Disease-specific medical examination day number estimation device, method, and program

Abstract

To provide a disease-specific medical examination day number estimation device, method, and program capable of performing three-element decomposition of medical expenses by disease using medical fee information such as receipts.SOLUTION: A disease-specific medical examination day number estimation device includes: a first balancing score calculation unit 12 that receives a plurality of pieces of analysis information including at least one piece of disease information of a patient, at least one piece of medical action information administered to the patient, and the number of medical examination days associated with the medical action information, and calculates a balancing score for each analysis information on the basis of the received pieces of analysis information; and a disease-specific medical examination day number estimation unit 13 that estimates the number of medical examination days for each disease information on the basis of the balancing score for each analysis information.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an apparatus, a method, and a program for estimating the number of days of medical treatment for each type of injury or illness based on a receipt or the like and estimating the number of days of medical treatment for each type of injury or illness.

Background Art

[0002] In recent years, medical expenses using the medical insurance system have been increasing year by year, and the burdens on the country, insurers, and patients have been increasing. In the future, in order to maintain the medical insurance system in a sound state, it is necessary to prevent injuries and illnesses, detect them at an early stage and provide early treatment, and suppress medical expenses.

[0003] When considering comparisons at a certain group or different points in time, "medical expenses per person" is one of the typical indicators. Medical expenses per person are obtained by the following formula. Medical expenses per person = (total medical expenses) / (number of subscribers)

[0004] Furthermore, medical expenses per person are decomposed into the "medical treatment rate (number of cases per person)", the "period" of medical treatment for diseases such as the number of days of outpatient visits, and the "unit price" of medical treatment for diseases such as medical expenses per day. Medical expenses per person can be decomposed into the product of the "medical treatment rate (number of cases per person)", the "number of days per case", and the "medical expenses per day" as shown in the following formula. Medical expenses per person = (total medical expenses) / (number of subscribers) = {(number of patients) / (number of subscribers)} × {(number of days of medical treatment) / (number of patients)} × {(total medical expenses) / (number of days of medical treatment)} = (medical treatment rate (number of cases per person)) × (number of days per case) × (medical expenses per day)

[0005] The "medical treatment rate (number of cases per person)", the "number of days per case", and the "medical expenses per day" are called the three elements of medical expenses and can be used as basic indicators for analyzing medical expenses.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

[0007] Here, if it is possible to further decompose the three elements of medical expenses by type of injury or illness, further analysis of medical expenses can be performed, and thereby, it may be possible to efficiently suppress medical expenses.

[0008] However, in order to decompose the three elements of medical expenses by type of injury or illness, the number of days of medical treatment by type of injury or illness is required. Although the actual number of days of medical treatment is shown in the medical fee information such as receipts, since it is not related to a specific injury or illness, for example, when a person with multiple injuries or illnesses receives medical treatment, it is unknown which injury or illness was treated for how many days, and the number of days of medical treatment by type of injury or illness cannot be obtained.

[0009] In Patent Document 1 and Patent Document 2, it is proposed to calculate the medical expenses by type of injury or illness using a receipt including information on the patient's injury or illness, information on medical acts performed on the patient, and information on medical expenses. However, no method for obtaining the number of days of medical treatment by type of injury or illness has been proposed.

[0010] In view of the above problems, an object of the present invention is to provide an apparatus, a method, and a program for estimating the number of days of medical treatment by type of injury or illness that can perform three-element decomposition of medical expenses by type of injury or illness using medical fee information such as receipts. [Means for Solving the Problems]

[0011] The injury / disease-specific medical visit days estimation device of the present invention receives a plurality of analysis information including at least one piece of injury / disease information of a patient, at least one piece of medical treatment information applied to the patient, and the medical visit days related to the medical treatment information, and calculates a balancing score for each piece of analysis information based on the received plurality of analysis information, and includes a balancing score calculation unit, and a medical visit days estimation unit that estimates the medical visit days for each injury / disease based on the balancing score for each piece of analysis information.

Advantages of the Invention

[0012] According to the injury / disease-specific medical visit days estimation device of the present invention, a plurality of analysis information including at least one piece of injury / disease information of a patient, at least one piece of medical treatment information applied to the patient, and the medical visit days related to the medical treatment information are received, a balancing score for each piece of analysis information is calculated based on the received plurality of analysis information, and the medical visit days for each injury / disease are estimated based on the balancing score for each piece of analysis information. Therefore, using the estimated medical visit days for each injury / disease, three-factor decomposition of medical expenses by injury / disease can be performed.

Brief Description of the Drawings

[0013]

Figure 1

Figure 2

Figure 3

Figure 4

Embodiments for Carrying Out the Invention

[0014] Hereinafter, an embodiment of an injury / disease-specific medical expense analysis system using an injury / disease-specific visit days estimation apparatus, method, and program of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of an injury / disease-specific medical expense analysis system 1 according to the present embodiment.

[0015] The injury / disease-specific medical expense analysis system 1 according to the present embodiment is a system that can estimate the number of visit days for each injury / disease based on analysis information such as receipt information, and perform three-factor decomposition of the medical expenses for each injury / disease using the estimated number of visit days for each injury / disease.

[0016] Specifically, the injury / disease-specific medical expense analysis system 1 according to the present embodiment estimates the number of visit days for each injury / disease and the medical expenses for each injury / disease using the balancing score method based on analysis information such as the patient's receipt information and electronic medical record information, and performs three-factor decomposition of the medical expenses for each injury / disease based on these.

[0017] As shown in FIG. 1, the injury / disease-specific medical expense analysis system 1 according to the present embodiment includes an injury / disease-specific medical expense analysis apparatus 10 (corresponding to the arithmetic apparatus of the present invention), an input apparatus 20, and a display apparatus 30.

[0018] The injury / disease-specific medical expense analysis apparatus 10 includes an analysis information acquisition unit 11, a first balancing score calculation unit 12, an injury / disease-specific visit days estimation unit 13, a second balancing score calculation unit 14, an injury / disease-specific medical expense estimation unit 15, an injury / disease-specific medical expense analysis unit 16, and a display control unit 17.

[0019] The analysis information acquisition unit 11 acquires analysis information used for estimating the number of visit days for each injury / disease and the medical expenses for each injury / disease, which will be described later.

[0020] As analysis information, for example, there is receipt information, electronic medical record information, medical treatment record information, etc. as shown in FIG. 2. Receipt information refers to information in a medical fee statement (receipt) that a medical institution bills to an insurer (such as a national health insurance association or a health insurance association) for the insured medical treatment received by a patient. A receipt is also called a medical fee statement in the case of medicine and dentistry, a dispensing fee statement in the case of dispensing at a pharmacy, and a home visit nursing care cost statement in the case of home visit nursing care. Note that the receipt information shown in FIG. 2 shows an outline of the information in the medical fee statement, and the receipt information includes at least one injury condition information of the patient, at least one medical treatment information performed on the patient, medical fee information related to the medical treatment information, and the actual number of days of medical treatment.

[0021] Injury condition information refers to information indicating an injury or illness such as a disease or trauma that a patient is suffering from.

[0022] In addition, medical treatment information includes medical treatment information indicating the content of medical treatment such as treatments, procedures, examinations, and surgeries performed on a patient, information indicating drugs administered to the patient, and information indicating equipment used on the patient.

[0023] In addition, medical fee information refers to information indicating the medical fees required for the content of medical treatment for a patient, drugs administered to the patient, and equipment used on the patient. As the information on medical fees, it may be information indicating the so-called sales amount, information indicating the medical cost, or information indicating the points of insured medical treatment.

[0024] In addition, the actual number of days of medical treatment is the number of days on which the above-mentioned medical treatment information is received. In the case of hospitalization, it is the number of days of hospitalization. In the case of outpatient treatment, it is the number of days on which a doctor's medical treatment is received in an outpatient clinic, a home visit, etc.

[0025] Electronic medical record information is information created by a doctor when examining a patient by inputting information using a computer or the like. As shown in FIG. 2, the electronic medical record information also includes at least one injury condition information of the patient and at least one medical act information performed on the patient. The injury condition information and the medical act information are as described above. In addition, the electronic medical record information also includes information indicating the results of a medical interview, findings, imaging findings, and test values.

[0026] The content of the medical record information is basically the same as that of the electronic medical record information. However, while the electronic medical record information is created from a format defined by the electronic medical record system of a medical institution, the medical record information is not created from a fixed format like the electronic medical record information. For example, it may be created by inputting injury condition information and medical act information handwritten by a doctor or the like on a recording medium such as paper into a computer or the like, or it may be electronic data obtained by optically reading the above-described recording medium with a scanner or the like. Note that the medical record information can include the content defined by the Medical Practitioners Act.

[0027] In addition, as a method for acquiring the analysis information in the analysis information acquisition unit 11, for example, for receipt information and electronic medical record information, it may be acquired from a receipt server storing the receipt information and an electronic medical record server storing the electronic medical record information in an electronic medical record system via a communication line such as the Internet line or a LAN, or it may be set and input using the input device 20.

[0028] Regarding the medical record information as well, similar to the receipt information and the electronic medical record information, it may be read and acquired from a predetermined server storing the medical record information, or it may be set and input using the input device 20, or it may be acquired as electronic data read using a scanner or the like as described above.

[0029] In addition, the analysis information acquired by the analysis information acquisition unit 11 does not necessarily have to be stored within a single medical institution, and it may be collected and acquired from a plurality of different medical institutions for the analysis information of a single patient. Also, regarding the information indicating the administered drug, the information stored in a server such as a dispensing pharmacy may be acquired.

[0030] In addition, as the range of the analysis information acquired by the analysis information acquisition unit 11, for example, the analysis information may be acquired within the range of a patient unit, a medical facility unit, an insurer unit, a regional unit, or a unit of a predetermined attribute. When acquiring the analysis information in patient units, identification information set for each patient is respectively added to the analysis information such as receipt information, electronic medical record information, and medical record information, and the analysis information in patient units may be collected using the identification information. Examples of the identification information set for each patient include, for example, a medical insurance card number or an insured person number, or information obtained by combining these with the date of birth and / or gender, etc., but any information may be used as long as it can identify an individual. Note that when acquiring the analysis information in patient units, the analysis information is acquired in units of a group consisting of a certain number of people, rather than a single patient.

[0031] When acquiring the analysis information in medical facility units, identification information set for each medical facility is respectively added to the analysis information such as receipt information, electronic medical record information, and medical record information, and the analysis information in medical facility units may be collected using the identification information. Examples of the identification information set for each medical facility include, for example, a medical institution number, etc., but any information may be used as long as it can identify a medical facility.

[0032] In addition, a time range such as a monthly unit and an annual unit may be set to acquire the analysis information. In this case, information capable of specifying the time such as the year and month of creation may be added to the analysis information.

[0033] The first balancing score calculation unit 12 receives a plurality of pieces of analysis information and calculates a first balancing score based on the received analysis information. In the present embodiment, a first tendency score is calculated as the first balancing score. The first tendency score is calculated for each of the above-described pieces of analysis information.

[0034] As shown in FIG. 1, the first balancing score calculation unit 12 includes a first information extraction unit 12a. In the present embodiment, the first information extraction unit 12a corresponds to the injury condition information extraction unit, the medical treatment information extraction unit, and the number of days of medical visits extraction unit of the present invention. The first information extraction unit 12a extracts injury condition information, medical treatment information, and the number of days of medical visits from each of the plurality of pieces of analysis information acquired by the analysis information acquisition unit 11.

[0035] As a method for extracting injury condition information from receipt information, for example, codes for specifying injuries such as ICD10 (International Classification of Diseases, 10th Revision) or standard disease name numbers may be extracted from the receipt information, or the injury disease name itself may be extracted using character recognition processing or the like.

[0036] Similarly, for medical treatment information, code information preset for each medical treatment may be extracted, or the characters themselves indicating the medical treatment may be extracted.

[0037] Regarding the number of days of medical visits, the actual number of days of medical treatment indicated in the receipt information is extracted. Note that only this actual number of days of medical treatment does not clarify for which injury the number of days of medical visits is.

[0038] Also, as described above, the electronic medical record information and the medical treatment record information basically do not include the number of days of medical visits. Therefore, for the electronic medical record information and the medical treatment record information, for example, the electronic medical record information and the medical treatment record information can be linked and stored in advance through counting the number of days when they are recorded or updated and a preset number or the like, and together with the electronic medical record information and the medical treatment record information, the linked number of days can be read out and obtained from a server or the like.

[0039] Then, the first balancing score calculation unit 12 calculates a first tendency score for each analysis information using the injury condition information, the medical treatment information, and the number of days of medical visits extracted by the first information extraction unit 12a as described above. Hereinafter, as a specific example of the method for calculating the first tendency score, a method for calculating the first tendency score used for estimating the number of days of medical visits for diabetes will be described.

[0040] First, as a disease-specific variable indicating whether or not the diabetes, which is the disease of interest, is included in the analysis information, z = 0 (no diabetes) and z = 1 (diabetes) are set.

[0041] Next, among the medical treatment information that may be included in the analysis information, the medical treatment information that is related to both the disease of interest (here, diabetes) and the number of days of medical visits and corresponds one-to-one with the disease is excluded, and the medical treatment information that is considered to be appropriate as a confounding factor is selected as the covariate x=(x 1 ,x 2 ,x 3 ,···). The variable (covariate) indicating the medical treatment information corresponding to this confounding factor may be 0 or 1, or a continuous variable may be used. Usually, the variables indicating this medical treatment information are usually multiple, such as x 1 ,x 2 ,x 3 ,···.

[0042] And, taking the number of days of medical visits included in the analysis information as yd, the number of days of medical visits in the case of no diabetes is set as yd 0 , and the number of days of medical visits yd 1 in the case of having diabetes is set.

[0043] Then, based on each piece of information included in the actually acquired analysis information under the above-mentioned preparations, a first tendency score is calculated.

[0044] FIG. 3 is a diagram for explaining the values of x, yd, and z when the analysis information is receipt information 1. In the case of receipt information 1 shown in FIG. 3, when diabetes is included in "injury / disease 1" and "injury / disease 2" which are injury / disease information, z = 1, and when diabetes is not included, z = 0. Also, depending on whether medical practice information corresponding to confounding factors is included in the medical practice information of "medical practice A", "medical practice B", "medical practice C", and "drug D" (administered drug), the values of the variables of the covariate x=(x 1 , x 2 , x 3 , ···) are set. Also, Dy of "actual medical treatment days Dy days" which is the number of days of medical treatment is set as the value of yd.

[0045] In the same manner as the receipt information 1 shown in FIG. 3, the values of x, yd, and z are also set for other analysis information such as receipt information, electronic medical record information, and medical treatment record information.

[0046] Then, using the plurality of pieces of analysis information acquired by the analysis information acquisition unit 11, logistic regression analysis is performed with x as the explanatory variable and z as the explained variable. Specifically, each coefficient β is estimated by numerical calculation such as the Newton-Raphson method. Then, from the following formula, the tendency score e i for each piece of analysis information i is calculated.

Equation

[0047] Returning to FIG. 1, the number of outpatient days estimated by injury / sickness unit 13 estimates the number of outpatient days for each injury / sickness information based on the first propensity score for each analysis information calculated by the first balancing score calculation unit 12. The number of outpatient days estimated by injury / sickness unit 13 in the present embodiment estimates the number of outpatient days for each injury / sickness using the matching method.

[0048] Specifically, first, for the injury / sickness of interest (diabetes here), it is divided into a group with z = 1 and a group with z = 0, and subjects with the same first propensity score are selected from each group to form pairs. Since it is difficult for the first propensity scores to be exactly the same value, it may be set in advance as to how close the values of the first propensity scores should be to be regarded as pairs with the same value. Also, what to do when the pairs with the same first propensity score are not one-to-one but many-to-many may be determined in advance. Specifically, nearest neighbor matching that pairs the ones with the shortest distance between two points or caliper matching that does not pair if they are separated by a preset distance or more may be used.

[0049] Also, if there are no subjects to be paired for a given first propensity score, the calculation of the difference in the number of outpatient days described later is not performed, and the data without pairs itself is discarded.

[0050] Here, assuming that a total of N pairs of pairs j are formed, for the pairs j with the same first propensity score, the difference in the number of outpatient days (yd 1j - yd 0j ) is calculated as follows.

Equation

[0051] And then, as shown in the following equation, the average value E(yd j - yd 1 - yd 0 ) of the difference W is calculated, and this is used as the estimated value of the number of outpatient days for the injury / sickness of interest (diabetes here).

Equation

[0052] In the above example, the number of days of medical visits for diabetes was estimated. For other injuries and illnesses, by setting z = 0 (no target injury or illness) and z = 1 (target injury or illness) as injury-specific variables indicating whether the injury or illness is included in the analysis information, the number of days of medical visits for each injury or illness can be estimated.

[0053] The second balancing score calculation unit 14 receives the above-described analysis information and calculates a second balancing score based on the received plurality of analysis information. In the present embodiment, a second propensity score is calculated as the second balancing score. Similar to the first propensity score, the second propensity score is calculated for each of the above-described analysis information such as receipt information, electronic medical record information, and medical record information.

[0054] As shown in FIG. 1, the second balancing score calculation unit 14 includes a second information extraction unit 14a. The second information extraction unit 14a extracts injury information, medical treatment information, and medical expense information from each of the plurality of analysis information acquired by the analysis information acquisition unit 11. The method of extracting these information is the same as that of the first information extraction unit 12a.

[0055] Then, the second balancing score calculation unit 14 calculates a second propensity score for each analysis information using the injury information, medical treatment information, and medical expense information extracted by the second information extraction unit 14a. Hereinafter, as a specific example of the method for calculating the second propensity score, a method for calculating the second propensity score used for estimating the medical expenses for diabetes will be described. Note that the basic method is the same as the method for calculating the first propensity score described above.

[0056] First, z = 0 (no diabetes) and z = 1 (diabetes) are set as injury-specific variables indicating whether diabetes, which is the target injury or illness, is included in the analysis information.

[0057] Next, among the medical act information that may be included in the analysis information, the medical act information related to both the target injury or illness (here, diabetes) and medical expenses, and those that correspond one-to-one with the injury or illness are excluded, and the medical act information that is considered appropriate as a confounding factor is selected as the covariate x = (x 1 , x 2 , x 3 , ···). The variable (covariate) indicating the medical act information corresponding to this confounding factor may be 0 or 1, or a continuous variable may be used. Usually, the variables indicating this medical act information are multiple, such as x 1 , x 2 , x 3 , ···.

[0058] Then, the medical expense information included in the analysis information is set as y, the medical expense in the case of no diabetes is set as y 0 , and the medical expense y 1 in the case of having diabetes.

[0059] Then, based on each piece of information included in the actually obtained analysis information under the above-mentioned preliminary preparations, the second propensity score is calculated.

[0060] When diabetes is included in the injury or illness information, z = 1, and when diabetes is not included, z = 0. Also, depending on whether the medical act information corresponding to the confounding factor is included in the medical act information, the values of the variables of the covariate x = (x 1 , x 2 , x 3 , ···) are set. Also, the medical expense information is set as the value of y.

[0061] Then, the second balancing score calculation unit 14 performs logistic regression analysis with x as the explanatory variable and z as the explained variable, and specifically estimates each coefficient β by numerical calculation such as the Newton-Raphson method. Then, from the following formula, the propensity score e i for each analysis information i is calculated.

Equation

[0062] The medical expense estimation unit 15 for each type of injury / sickness estimates medical expense information for each type of injury / sickness based on the second propensity score for each piece of analysis information calculated by the second balancing score calculation unit 14. The medical expense estimation unit 15 for each type of injury / sickness in the present embodiment estimates medical expense information for each type of injury / sickness using the matching method.

[0063] Specifically, first, for the injury / sickness of interest (diabetes here), it is divided into a group with z = 1 and a group with z = 0, and objects with the same second propensity score are selected from each group to form pairs. Since it is difficult for the second propensity scores to be exactly the same value, it may be set in advance regarding how close the propensity score values should be to be regarded as pairs with the same value. Also, what to do when the pairs with the same second propensity score are not one-to-one but many-to-many may be determined in advance. Specifically, it may be possible to use nearest neighbor matching that pairs the ones with the shortest distance between two points or caliper matching that does not pair them when they are separated by a preset distance or more.

[0064] Also, if there are no paired objects for a predetermined second propensity score, the calculation of the difference in medical expenses described later is not performed, and the data without pairs itself is discarded.

[0065] Here, assuming that a total of N pairs of pairs j are formed, for the pairs j with the same propensity score, the difference in medical expenses (y 1j -y 0j ) is calculated as follows.

Equation

[0066] Then, as follows, the average value E(y j of the difference W 1 -y 0 ) is calculated, and this is used as the estimated value of the medical expense information for the injury / sickness of interest (diabetes here).

Equation

[0067] In the above example, the medical expense information for diabetes was estimated. For other injuries and illnesses, by setting z = 0 (no targeted injury or illness) and z = 1 (targeted injury or illness) as variables for each injury or illness indicating whether the injury or illness is included in the analysis information, the medical expense information for each injury or illness can be estimated.

[0068] Next, the medical expense analysis unit 16 by injury or illness performs a three-factor decomposition of the medical expenses by injury or illness based on the number of days of medical treatment by injury or illness estimated by the number of days of medical treatment by injury or illness estimation unit 13 and the medical expenses by injury or illness estimated by the medical expense estimation unit 15 by injury or illness.

[0069] The medical expense analysis unit 16 by injury or illness calculates the number of cases per person by injury or illness, the number of days per case by injury or illness, and the medical expenses per day by injury or illness within a predetermined fixed period.

[0070] Regarding the number of cases per person by injury or illness, it is obtained by calculating (the number of cases by injury or illness) / (the number of people). Regarding the number of cases by injury or illness, it is obtained by counting the number of receipt information including a specific injury or illness (for example, diabetes). Also, regarding the number of people by injury or illness, it is obtained by counting the number of people indicated in the receipt information including a specific injury or illness (for example, diabetes).

[0071] Regarding the number of days per case by injury or illness, it is obtained by calculating (the number of days by injury or illness) / (the number of cases). Regarding the number of days by injury or illness, the number of days of medical treatment estimated by the number of days of medical treatment by injury or illness estimation unit 13 is used. Regarding the number of cases by injury or illness, as described above, it is obtained by counting the number of receipt information including a specific injury or illness.

[0072] Regarding the medical expenses per day by type of injury or illness, it is obtained by calculating (medical expenses by type of injury or illness) / (number of days by type of injury or illness). For the medical expenses by type of injury or illness, the medical expenses by type of injury or illness estimated by the medical expense estimation unit 15 by type of injury or illness are used. For the number of days by type of injury or illness, the number of medical visit days by type of injury or illness estimated by the medical visit days estimation unit 13 by type of injury or illness are used.

[0073] In addition to the number of cases per person by type of injury or illness, the number of days per case by type of injury or illness, and the medical expenses per day by type of injury or illness described above, the medical expense analysis unit 16 by type of injury or illness calculates (number of cases per person by type of injury or illness) × (number of days per case by type of injury or illness) to calculate the number of days per person by type of injury or illness, or calculates (number of days per case by type of injury or illness) × (medical expenses per day by type of injury or illness) to calculate the medical expenses per case by type of injury or illness, etc., or calculates (number of cases per person by type of injury or illness) × (number of days per case by type of injury or illness) × (medical expenses per day by type of injury or illness) to calculate the medical expenses per person by type of injury or illness.

[0074] In addition, the medical expense analysis unit 16 by type of injury or illness obtains the result of classifying the above calculation results by age group based on the patient's age information included in the receipt information.

[0075] The display control unit 17 causes the analysis result in the medical expense analysis unit 16 by type of injury or illness to be displayed on the display device 30. Specifically, the display control unit 17 text-displays the numerical values calculated by the medical expense analysis unit 16 by type of injury or illness, or graphically displays, as shown in FIG. 4, the medical visit rate (number of cases per person) by type of injury or illness, the number of days per case by type of injury or illness, and the medical expenses per day by type of injury or illness with three axes. In addition, the display control unit 17 graphically displays the analysis result by age group.

[0076] In addition, the display control unit 17 can cause the receipt information, electronic medical record information, and medical treatment record information acquired by the analysis information acquisition unit 11, and the first and second tendency scores calculated by the first and second balancing score calculation units 12, 14, etc. to be displayed on the display device 30.

[0077] The injury / disease-specific medical expense analysis device 10 is composed of a computer equipped with a central processing unit (CPU (Central Processing Unit)), a semiconductor memory, and a storage device such as a hard disk or an SSD (Solid State Drive), and an injury / disease-specific medical expense analysis program including an embodiment of the injury / disease-specific visit days estimation program of the present invention is installed. By executing this injury / disease-specific medical expense analysis program by the central processing unit, the analysis information acquisition unit 11, the first balancing score calculation unit 12, the injury / disease-specific visit days estimation unit 13, the second balancing score calculation unit 14, the injury / disease-specific medical expense estimation unit 15, the medical expense analysis unit 16, and the display control unit 17 shown in FIG. 1 operate.

[0078] The injury / disease-specific medical expense analysis program is recorded and distributed on a recording medium such as a DVD (Digital Versatile Disc) and a CD-ROM (Compact Disc Read Only Memory), and can be installed on a computer from the recording medium. Alternatively, the injury / disease-specific medical expense analysis program is stored in a storage device of a server computer connected to a network or a network storage in a state accessible from the outside. Then, it can be downloaded to and installed on a computer in response to a request from the outside.

[0079] Also, in the present embodiment, the functions of each unit are executed by executing the injury / disease-specific medical expense analysis program by the central processing unit. However, some or all of the functions executed by the injury / disease-specific medical expense analysis may be configured by hardware such as an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or other electric circuits.

[0080] The input device 20 is equipped with devices such as a keyboard and a mouse. The display device 30 is equipped with devices such as a liquid crystal display. The input device 20 and the display device 30 may be configured from a touch panel and used in common.

[0081] In the injury / disease-specific medical expense analysis system 1 of the above-described embodiment, medical act information that is considered appropriate as a confounding factor is selected as the covariate x=(x 1 ,x 2 ,x 3 ,···). However, when the analysis information is the above-described electronic medical record information or medical record information, since the information indicating the interview results, findings, image findings, and test values is also included, these may also be included in the covariate. For the interview results and findings, for example, information on the presence or absence of symptoms such as "cough" can be used as a numerical value as the covariate, and for the image findings, for example, information on the presence or absence of diseases such as "tumor in the image" can be used as a numerical value as the covariate.

[0082] Conversely, for the receipt information, since the information indicating the interview results, findings, image findings, and test values as described above is not included, for example, at least one of the health examination information and the screening information of the target patient of the receipt information is separately obtained, and information indicating the interview results, findings, image findings, and test values is extracted from the information and included in the covariate. Thereby, the information of the covariate can be increased, so that more appropriate first and second propensity scores can be calculated, and the number of days of medical visits by injury / disease and the medical expenses by injury / disease can be estimated with higher accuracy.

[0083] In addition, the receipt information includes information related to the functions of the medical institution that provides medical treatment to the patient. Since this information also includes the medical supply system and functions related to diseases, this information may be extracted from the receipt information and included in the covariate.

[0084] In the medical expense analysis system 1 by injury type of the above-described embodiment, the first and second propensity scores are calculated using logistic regression analysis. However, the method for calculating the first and second propensity scores is not limited to this. Basically, as long as it is defined by the following formula e i any calculation method may be used. In the following formula, x i is the value of the covariate of the analysis information i, and z i is the value of the assignment variable (in the above-described embodiment, the variable by injury type, 1 or 0 depending on the presence or absence of injury), and the probability e i assigned to group 1 is the propensity score.

Number

[0085] For example, instead of the logistic regression model as in the above-described embodiment, a probit regression model may be used. Also, methods using LASSO (sparse estimation) and methods obtained from kernel regression models can also be used.

[0086] In the medical expense analysis system 1 by injury type of the above-described embodiment, after calculating the first and second propensity scores of each analysis information, the matching method is used to estimate the number of days of medical treatment by injury type and the medical expense information by injury type. However, the method for estimating the number of days of medical treatment by injury type and the medical expense by injury type is not limited to the matching method, and any other method may be used.

[0087] Specifically, methods using stratified analysis and covariance analysis can be used. In stratified analysis, the first propensity score or the second propensity score is used to divide into K subclasses (the sizes of the subclasses are the same), perform the same operations as in matching, and finally take the average value of the differences to obtain the estimated value of the difference in the number of days of medical treatment or the estimated value of the difference in medical expenses. Also, the method using covariance analysis is a method of performing linear regression analysis with the assignment variable z of the group with injury and the group without injury and the first or second propensity score e as explanatory variables and the number of days of medical treatment or medical expenses as the objective variable.

[0088] In addition, the IPW (Inverse probability weighting) method and the DR (Doubly Robust) method can also be used. Further, for example, when using the IPW method, as the number of medical visits, the average number of medical visits y1 of the group with injuries and the average number of medical visits y0 of the group without injuries, as well as their variances, etc. can be estimated, or as the medical expenses, the average medical expenses y1 of the group with injuries and the average medical expenses y0 of the group without injuries, as well as their variances, etc. can be estimated.

[0089] Note that in the IPW method and the DR method, the number of medical visits of the analysis information belonging to the two divided groups, which has an explanatory variable indicating that it includes at least a predetermined injury condition information, and the balancing score of the analysis information having an explanatory variable indicating that it includes at least a predetermined injury condition information are used to estimate the number of medical visits by injury type. Similarly, in the IPW method and the DR method, the medical expenses of the analysis information belonging to the two divided groups, which has an explanatory variable indicating that it includes at least a predetermined injury condition information, and the balancing score of the analysis information having an explanatory variable indicating that it includes at least a predetermined injury condition information are used to estimate the medical expenses by injury type.

[0090] Also, regarding which method to use to estimate the number of medical visits by injury type and the medical expenses by injury type, it is appropriately selected based on the focused injury condition information, the nature and data structure of the available analysis information data, and its purpose of use (whether only the difference between with and without injuries is sufficient, or individual estimation is desired, etc.).

[0091] In addition, in the injury - specific medical expense analysis system 1 of the above - described embodiment, the propensity score is used as the balancing score, but not limited to the propensity score, other balancing scores may be used.

[0092] Specifically, for the group with injuries (z i = 1) and the group without injuries (z iFor all covariates x that are characterized to be independent of each other with respect to (x = 0), i they may be directly used as a "vector" as the balancing score. That is, in the above embodiment, the selected covariate x i is used to perform a logistic regression analysis on the group with injuries (z i = 1) to calculate the propensity score. However, this operation can be omitted, and the vector (x 1 , x 2 , …, x n ) can be directly used as the balancing score, and pairs with exactly the same vector can be selected from the group without injuries and matching can be performed.

[0093] Regarding the balancing score, for example, the balancing score defined in "Causal Inference for Statistics, Social, and Biomedical Sciences: An Introduction" by Guido W. Imbens and Donald B. Rubin can be used.

[0094] Note that the present invention is not limited to the above embodiment, and components can be modified and embodied without departing from the gist thereof at the implementation stage. Also, various inventions can be formed by appropriately combining a plurality of components disclosed in the above embodiment. For example, all the components shown in the embodiment may be appropriately combined. Of course, various modifications and applications are possible within the scope not departing from the gist of the invention.

[0095] The following additional remarks are further disclosed regarding the present invention.

[0096] (Additional Remark 1) The injury / disease-specific medical visit days estimation device of the present invention receives a plurality of analysis information including at least one injury / disease information of a patient, at least one medical treatment information applied to the patient, and the medical visit days related to the medical treatment information, and based on the received plurality of analysis information, a balancing score calculation unit that calculates a balancing score for each analysis information, and a medical visit days estimation unit that estimates the medical visit days for each injury / disease information based on the balancing score for each analysis information.

[0097] (Appendix 2) In the injury / disease-specific medical visit days estimation device described in Appendix 1, receipt information can be included in the plurality of analysis information.

[0098] (Appendix 3) In the injury / disease-specific medical visit days estimation device described in Appendix 1 or Appendix 2, electronic medical record information can be included in the plurality of analysis information.

[0099] (Appendix 4) In the injury / disease-specific medical visit days estimation device described in any one of Appendices 1 to 3, a medical visit days extraction unit that extracts the medical visit days from the analysis information is provided, and the balancing score calculation unit can calculate the balancing score using the medical visit days extracted by the medical visit days extraction unit.

[0100] (Appendix 5) In the injury / disease-specific medical visit days estimation device described in any one of Appendices 1 to 4, an injury / disease information extraction unit that extracts the injury / disease information from the analysis information is provided, and the balancing score calculation unit can calculate the balancing score using the injury / disease information extracted by the injury / disease information extraction unit.

[0101] (Appendix 6) In the injury / disease-specific medical visit days estimation device described in any one of Appendices 1 to 5, a medical treatment information extraction unit that extracts the medical treatment information from the analysis information is provided, and the balancing score calculation unit can calculate the balancing score using the medical treatment information extracted by the medical treatment information extraction unit.

[0102] (Appendix 7) In the injury / disease-specific medical visit days estimation device according to any one of Appendices 1 to 6, the balancing score calculation unit can receive at least one of the patient's health diagnosis information and screening information, and calculate a balancing score using the at least one piece of information.

[0103] (Appendix 8) In the injury / disease-specific medical visit days estimation device according to any one of Appendices 1 to 7, a propensity score can be used as the balancing score.

[0104] (Appendix 9) The method for estimating the number of medical visits by injury / disease of the present invention prepares a plurality of analysis information including at least one piece of injury / disease information of a patient, at least one piece of medical treatment information applied to the patient, and the number of medical visits related to the medical treatment information. Using an arithmetic device, based on the plurality of analysis information, a balancing score for each analysis information is calculated, and based on the calculated balancing score for each analysis information, the number of medical visits for each injury / disease information is estimated.

[0105] (Appendix 10) The program for estimating the number of medical visits by injury / disease of the present invention causes a computer to function as a balancing score calculation unit that receives a plurality of analysis information including at least one piece of injury / disease information of a patient, at least one piece of medical treatment information applied to the patient, and the number of medical visits related to the medical treatment information, and calculates a balancing score for each analysis information based on the received plurality of analysis information, and a medical visit days estimation unit that estimates the number of medical visits for each injury / disease information based on the balancing score for each analysis information.

[0106] (Appendix 11) The injury - specific number of outpatient visits estimation device of the present invention receives a plurality of analysis information including at least one piece of injury information of a patient, at least one piece of medical treatment information performed on the patient, and the number of outpatient visits related to the medical treatment information. Based on each received analysis information, it has an explanatory variable indicating whether a predetermined injury information is included in each analysis information, and calculates a balancing score using the medical treatment information included in each analysis information as a covariate. The balancing score calculation unit divides the analysis information into a group of analysis information having an explanatory variable indicating that it includes a predetermined injury information and a group of analysis information having an explanatory variable indicating that it does not include a predetermined injury information, and uses at least the number of outpatient visits of the analysis information having an explanatory variable indicating that it includes a predetermined injury information among the number of outpatient visits of the analysis information belonging to the two divided groups and the balancing score of the analysis information having an explanatory variable indicating that it includes at least a predetermined injury information to estimate the number of outpatient visits for each type of injury information.

Explanation of reference numerals

[0107] 1 Medical expense analysis system by injury type 10 Medical expense analysis device by injury type 11 Analysis information acquisition unit 12 First balancing score calculation unit 12a First information extraction unit 13 Injury - specific number of outpatient visits estimation unit 14 Second balancing score calculation unit 14a Second information extraction unit 15 Injury - specific medical expense estimation unit 16 Injury - specific medical expense analysis unit 17 Display control unit 20 Input device 30 Display device

Claims

1. A balancing score calculation unit that receives a plurality of analysis information including at least one injury condition information of a patient, at least one medical treatment information applied to the patient, and the number of days of medical treatment related to the medical treatment information, and calculates a balancing score for each of the received analysis information; An injury-specific medical treatment days estimation device comprising a medical treatment days estimation unit that estimates the number of days of medical treatment for each injury condition information based on the balancing score for each of the analysis information.

2. The injury-specific medical treatment days estimation device according to claim 1, wherein the plurality of analysis information includes receipt information.

3. The injury-specific medical treatment days estimation device according to claim 1, wherein the plurality of analysis information includes electronic medical record information.

4. Comprising a medical treatment days extraction unit that extracts the number of days of medical treatment from the analysis information, The injury-specific medical treatment cost estimation device according to claim 1, wherein the balancing score calculation unit calculates the balancing score using the number of days of medical treatment extracted by the medical treatment days extraction unit.

5. Comprising an injury condition information extraction unit that extracts the injury condition information from the analysis information, The injury-specific medical treatment days estimation device according to claim 1, wherein the balancing score calculation unit calculates the balancing score using the injury condition information extracted by the injury condition information extraction unit.

6. Comprising a medical treatment information extraction unit that extracts the medical treatment information from the analysis information, The injury-specific medical treatment days estimation device according to claim 1, wherein the balancing score calculation unit calculates the balancing score using the medical treatment information extracted by the medical treatment information extraction unit.

7. The injury-specific medical treatment days estimation device according to claim 1, wherein the balancing score calculation unit receives at least one of the patient's health examination information and screening information, and calculates the balancing score using the at least one information.

8. The injury-specific medical treatment days estimation device according to claim 1, wherein the balancing score is a propensity score.

9. Prepare a plurality of analysis information including at least one injury condition information of a patient, at least one medical treatment information applied to the patient, and the number of days of medical treatment related to the medical treatment information, A method for estimating the number of days of medical visits for each type of injury or illness, which uses a computing device to calculate a balancing score for each piece of analysis information based on the plurality of pieces of analysis information, and estimates the number of days of medical visits for each piece of injury or illness information based on the calculated balancing scores for each piece of analysis information.

10. A computer, receives a plurality of pieces of analysis information including at least one piece of injury or illness information of a patient, at least one piece of medical treatment information performed on the patient, and the number of days of medical visits related to the medical treatment information; a balancing score calculation unit that calculates a balancing score for each piece of analysis information based on the received plurality of pieces of analysis information; An injury / illness-specific medical visit days estimation program that functions as a medical visit days estimation unit that estimates the number of days of medical visits for each piece of injury or illness information based on the balancing scores for each piece of analysis information.

11. receives a plurality of pieces of analysis information including at least one piece of injury or illness information of a patient, at least one piece of medical treatment information performed on the patient, and the number of days of medical visits related to the medical treatment information; has an explanatory variable indicating whether a predetermined piece of injury or illness information is included in each piece of analysis information based on each received piece of analysis information, and calculates a balancing score using the medical treatment information included in each piece of analysis information as a covariate; a balancing score calculation unit; An injury / illness-specific medical visit days estimation device comprising: a medical visit days estimation unit that divides the group of analysis information having an explanatory variable indicating inclusion of the predetermined injury or illness information and the group of analysis information having an explanatory variable indicating non-inclusion of the predetermined injury or illness information, and uses at least the number of days of medical visits of the analysis information having an explanatory variable indicating inclusion of the predetermined injury or illness information among the number of days of medical visits of the analysis information belonging to the two divided groups and the balancing scores of the analysis information having an explanatory variable indicating inclusion of at least the predetermined injury or illness information to estimate the number of days of medical visits for each piece of injury or illness information.

Images