later.

The questionnaire 31 shown in FIG. 5 is an example, and may have other layouts or may include other information.

The patient fills in the answer column 313 with the answer to the question 312 on the questionnaire 31. Next, the medical staff or the patient cuts a part of the questionnaire 31 including the fixture F, and divides the questionnaire 31 into sheets of paper. Then, the medical staff or the patient has the reading device 3 read the questionnaire 31. The reading device 3 reads the questionnaire 31 answered by the patient and transmits the image of the questionnaire 31 to the medical care support device 1.

When the patient's answer is accepted using the patient's terminal, the patient's terminal displays the contents of the questionnaire 31 on the display unit. The patient fills in the answer by handwriting with a finger or a stylus on the patient terminal displaying the electronic questionnaire. The patient terminal transmits the handwritten input information (for example, the trajectory of the handwritten input) performed by the patient on the electronic questionnaire 31 to the medical care support device 1 as an image of the questionnaire 31. As the image of the questionnaire 31, vector data indicating the input made by the patient may be used.

In the medical care support device 1, first, the entry content acquisition unit 111 detects the symbol 315 and the code 316 in the image of the received questionnaire 31. Here, the information associated with the code 316 is associated with the order information of which page is read first, which is easier for the doctor to read, and the entry content acquisition unit 111 uses the questionnaire 31 based on the order information. You may rearrange a plurality of pages included in. As a result, even if a plurality of pages included in the questionnaire 31 are scanned in an erroneous order, the plurality of pages can be rearranged in the correct order. Then, the entry content acquisition unit 111 acquires the identification information of the questionnaire 31 indicated by the detected code 316, and identifies the position of the answer area 311 based on the detected symbol 315.

6 (a) and 6 (b) are schematic views of a method in which the entry content acquisition unit 111 specifies the position of the response area 311. FIG. 6A shows a partially cut-out questionnaire 31, and FIG. 6B shows an uncut questionnaire 31. In the state of FIG. 6A, the entry content acquisition unit 111 does not detect the symbol 315 on the upper left, but detects the symbol 315 on the center and the symbol 315 on the upper right on the questionnaire 31.

The entry content acquisition unit 111 identifies the vector V1 (line segment) from the symbol 315 on the upper right to the symbol 315 on the center. Then, the entry content acquisition unit 111 specifies the position of the upper side E by calculating the length of the upper side E that divides the response area 311 by multiplying the length of the vector V1 by a predetermined coefficient. Specifically, the entry content acquisition unit 111 sets an area having the width obtained by multiplying the length of the vector V1 by a coefficient from the same position as the symbol 315 in the horizontal direction of the questionnaire 31 as the width of the answer area 311. Identify the position in the horizontal direction. The coefficient is preset in the layout storage unit 121 of the medical care support device 1. The coefficient may be set for each identification information of the questionnaire 31, or may be set in common for all the questionnaires 31.

For example, if the symbol 315 on the center is located above the midpoint of the upper side E of the answer area 311 and the symbol 315 on the upper right is located above the right edge of the upper side E of the answer area 311, the coefficient is It is 2. The symbol 315 on the center is placed above the right point of the two points that divide the upper side E of the answer area 311 into three equal parts, and the symbol 315 on the upper right is placed above the right end of the upper side E of the answer area 311. If so, the coefficient is 3.

The entry content acquisition unit 111 multiplies the length of the vector V1 from the symbol 315 on the upper right to the symbol 315 on the center by these coefficients set for the questionnaire 31, and then adds the answer area 311 and the symbol 315. The position of the upper side E of the answer area 311 can be specified by moving a predetermined value corresponding to the distance between the two.

In the state of FIG. 6B, the entry content acquisition unit 111 detects the symbol 315 on the upper left, the symbol 315 on the center, and the symbol 315 on the upper right on the questionnaire 31. The entry content acquisition unit 111 specifies the vector V2 (line segment) from the symbol 315 on the upper right to the symbol 315 on the upper left. Then, the entry content acquisition unit 111 can specify the position of the upper side E of the answer area 311 by moving the vector V2 by a predetermined value corresponding to the distance between the answer area 311 and the symbol 315.

In the questionnaire 31, the symbol 315 on the upper left (that is, the symbol 315 that may be cut off) may be omitted from the beginning. In this case, the entry content acquisition unit 111 determines the length of the vector V1 between the symbol 315 on the center and the symbol 315 on the upper right regardless of whether or not a part of the questionnaire 31 is cut off. The position of the upper side E of the answer area 311 is specified by multiplying the coefficients.

Here, the method of specifying the position of the upper side that divides the answer area 311 by the entry content acquisition unit 111 has been described, but the positions of the left side, the right side, and the lower side that divide the answer area 311 can also be specified by the same method.

The entry content acquisition unit 111 corrects the received image of the questionnaire 31 based on the position of the specified answer area 311. As a result, the entry content acquisition unit 111 can correct the distortion and inclination of the image generated when the reading device 3 reads the questionnaire 31, and can acquire the characters entered in the answer column 313 with high accuracy. ..

In this way, the entry content acquisition unit 111 has a symbol at a position (for example, the upper left of the questionnaire 31) where a part of the plurality of symbols 315 is missing or may be cut due to cutting the fixture F or the like. Even when the 315 is not arranged, the position of the response area 311 can be specified and the image can be corrected. The entry content acquisition unit 111 is not limited to the specific method shown here, and may specify the position of the response area 311 based on the position of the symbol 315 by another method.

Next, the entry content acquisition unit 111 acquires the content entered in the answer column 313 based on the image of the questionnaire 31. The entry content acquisition unit 111 acquires the items selected in the answer column 313, which is a selection column such as a check box, as the entry content, and also uses a group of characters entered in the answer column 313, which is a free entry column, as the entry content. get. Since the reading content of the free entry field is likely to contain errors, the method of identifying the patient's answer based on the group of characters entered in the answer field 313, which is the free entry field, will be described below.

The entry content acquisition unit 111 acquires the layout information associated with the identification information of the questionnaire 31 indicated by the code 316 detected from the image from the layout storage unit 121. The layout information indicates the relative position of each answer column 313 included in the questionnaire 31 with reference to a predetermined position on the questionnaire 31. The predetermined position as a reference may be, for example, the position of the symbol 315 on the upper right, the position of the code 316, or the position of any apex of the answer area 311. The position of the answer column 313 is represented by, for example, the coordinate range of the answer column 313 in the image of the questionnaire 31. Further, the layout information may indicate the relative position of each answer column 313 in the order of the plurality of answer columns 313 with reference to a predetermined position (for example, the left end or the upper end) on the questionnaire 31.

The entry content acquisition unit 111 takes out an image including the position of each answer column 313 indicated by the layout information in the image of the questionnaire 31. The entry content acquisition unit 111 acquires a group of characters (one or more characters) entered in the answer column 313 by applying a known character recognition process such as OCR to the image extracted for each answer column 313. To do. The entry content acquisition unit 111 associates the acquired group of characters with the answer column 313 of the acquisition source.

[Explanation of processing to narrow down candidate words]
Among the medical care support methods executed by the medical care support system S, the process of narrowing down the candidate words will be described below. The candidate word storage unit 122 stores in advance a plurality of candidate words that may be entered in the questionnaire 31. The candidate word storage unit 122 may memorize not only the candidate words that are expected to be described in the answer column 313 but also the candidate words acquired by learning the words that are described in the answer column 313 in the past. Good. Learning is performed not only on whether or not a candidate word appears, but also on the frequency of appearance of the candidate word and the like.

Further, the candidate word storage unit 122 has at least one of a state of one word itself, a state of dividing one word into a plurality of parts, and a state of combining a plurality of words for each of the plurality of candidate words. It may be memorized in the state. Specifically, the candidate word storage unit 122 divides and stores the candidate word into one or more consecutive character-by-character parts such as monogram (1 character), bigram (2 characters), and trigram (3 characters). You may. In this case, the candidate word storage unit 122 stores a plurality of divided parts of the candidate word by combining them with a blank character or a comma (a reading point). Further, the candidate word storage unit 122 may combine and memorize a plurality of different words as one word. In this case, the candidate word storage unit 122 stores a plurality of words by combining them with a blank character or a comma (a reading point).

The narrowing-down unit 112 performs a narrowing-down process on a plurality of candidate words stored in the candidate word storage unit 122. FIG. 7 is a schematic view of the narrowing process by the narrowing section 112. First, the narrowing unit 112 acquires a candidate word from the candidate word storage unit 122. Further, in the narrowing-down unit 112, for each of the group of characters acquired by the entry content acquisition unit 111 by the character recognition process, the position where the group of characters is acquired (that is, the group of characters is entered in any of the answer columns 313). Was it) to get.

As a first method, the narrowing section 112 is based on the medical term heading 314 (ie, the medical term described in or near the answer column 313) associated with the answer column 313 in which a group of letters is entered. To narrow down the candidate words. In this case, the plurality of candidate words stored in the candidate word storage unit 122 are previously associated with one or more medical terms that can be used as headings 314, respectively.

Based on the image of the questionnaire 31, the entry content acquisition unit 111 acquires the content of the heading 314 associated with the answer column 313 in addition to the group of characters of the answer column 313. In the example of FIG. 7, the heading 314 of "allergy" is associated with the answer column 313 in which "loxoprofen" is written. Here, the answer column 313 and the heading 314 are arranged in one frame, so that they are related to each other. A plurality of answer columns 313 may be associated with one heading 314. The method is not limited to the method shown here, and the answer column 313 and the heading 314 may be associated with each other by other methods such as association by a database.

The drug name may be entered in the answer column 313, which is a free entry column associated with the heading 314 of "allergy". Therefore, the narrowing-down unit 112 narrows down the candidate words stored in the candidate word storage unit 122 to the candidate words associated with the heading 314 of "allergy" (specifically, the candidate words related to the drug name).

Further, in the example of FIG. 5, the time may be entered in the answer column 313, which is a free entry column associated with the heading 314 of "fever". Therefore, the narrowing unit 112 narrows down the candidate words stored in the candidate word storage unit 122 to the candidate words associated with the heading 314 of "fever" (specifically, the candidate words related to time).

The medical care support device 1 stores medical terms and candidate words in association with each other in the candidate word storage unit 122 by performing machine learning in advance. Specifically, the medical care support device 1 is set for each group of characters (that is, learning data) read from the answer column 313 of the exemplary questionnaire 31 and the medical term of the heading 314 associated with the answer column 313. Machine learning is performed in advance using the words (that is, teacher data) indicated by the group of letters, and a model showing the relationship between the group of letters and words is generated for each medical term of the heading 314 associated with the answer column 313. To do. Then, the medical care support device 1 stores the generated model in the candidate word storage unit 122. For example, machine learning is performed by continuously combining Convolutional Neural Network, Bidirectional Long Short-term Neural Network, and Connectionist Temporal Classification. Convolutional Neural Network learns each character character by character, Bidirectional Long Short-term Neural Network converts it into a string of probability pairs of character candidates and each character, and Connectionist Temporal Classification converts it into a character string. The search algorithm of Connectionist Temporal Classification is restricted so that the output that comes out is narrowed down to the candidate words. The medical care support device 1 may use other known machine learning methods. For example, the medical care support device 1 divides the entered character string into one or more characters by using a classifier learned by machine learning, and then divides one or more characters into excessively divided characters. Combine characters. Next, the medical care support device 1 acquires a character candidate and its probability as a result of recognizing each of one or a plurality of divided or combined characters by using a classifier learned as a result of machine learning. By combining them, it becomes a candidate group of medical words. Then, the medical care support device 1 may acquire candidate words by converting the character strings of those candidate groups by an algorithm such as spell check of medical terms.

The narrowing-down unit 112 acquires a model as a learning result from the candidate word storage unit 122. Next, the narrowing-down unit 112 uses the group of characters acquired by the entry content acquisition unit 111 and the medical term of the heading 314 associated with the answer column 313 as a learning result acquired from the candidate word storage unit 122. input. Then, the narrowing unit 112 narrows down the output from the model as a candidate word.

When the candidate word storage unit 122 narrows down the candidate words based on the medical term heading 314 associated with the answer column 313 in this way, each candidate word does not have to be associated with the answer column 313 one by one, and the heading It may be associated with a limited number of medical terms used in 314. Therefore, for example, even if the headings of medical terms are the same and the question is changed, the medical staff can divert the data for narrowing down without any effort.

As a second method, the narrowing unit 112 may narrow down the candidate words based on the answer column 313 itself in which a group of characters is entered without using the heading 314. In this case, each of the plurality of candidate words stored in the candidate word storage unit 122 is associated with each of the plurality of answer columns 313 of the plurality of questionnaires 31 in advance. The narrowing-down unit 112 identifies the answer column 313 based on the identification information of the questionnaire 31 indicated by the code 316 and the position of the answer column 313, and identifies the candidate word stored in the candidate word storage unit 122 in the specified answer column 313. Narrow down to candidate words associated with. In this case, the candidate word storage unit 122 stores a model showing the relationship between a group of characters and words for each answer column 313 (position of the answer column 313) instead of the medical term of the heading 314.

When the candidate word storage unit 122 narrows down the candidate words based on the answer column 313 itself in this way, each candidate word needs to be associated with the answer column 313, but the heading 314 can be omitted from the questionnaire 31. There is an advantage that the conventional questionnaire 31 can be used as it is.

As described above, the narrowing unit 112 takes to identify the answer in order to narrow down the candidate words that the group of characters may indicate by using the position where the group of characters was acquired and the learning result of machine learning. The processing time can be shortened and the accuracy of identifying the answer can be improved.

[Explanation of the process to identify the answer]
Among the medical care support methods executed by the medical care support system S, the process of specifying the answer will be described below. The similar information storage unit 123 stores in advance a matrix (matrix) indicating the inter-character distance of each combination of characters.

FIG. 8 is a schematic diagram of a matrix M showing the distance between characters. The inter-character distance is a value indicating the appearance similarity between two characters, and is defined so that the value becomes smaller as the two characters are similar.

In the example of FIG. 8, for example, "so" and "n" are similar in appearance, so the distance between letters is small, whereas "a" and "i" are not similar in appearance. Therefore, the distance between characters is large. The distance between characters of the same character is 0.

The inter-character distance is not limited to the definition shown here, and may be defined so that the value becomes larger as the two characters are similar, or the binary value of whether or not the two characters are similar is used. It may be defined using.

In FIG. 8, the matrix M is represented by a table of characters for visibility, but the matrix M may be in other formats such as character string data, numerical data, and binary data. The matrix M may be recorded as a database or as a list listing data.

The answer specifying unit 113 acquires the matrix M of the inter-character distance from the similar information storage unit 123. The answer specifying unit 113 uses the acquired matrix M to be between each character included in the group of characters acquired by the entry content acquisition unit 111 and each character included in each of the candidate words narrowed down by the narrowing unit 112. Calculate the similarity by summing the distances between characters. In the example of FIG. 8, the similarity becomes smaller as the group of characters and the candidate word are similar, and becomes larger as the group of characters and the candidate word are similar.

Specifically, the answer specifying unit 113 acquires the inter-character distance between the Xth character of the group of characters to be compared (X: the number of characters of the 1st group of characters) and the Xth character of the candidate word from the matrix M. Then, the acquired distances between characters are summed up in order. When the number of characters of the group of characters to be compared and the number of characters of the candidate word are different, the answer identification unit 113 shifts the characters and the distance between the characters (for example, the Xth character of the group of characters and the X-1th character or the X + 1 character of the candidate word) The distance between letters between the eyes) may be totaled.

For example, when the group of characters acquired by the entry content acquisition unit 111 is "loxoprofen" and the candidate word is "loxoprofen", there is only one difference between them (and "so" with a small inter-character distance). Since it is "n"), the similarity is relatively small. When the group of characters acquired by the entry content acquisition unit 111 is "Loxoprofen" and the candidate word is "pranoprofen", there are four differences between them, so the similarity is relatively large. It becomes.

Then, the answer specifying unit 113 specifies, among the candidate words narrowed down by the narrowing unit 112, the candidate words whose calculated similarity satisfies a predetermined condition as an answer. In the example of FIG. 8, the similarity becomes smaller as the characters in the group and the candidate word are similar to each other. Therefore, the answer specifying unit 113 specifies the candidate word having the lowest similarity as the answer. When the similarity becomes so large that the characters in the group and the candidate word are similar to each other, the answer specifying unit 113 may specify the candidate word having the highest similarity as the answer.

In this way, the answer specifying unit 113 identifies the answer corresponding to the group of characters acquired from the image from the candidate words based on the value of the inter-character distance indicating the similarity between the two characters. Therefore, the answer identification unit 113 can identify the answer entered in the questionnaire 31 with high accuracy even if the group of characters acquired from the image contains an error in the character recognition process or an error by the patient. Further, when there are a large number of candidate words, it takes a lot of time to calculate the similarity between a group of characters and all the candidate words, but in the present embodiment, the narrowing unit 112 narrows down the candidate words. Therefore, the answer identification unit 113 can specify the answer in a realistic processing time.

When the candidate word storage unit 122 divides the candidate word into parts for each of two or more characters and stores the candidate word, the answer identification unit 113 includes one character or a plurality of characters included in each of the plurality of candidate words. The value of the inter-character distance between the combination of the above and the combination of the plurality of characters in the relevant part and the combination of the plurality of characters included in the group of characters acquired by the entry content acquisition unit 111 is acquired, and the similarity is calculated. To do.

The answer specifying unit 113 performs a beam search to search for a group of characters acquired by the entry content acquisition unit 111, searches for the distance between characters with the candidate word in the beam search results, and selects from the shortest one. You may. As for the distance between characters, the result of machine learning based on the correct answer data may be used, or the parameter obtained by learning autoencoding may be used for the font list information.

Further, even if the narrowing section 112 narrows down the candidate words for another answer column 313 (second answer column) based on the answer specified for the answer column 313 (first answer column) in which the answer specifying section 113 is located. Good. In this case, the candidate word storage unit 122 stores the answer in the first answer column 313 in advance in association with the candidate word in the second answer column. Then, the narrowing unit 112 limits the second answer column 313 to the candidate words associated with the answer of the first answer column 313, or limits the candidate words associated with the answer of the first answer column 313. Weighting is performed to perform the above-mentioned narrowing down.

For example, when the answer identification unit 113 identifies the answer of "muscle tone headache" in the answer column 313 of the medical history, the answer column 313 of the drug history of the same questionnaire 31 is related to the muscle tone headache such as "loxoprofen". There is a high possibility that the name of the drug to be used is listed. Therefore, the narrowing-down unit 112 narrows down the answer column 313 of the drug history by limiting or weighting the drug names related to muscle tone headache among the candidate words.

The output unit 114 stores the answer specified by the answer identification unit 113 in the answer storage unit 124 in association with the patient (patient identification information), the medical term associated with the answer column 313 and the answer column 313 (heading 314). Output by letting. Further, the output unit 114 outputs the information for displaying the answer specified by the answer specifying unit 113 by transmitting the information to the medical care terminal 2. The output unit 114 may output information by other methods. For example, the output unit 114 may print the information using a printer, or may transmit the information to the outside by using a communication means.

FIG. 9 is a front view of the display unit 21 of the medical terminal 2. The medical care terminal 2 displays the answer specified by the answer specifying unit 113 on the display unit 21 according to the information received from the medical care support device 1. The display unit 21 displays the first button 211 and the second button 212. The first button 211 and the second button 212 are virtual buttons that can be pressed by a medical professional. When the first button 211 is pressed, the screen shown in the upper part of FIG. 9 is displayed, and when the second button 212 is pressed, the screen shown in the lower part of FIG. 9 is displayed.

The display unit 21 displays the medical term 213 corresponding to the question 312 of the questionnaire 31 and the answer 214 specified by the answer specifying unit 113 in association with each other (that is, in the vicinity of each other) on the screen shown in the upper part of FIG. .. In this case, the output unit 114 transmits a template showing the medical term 213 corresponding to the question 312 of the questionnaire 31 to the medical terminal 2. This template is associated with the questionnaire 31 (identification information of the questionnaire 31) and is stored in the storage unit 12 in advance. The display unit 21 displays the medical term 213 and the answer 214 according to the received template. The display unit 21 displays the answer specified by the answer specifying unit 113 as the answer 214 for the answer column 313 which is a free entry field, and displays the answer column 313 which is a check box according to whether the check box is on or off. Or "No" is displayed as answer 214.

On the screen shown in the lower part of FIG. 9, the display unit 21 associates the original question 215 represented by the questionnaire 31 with the entry content 216 of the questionnaire 31 acquired by the entry content acquisition unit 111 (that is, in the vicinity of each other). indicate. In this case, the layout storage unit 121 stores the medical term as the heading 314 of the questionnaire 31 and the question 312 of the questionnaire 31 in association with each other so that the question of the plain expression corresponding to the medical term can be referred to. ing. The output unit 114 transmits a template showing the question 312 of the questionnaire 31 to the medical care terminal 2. This template is associated with the questionnaire 31 (identification information of the questionnaire 31) and is stored in the storage unit 12 in advance. The display unit 21 displays the original question 215 and the entry contents 216 according to the received template. The display unit 21 displays a group of characters acquired by the entry content acquisition unit 111 as the entry content 216 for the answer column 313, which is a free entry field, and displays an on or off check box for the answer column 313, which is a check box. It is displayed as the entry content 216.

In this way, the output unit 114 can switch between the answer 214 specified by the answer specifying unit 113 and the original entry content 216 of the questionnaire 31 and display it on the display unit 21, so that the medical staff makes an error. You can check not only the answer after the correction is made, but also the contents before the error is corrected. Further, since the output unit 114 can display the medical terms corresponding to the question 312 of the questionnaire 31 on the display unit 21, the medical staff can medically understand the contents of the question 312 of the questionnaire 31. You can save time.

Further, in the output unit 114, when the response identification unit 113 specifies a predetermined drug name (rare drug, antidiabetic drug, etc.) in the response column 313 (first answer column) of the drug history, the response identification unit 113 It is determined whether or not the disease name (disease name) corresponding to the drug name is specified in the answer column 313 (second answer column) of the medical history. When the disease name corresponding to the drug name is not specified in the answer column 313 of the medical history, the output unit 114 outputs information indicating that fact by transmitting it to the medical care terminal 2. As a result, the medical care support system S can notify the medical staff as an alert that the plurality of answers on the questionnaire 31 are inconsistent.

[Flowchart of medical care support method]
FIG. 10 is a diagram showing a flowchart of a medical care support method (information processing method) according to the present embodiment. First, the patient fills in the answer to the question on the questionnaire 31. The medical staff or the patient cuts a part of the questionnaire 31 including the fixture F, and divides the questionnaire 31 into sheets of paper. Then, the medical staff or the patient has the reading device 3 read the questionnaire 31. In the medical care support device 1, the entry content acquisition unit 111 receives the image of the questionnaire 31 from the reading device 3 that has read the questionnaire 31.

The entry content acquisition unit 111 detects the symbol 315 in the image of the received questionnaire 31, and identifies the position of the answer area 311 based on the detected symbol 315 (S11). The entry content acquisition unit 111 corrects the received image of the questionnaire 31 based on the position of the specified answer area 311.

The entry content acquisition unit 111 detects the code 316 in the received image of the questionnaire 31, and acquires the layout information associated with the identification information of the questionnaire 31 indicated by the detected code 316 from the layout storage unit 121. .. The entry content acquisition unit 111 takes out an image including the position of each answer column 313 indicated by the layout information in the image of the questionnaire 31. The entry content acquisition unit 111 acquires a group of characters entered in the answer column 313 by applying the character recognition process to the images taken out from each answer column 313 (S12).

The narrowing-down unit 112 acquires a candidate word from the candidate word storage unit 122 (S13). The narrowing-down unit 112 includes a position where a group of characters acquired by the entry content acquisition unit 111 by character recognition processing is acquired (that is, which answer column 313 the group of characters was entered in) and a candidate word storage unit 122. The candidate words are narrowed down based on the learning result stored in advance in (S14).

The answer specifying unit 113 acquires a matrix of inter-character distances from the similar information storage unit 123. Using the acquired matrix, the answer specifying unit 113 between each character included in the group of characters acquired by the entry content acquisition unit 111 and each character included in each of the candidate words narrowed down by the narrowing unit 112. The distance between characters is totaled to calculate the similarity (S15).

The answer specifying unit 113 specifies as an answer a candidate word whose calculated similarity satisfies a predetermined condition among the candidate words narrowed down by the narrowing unit 112 (S16). The output unit 114 outputs the answer specified by the answer specifying unit 113 by storing it in the answer storage unit 124 and displaying it on the display unit 21 of the medical care terminal 2 (S17).

[Effect of this embodiment]
The medical care support system S according to the present embodiment narrows down the candidate words that may be entered in the questionnaire based on the position where the characters are entered and the learning result, and then identifies the answer from the candidate words. Therefore, the medical care support system S can shorten the processing time required to identify the answer and improve the accuracy of identifying the answer. Further, the medical care support system S identifies an answer corresponding to a group of characters acquired from the image from the candidate words based on the value of the inter-character distance indicating the similarity between the two characters. Therefore, the medical care support system S can identify the answer entered in the questionnaire 31 with high accuracy even if the group of characters acquired from the image contains an error in the character recognition process or an error by the patient.

Further, the medical care support system S identifies the position of the answer area in the questionnaire by multiplying the line segment between the two symbols by a predetermined coefficient, and corrects the image. Therefore, the medical care support system S reads even if a part of the questionnaire is missing due to cutting a fixture such as a stapler or the symbol is not placed at a position where the questionnaire may be cut. It is possible to correct the distortion and inclination of the image generated when the device is made to read the questionnaire, and it is possible to acquire the characters written on the questionnaire with high accuracy.

Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes can be made within the scope of the gist. is there. For example, the specific embodiment of the distribution / integration of the device is not limited to the above embodiment, and all or a part thereof may be functionally or physically distributed / integrated in an arbitrary unit. Can be done. Also included in the embodiments of the present invention are new embodiments resulting from any combination of the plurality of embodiments. The effect of the new embodiment produced by the combination has the effect of the original embodiment together.

The processor of the medical care support device 1 is the main body of each step included in the method shown in FIG. That is, the processor of the medical care support device 1 reads a program for executing the method shown in FIG. 10 from the storage unit, executes the program, and controls each part of the medical care support system S, whereby the method shown in FIG. 10 To execute. Some of the steps included in the method shown in FIG. 10 may be omitted, the order between the steps may be changed, and a plurality of steps may be performed in parallel.

S Medical treatment support system 1 Medical treatment support device 11 Control unit 111 Entry content acquisition unit 112 Narrowing unit 113 Answer specific unit 114 Output unit 12 Storage unit 121 Layout storage unit 122 Candidate word storage unit 123 Similar information storage unit 31 Questionnaire 311 Answer area 313 Answer column 315 Symbol

Claims (12)

A layout storage unit that stores the layout of the questionnaire in association with the identification information that identifies the questionnaire, and
Candidate word storage unit that stores multiple candidate words,
An acquisition unit that acquires the identification information from the image of the questionnaire and acquires one or more characters included in the image based on the layout associated with the acquired identification information.
A narrowing section that narrows down the plurality of candidate words based on the position where one or more characters are acquired, and
A specific unit that identifies the answer entered in the questionnaire by comparing the one or more characters acquired by the acquisition unit with the plurality of candidate words narrowed down by the narrowing unit.
An output unit that outputs the answer specified by the specific unit, and an output unit that outputs the answer.
Information processing device with. The specific unit calculates the similarity between the one or more characters acquired by the acquisition unit and each of the plurality of candidate words narrowed down by the narrowing unit, and selects the specific unit based on the calculated similarity. The information processing apparatus according to claim 1, wherein one of the plurality of candidate words is specified as the answer. It further has a similar information storage unit that stores a value indicating similarity between two characters.
The specific unit is a combination of one character or a plurality of characters included in the one or more characters acquired by the acquisition unit and a combination of one character or a plurality of characters included in each of the plurality of candidate words. The information processing apparatus according to claim 2, wherein the value between the two is acquired, and the similarity is calculated using the acquired value. The layout storage unit stores the layout indicating the position of the answer column included in the questionnaire with reference to the predetermined position.
The information processing device according to any one of claims 1 to 3, wherein the acquisition unit acquires the one or more characters at the position of the answer column indicated by the layout in the image. The candidate word storage unit stores the plurality of candidate words in association with the answer column, and stores the plurality of candidate words.
The narrowing unit narrows down the plurality of candidate words by acquiring the plurality of candidate words associated with the answer column in which one or more characters are acquired from the candidate word storage unit. The information processing apparatus according to any one of 4 to 4. The candidate word storage unit stores the plurality of candidate words in association with medical terms, and stores the plurality of candidate words.
In addition to the one or more characters, the acquisition unit acquires the medical term associated with the answer column in which the one or more characters are acquired in the image.
The narrowing unit narrows down the plurality of candidate words by acquiring the plurality of candidate words associated with the medical term acquired by the acquisition unit from the candidate word storage unit.
The information processing device according to any one of claims 1 to 4, wherein the output unit outputs the answer in association with the medical term. The layout storage unit stores the questions described in the questionnaire in association with the medical terms.
The information processing device according to claim 6, wherein the output unit outputs the question in addition to the answer and the medical terminology. The narrowing-down unit is based on the answer specified by the specific unit for the first answer column at the position where one or more characters are acquired, and the second answer column is different from the first answer column. , The information processing apparatus according to any one of claims 1 to 7, which narrows down the plurality of candidate words. The candidate word storage unit is in at least one state of each of the plurality of candidate words being one word, one word divided into a plurality of parts, and a plurality of words combined. The information processing apparatus according to any one of claims 1 to 8, which is stored. The acquisition unit identifies a line segment connecting two predetermined symbols included in the image, and the answer is entered in the questionnaire by multiplying the length of the line segment by a predetermined coefficient. The information processing apparatus according to any one of claims 1 to 9, which calculates the length of a side that divides a possible region and corrects the image based on the length. A computer having a layout storage unit that stores the layout of the questionnaire in association with the identification information that identifies the questionnaire, and a candidate word storage unit that stores a plurality of candidate words.
A step of acquiring the identification information from the image of the questionnaire and acquiring one or more characters included in the image based on the layout associated with the acquired identification information.
A step of narrowing down the plurality of candidate words based on the position where one or more characters are acquired, and
A step of identifying the answer entered in the questionnaire by comparing the one or more characters acquired by the acquisition step with the plurality of candidate words narrowed down by the narrowing step.
The step of outputting the answer specified by the specific step and the step of outputting the answer
Information processing method having. A computer having a layout storage unit that stores the layout of the questionnaire in association with the identification information that identifies the questionnaire, and a candidate word storage unit that stores a plurality of candidate words.
A step of acquiring the identification information from the image of the questionnaire and acquiring one or more characters included in the image based on the layout associated with the acquired identification information.
A step of narrowing down the plurality of candidate words based on the position where one or more characters are acquired, and
A step of identifying the answer entered in the questionnaire by comparing the one or more characters acquired by the acquisition step with the plurality of candidate words narrowed down by the narrowing step.
The step of outputting the answer specified by the specific step and the step of outputting the answer
An information processing program that executes.

Images