Device, program, and method for supporting the estimation of suspected drugs for side effects

Abstract

To support the estimation of a suspect drug for a side effect expressing in a patient.SOLUTION: A suspect drug estimation support device 1 for a side effect according to the present invention is used to estimate a suspect drug having the possibility of a causative drug being the cause of an expression of a side effect expressing in an object patient from among a plurality of administering drugs administered to the object patient. An information processing device includes an information acquisition part 134 for acquiring candidate side effect information showing candidate side effects to be a candidate for the expression of the side effect, and administering drug information showing administering drugs, a probability calculation part 135 for calculating a conditional probability that an administering drug is a suspect drug in each of the plurality of administering drugs on condition that the candidate side effects are the expression of the side effect on the basis of the candidate side effect information and the administering drug information, and a storage part 12 for storing the calculated conditional probability.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a suspect drug estimation support device for side effects, a suspect drug estimation support program, and a suspect drug estimation support method. [Background technology]

[0002] In recent years, with the advancement of drug development, there has been an increase in the number of cases where multiple drugs are administered simultaneously or for overlapping periods for various diseases. In addition, with an aging society, the number of patients with multiple comorbidities is increasing. As a result, there has been an increase in cases where a single patient is administered multiple drugs (so-called polypharmacy).

[0003] Patients administered a drug may experience not only medicinal effects but also side effects. If a side effect occurs, depending on the type or severity of the side effect, administration of the drug causing the side effect (the causative drug) must be suspended or discontinued. Therefore, when a side effect occurs, it is necessary to identify the drug causing the side effect. However, as mentioned above, when a patient is administered multiple drugs, side effects may occur in common with multiple drugs. Therefore, identifying the drug causing the side effect is not easy.

[0004] A system has been proposed to identify a drug (suspect drug) that is suspected to be the cause of symptoms based on symptoms that appear in a patient who has been administered multiple drugs (see, for example, Patent Document 1).

[0005] In the system disclosed in Patent Document 1, symptoms that a patient experiences due to side effects are input, combinations of drugs and side effects corresponding to the input symptoms are extracted, a ranking of the drugs and side effects corresponding to the symptoms is determined, and the combinations are displayed in order of highest ranking. As a result, doctors or pharmacists can sequentially identify the combinations that cause the symptoms, improving work efficiency. However, while this system directly extracts and infers the causative drug from the symptoms, it does not extract or infer the causative drug from the side effects. In other words, side effects diagnosed by doctors are not input (or taken into account) in this system.

[0006] Furthermore, in calculating specificity, which is an index for ranking, it is important to determine whether the administered drug is in the "same group." However, the system does not clearly define "the same group," making specificity meaningless as an index.

[0007] Furthermore, the system identifies the frequency of side effects, which is an index for ranking, based on the drug's package insert. However, the system does not reflect the actual clinical incidence of side effects for each drug, making it difficult to automatically extract the frequency of side effects from the package insert. Furthermore, manually inputting the frequency of side effects for each of the vast number of drugs is not practical, and package inserts list many side effects whose frequency of occurrence is unknown. As a result, the system's estimation accuracy is extremely low. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-21081 Summary of the Invention [Problem to be solved by the invention]

[0009] The present invention aims to support the estimation of suspected drugs that cause side effects in patients. [Means for solving the problem]

[0010] The suspected drug estimation support device for side effects of the present invention is a device used to estimate a suspected drug that may be the causative drug that is causing the side effect that has occurred in a target patient, from among multiple administered drugs to the target patient, and is characterized by having an information acquisition unit that acquires candidate side effect information indicating candidate side effects that are candidates for the side effect that has occurred and administered drug information indicating the administered drug, a probability calculation unit that calculates, based on the candidate side effect information and the administered drug information, the condition that the candidate side effect is an occurred side effect, for each of the multiple administered drugs, that the administered drug is a suspected drug, and a memory unit that stores the calculated conditional probability. [Effects of the Invention]

[0011] According to the present invention, it is possible to assist in estimating a drug suspected of causing a side effect occurring in a patient. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a network configuration diagram showing an embodiment of a support device for estimating a suspected drug for a side effect according to the present invention. [Figure 2] 2 is a functional block diagram showing an embodiment of the suspect drug estimation support device of FIG. 1. FIG. [Figure 3] 2 is a schematic diagram showing an example of information stored in a storage unit included in the suspect drug estimation support device of FIG. 1. FIG. [Figure 4] FIG. 4 is a schematic diagram showing another example of the information in FIG. 3. [Figure 5] FIG. 4 is a schematic diagram showing yet another example of the information in FIG. 3. [Figure 6] 2 is a flowchart showing the operation of the suspect drug estimation support device of FIG. 1. [Figure 7] 2 is a flowchart of a candidate side effect extraction process executed by the suspect drug estimation support device of FIG. 1. [Figure 8] FIG. 8 is a schematic diagram visually showing an example of calculation of the incidence rate in the candidate side effect extraction process of FIG. 7. [Figure 9] 2 is a schematic diagram showing an example of information displayed on a display unit included in the suspect drug estimation support device of FIG. 1. FIG. [Figure 10]2 is a flowchart of a probability calculation process executed by the suspect drug estimation support device of FIG. 1. [Figure 11] 10 is a schematic diagram showing another example of information displayed on the display unit of FIG. 9. FIG. [Figure 12] 1 is a graph showing the ranking of conditional probabilities for each of the causative drug and the concomitant drug. [Figure 13] (a) is a graph showing the number of times that the causative drug was counted in the ranking assigned by the suspect drug estimation support device in Figure 1, and (b) is a graph showing the number of times that the concomitant medication was counted in the ranking assigned by the same device. DETAILED DESCRIPTION OF THE INVENTION

[0013] Below, with reference to the drawings, we will explain the embodiments of the side effect suspect drug estimation support device (hereinafter referred to as "this device"), side effect suspect drug estimation support program (hereinafter referred to as "this program"), and side effect suspect drug estimation support method (hereinafter referred to as "this method") related to the present invention.

[0014] The present invention supports specific estimation of candidate side effects by extracting candidate side effects (hereinafter referred to as "candidate side effects") for side effects occurring in a target patient (hereinafter referred to as "appearing side effects") from symptoms occurring in the target patient to whom multiple drugs have been administered (hereinafter referred to as "appearing symptoms"). The present invention also supports specific estimation of suspect drugs by indicating an index of the possibility that each administered drug is a suspect drug from the appeared side effects or candidate side effects.

[0015] The "subject patient" is a patient for whom a suspected drug causing the manifested symptoms is suspected by the present invention, and is a patient to whom the suspected drug is administered.

[0016] An "administered drug" is a drug administered to a subject patient. In the present invention, a subject patient is administered multiple drugs.

[0017] A "suspected drug" is a drug that is administered and is suspected (or may be suspected) of being the drug that is causing the symptoms of the side effect (hereinafter referred to as the "causative drug").

[0018] ●Support device for estimating suspected drugs for side effects● First, an embodiment of the present device will be described.

[0019] FIG. 1 is a network configuration diagram showing an embodiment of the present device.

[0020] The device 1 specifically extracts candidate side effects from the symptoms that have appeared, and also specifically extracts suspect drugs from the candidate side effects. The specific configuration of the device 1 will be described later.

[0021] The external device 2 is a device that transmits and receives information to and from the present device 1 via the network N. The external device 2 is, for example, a server that stores information that forms the basis of a database described below, an information processing terminal that receives each piece of information output from the present device 1, or a printer.

[0022] The network N is, for example, a communication network such as the Internet, a mobile communication network, a local area network (LAN), a wide area network (WAN), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0023] Configuration of the device to support the estimation of suspected drugs for side effects FIG. 2 is a functional block diagram showing an embodiment of the device 1. As shown in FIG.

[0024] The device 1 is realized, for example, by a personal computer. The program runs on the device 1, and the program cooperates with the hardware resources of the device 1 to realize the method.

[0025] Here, by causing a computer (not shown) to execute this program, the program can cause the computer to function in the same manner as the device 1, and cause the computer to execute this method.

[0026] The device 1 includes a communication unit 11, a storage unit 12, a control unit 13, an operation unit 14, and a display unit 15.

[0027] The communication unit 11 is connected to the external device 2 via a network N. The communication unit 11 is configured by, for example, a communication module and a communication interface.

[0028] The storage unit 12 stores information (for example, DB1, DB2, etc., which will be described later) necessary for the device 1 to execute the method, which will be described later. The storage unit 12 is configured, for example, by a recording device such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive) provided in the device 1, and / or a portable storage medium such as a flash memory.

[0029] The control unit 13 controls the overall operation of the device 1 and executes the method described below. The control unit 13 is configured, for example, with a CPU (Central Processing Unit) provided in the device 1, a RAM (Random Access Memory) that functions as a work area for the CPU, and a ROM (Read Only Memory) that stores various information such as the program. The control unit 13 includes a symptom acquisition unit 131, a side effect extraction unit 132, a ratio calculation unit 133, an information acquisition unit 134, a probability calculation unit 135, an index generation unit 136, a ranking unit 137, and an output unit 138.

[0030] The symptom acquisition unit 131 acquires information indicating symptoms (present symptoms) occurring in the target patient (hereinafter referred to as "present symptom information"). Specific operations of the symptom acquisition unit 131 will be described later.

[0031] The side effect extraction unit 132 extracts side effects corresponding to the manifested symptoms as candidate side effects based on the manifested symptom information and the side effect-subjective symptom correspondence information DB 1. The specific operation of the side effect extraction unit 132 will be described later.

[0032] The "side effect-subjective symptom correspondence information DB1 (hereinafter simply referred to as "DB1")" is information (database) showing the correspondence between information indicating each of a plurality of side effects (hereinafter referred to as "side effect information") and information indicating subjective symptoms that may occur due to each side effect (hereinafter referred to as "symptom information"). That is, DB1 stores (registers) subjective symptoms that may occur for each side effect. In this embodiment, DB1 is generated based on, for example, the "side effect glossary for patients" of the Pharmaceuticals and Medical Devices Agency (PMDA). The side effects stored (registered) in DB1 are side effects that occurred in various patients who underwent drug treatment in the past and were reported as side effects caused by the administered drugs. As described above, candidate side effects are extracted based on DB1, and therefore the side effect information stored in DB1 includes information indicating the extracted candidate side effects (hereinafter referred to as "candidate side effect information"). Furthermore, the subjective symptoms stored (registered) in DB1 are subjective symptoms that actually occurred in patients who experienced the side effects registered in DB1. DB1 is stored in the storage unit 12, for example.

[0033] FIG. 3 is a schematic diagram showing an example of information (DB1) stored in the storage unit 12. As shown in FIG. The figure shows that, for example, the side effect of side effect information "X01" is associated with subjective symptoms of symptom information "Y01," "Y02," and "Y03" and stored in the storage unit 12.

[0034] FIG. 4 is a schematic diagram showing another example of information (DB1) stored in the storage unit 12. As shown in FIG. The right axis of the figure lists "m (m is an integer)" types of subjective symptom information, and the left axis lists "n (n is an integer)" types of side effect information. The lines connecting the subjective symptom information and side effect information show the mutual association.

[0035] 3 and 4, the device 1 can, for example, refer to DB1 using the subjective symptom information "Y01" to read out the side effect information "X01" and "X02" that is associated with the subjective symptom information "Y01" and stored in the storage unit 12. In addition, the device 1 can, for example, refer to the side effect-subjective symptom correspondence information DB1 using the side effect information "X01" to read out the subjective symptom information "Y01", "Y02", and "Y03" that is associated with the side effect information "X01" and stored in the storage unit 12.

[0036] Return to Figure 2. For each extracted candidate side effect, the ratio calculation unit 133 calculates the ratio (hereinafter referred to as "occurrence ratio") of the manifested symptom to all subjective symptoms registered in DB1 as subjective symptoms corresponding to the candidate side effect (manifested by the candidate side effect). The specific operation of the ratio calculation unit 133 will be described later.

[0037] The information acquiring unit 134 acquires candidate side effect information and information indicating the administered drug administered to the subject patient (hereinafter referred to as "administered drug information"). Specific operations of the information acquiring unit 134 will be described later.

[0038] The probability calculation unit 135 calculates, for each of multiple administered drugs, a conditional probability that the administered drug is a suspect drug, under the condition that the candidate side effect is a manifested side effect, based on the candidate side effect information, administered drug information, and side effect-related drug report information DB 2. The specific operation of the probability calculation unit 135 will be described later.

[0039] "Adverse Reaction-Related Drug Report Information DB2 (hereinafter simply referred to as "DB2")" is information (database) showing the correspondence between information indicating each of a plurality of adverse reactions (adverse reaction information) and related drug information indicating related drugs associated with the adverse reaction information. That is, in DB2, adverse reaction information for each patient and related drug information are associated and stored (registered). In this embodiment, DB2 is generated based on, for example, the "Adverse Reaction Report Database: JADER (Japanese Adverse Drug Event Report database)" published by the PMDA. The adverse reaction information stored in DB2 includes the adverse reaction information stored in DB1. That is, the adverse reaction information stored in DB2 includes candidate adverse reaction information indicating extracted candidate adverse reactions.

[0040] "Related drugs" are drugs reported as being related to the adverse drug reactions experienced by various patients in the past who experienced adverse drug reactions due to medication (i.e., drugs administered to the patient at the time of the adverse drug reaction). When a drug identified as the cause of the adverse drug reaction (hereinafter referred to as a "specific causative drug") is reported, the related drugs for the adverse drug reaction include the specific causative drug and drugs administered in combination with the specific causative drug (hereinafter referred to as a "concomitant drug"). On the other hand, when a specific causative drug is not reported, the related drugs include only the concomitant drug. That is, for each adverse drug reaction information, DB2 stores information indicating the specific causative drug (hereinafter referred to as "specific causative drug information") and information indicating the concomitant drug (hereinafter referred to as "concomitant drug information"), or concomitant drug information, in association with each adverse drug reaction. Here, adverse drug reactions and related drugs may vary from patient to patient (depending on the patient's constitution). Therefore, DB2 may store multiple pieces of the same type of adverse drug reaction information.

[0041] FIG. 5 is a schematic diagram showing an example of information (DB2) stored in the storage unit 12. As shown in FIG. The figure shows that the related drug information "Z01," "Z02," "Z03," and "Z04" for the side effect of the side effect information "X11" are associated with each other and stored in the memory unit 12. The figure also shows that the related drug information "Z01," "Z02," "Z03," and "Z04" for the side effect information "X11" includes the specific causative drug information "Z01" and the concomitant drug information "Z02," "Z03," and "Z04." The figure also shows that no causative drug has been identified for the side effect of the side effect information "X01." As shown in FIG. 5, the device 1 can, for example, refer to DB2 using the side effect information "X11" to read the related drug information (specific causative drug information "Z01" and concomitant drug information "Z02," "Z03," and "Z04") associated with the side effect information "X11" and stored in the memory unit 12. Furthermore, the device 1 can obtain the number of all pieces of side effect information "X11" stored in DB2 by referring to DB2 using the side effect information "X11", for example.

[0042] Return to Figure 2. The index generating unit 136 generates an index (hereinafter referred to as "estimation index") used to estimate a suspect drug based on the conditional probability. The specific operation of the index generating unit 136 and the estimation index will be described later.

[0043] The ranking unit 137 ranks the candidate side effects based on the calculated occurrence rates. The ranking unit 137 also ranks the administered drugs based on the estimated information. The specific operation of the ranking unit 137 will be described later.

[0044] The output unit 138 outputs various information (for example, estimated indexes, candidate side effect information, etc.) obtained by executing the method described later. Specific operations of the output unit 138 will be described later.

[0045] The operation unit 14 is a device that is operated (for example, to input and select information) by a user (for example, a doctor or nurse; hereinafter simply referred to as "user") of the device 1. The operation unit 14 is, for example, a keyboard, a mouse, or a touch panel.

[0046] The display unit 15 is a device that displays information (for example, evaluation information) output by the output unit 138. The display unit 15 is, for example, a monitor or a display.

[0047] The operation unit and the display unit in the present invention may be configured, for example, as a touch panel display.

[0048] Operation of the device for supporting the estimation of suspected drugs causing side effects (method for supporting the estimation of suspected drugs causing side effects) Next, a description will be given of the operation of the device 1, that is, the method executed by the device 1. In the following description of the method, reference will also be made to FIG.

[0049] FIG. 6 is a flowchart showing the operation of the device 1.

[0050] The present device 1 executes a candidate side effect extraction process (S1) and a probability calculation process (S2). As shown in Fig. 6, in the present device 1, the probability calculation process (S2) may be executed after the candidate side effect extraction process (S1), or may be executed without executing the candidate side effect extraction process (S1). In other words, the present device 1 can selectively execute a process that executes the candidate side effect extraction process (S1) and the probability calculation process (S2) (hereinafter referred to as "first process"), and a process that executes only the probability calculation process (S2) (hereinafter referred to as "second process").

[0051] ●Candidate side effect extraction process FIG. 7 is a flowchart of the candidate side effect extraction process (S1).

[0052] The "candidate side effect extraction process (S1)" is a process for extracting candidate side effects occurring in a target patient from the symptoms (present symptoms) occurring in the target patient. The candidate side effect extraction process (S1) is a process executed before the probability calculation process (S2).

[0053] First, the symptom acquisition unit 131 acquires the onset symptom information of the target patient (S101). Specifically, when the onset symptom information is stored in advance in the storage unit 12, the symptom acquisition unit 131 acquires the onset symptom information of the target patient from the storage unit 12, for example, by using information (such as an ID) indicating the target patient. On the other hand, when the onset symptom information is not stored in advance in the storage unit 12, the symptom acquisition unit 131 acquires, for example, the onset symptom information input by the user via the operation unit 14.

[0054] Next, the side effect extracting unit 132 selects one piece of onset symptom information from which no candidate side effect has been extracted, from among the acquired onset symptom information (S102).

[0055] Next, the side effect extraction unit 132 extracts side effects corresponding to the manifested symptoms as candidate side effects based on the selected manifested symptom information and DB1 (S103). Specifically, the side effect extraction unit 132 refers to DB1 using the manifested symptom information (i.e., subjective symptom information) to extract side effect information associated with the manifested symptom information. Next, the side effect extraction unit 132 identifies side effects indicated (identified) by the extracted side effect information as candidate side effects.

[0056] Next, the side effect extraction unit 132 determines whether or not there is any information on developed symptoms that has not been selected (unselected) in step S102 among the acquired information on developed symptoms (S104).

[0057] If there is unselected onset symptom information ("Y" in S104), the side effect extraction unit 132 repeats the processes S102 to S104.

[0058] On the other hand, when there is no unselected symptom information ("N" in S104), the ratio calculation unit 133 acquires the number of all subjective symptoms (hereinafter referred to as "total symptom count") that are associated with and stored (registered) for each extracted candidate side effect (S105). Specifically, for each extracted candidate side effect, the ratio calculation unit 133 identifies all subjective symptom information associated with the side effect information corresponding to the candidate side effect based on the side effect information corresponding to the candidate side effect and DB1, and acquires this number as the total symptom count.

[0059] Next, the ratio calculation unit 133 acquires, for each extracted candidate side effect, the number of manifested symptoms that correspond to subjective symptoms that may occur due to the candidate side effect (hereinafter referred to as the "number of relevant symptoms") from among the manifested symptoms (S106). Specifically, for each candidate side effect, the ratio calculation unit 133 acquires, from the manifested symptom information acquired in process S101, the number that corresponds to subjective symptom information associated with side effect information corresponding to the candidate side effect, as the number of relevant symptoms.

[0060] Next, the rate calculation unit 133 calculates the incidence rate for each of the extracted candidate side effects (S107). Specifically, the rate calculation unit 133 calculates the incidence rate as the rate of the number of relevant symptoms to the total number of symptoms ((number of relevant symptoms / total number of symptoms) × 100(%)). The calculated incidence rate is stored in the storage unit 12 in association with the corresponding side effect information.

[0061] FIG. 8 is a schematic diagram visually showing an example of calculation of the expression ratio. For ease of explanation, in the figure, of the subjective symptom information, the manifested symptom information is shown enclosed in a solid frame, and the subjective symptom information indicating a non-manifested symptom (a subjective symptom not manifested in the subject patient) is shown enclosed in a dashed frame. The numbers in the figure indicate the incidence rate. The figure shows that the incidence rate of the side effect corresponding to side effect information "X11" is 100%, and the incidence rate of the side effect corresponding to side effect "X01" is 50%.

[0062] Return to Figure 7. Next, the ranking unit 137 ranks all of the extracted candidate side effects based on the calculated occurrence rates (S108). Specifically, the ranking unit 137 compares the occurrence rates for each candidate side effect and ranks the candidate side effects in descending order of occurrence rate. Here, candidate side effects with the same occurrence rate are assigned the same rank. The ranks are stored in the storage unit 12 in association with the corresponding occurrence rates and corresponding side effect information.

[0063] Next, the output unit 138 outputs the side effect information, incidence rate, and ranking, which are associated with each other, to the display unit 15 (S109).

[0064] Next, the display unit 15 displays the candidate side effects and the incidence rates in order of rank (S110).

[0065] Figure 9 is a schematic diagram showing an example of information (candidate side effects, occurrence rates) displayed on the display unit 15. The figure shows that five side effects are arranged in descending order of occurrence rate as candidate side effects and displayed on the display unit 15. For ease of explanation, the figure indicates the candidate side effects with the symbols used in the candidate side effect information.

[0066] As described above, candidate side effects are extracted based on the subjective symptoms (present symptoms) actually occurring in the target patient. The candidate side effects are ranked based on the proportion (present symptom) of present symptoms relative to the subjective symptoms that may occur for each candidate side effect. Therefore, by displaying the candidate side effects in ranked order on the display unit 15, the device 1 can provide the user with useful information for specifically estimating candidate side effects from present symptoms (for supporting the estimation of candidate side effects). That is, for example, if the user is someone who cannot diagnose the target patient (e.g., a nurse), the user can specifically estimate candidate side effects from present symptoms by viewing the display unit 15. Furthermore, for example, if the user is a doctor, the user can obtain information on other candidate side effects that are different from the results of their own diagnosis (side effects) by viewing the display unit 15.

[0067] In the candidate side effect extraction process, the display unit of the present invention may display only the candidate side effects in order of rank.

[0068] ●Probability calculation processing Next, the probability calculation process (S2) will be described.

[0069] FIG. 10 is a flowchart of the probability calculation process (S2).

[0070] First, the information acquisition unit 134 acquires administered drug information for the target patient (S201). Specifically, when administered drug information is pre-stored in the storage unit 12, the information acquisition unit 134 acquires the administered drug information for the target patient from the storage unit 12, for example, by using information indicating the target patient. On the other hand, when administered drug information is not pre-stored in the storage unit 12, the information acquisition unit 134 acquires administered drug information input by the user via the operation unit 14, for example.

[0071] Next, the information acquisition unit 134 acquires candidate side effect information for the target patient (S202). Specifically, when candidate side effect information has been extracted in advance by the candidate side effect extraction process (S1) and stored in the storage unit 12, the information acquisition unit 134 acquires the candidate side effect information for the target patient from the storage unit 12, for example, by using information indicating the target patient. On the other hand, when candidate side effect information is not stored in the storage unit 12, the information acquisition unit 134 acquires, for example, side effect information input by the user via the operation unit 14 as the candidate side effect information. In this case, the side effect information input by the user is obtained, for example, by a doctor's diagnosis.

[0072] Next, the probability calculation unit 135 selects one piece of candidate side effect information for which a conditional probability, which will be described later, has not been calculated from the acquired candidate side effect information (S203).

[0073] Next, the probability calculation unit 135 calculates, for each of the multiple administered drugs, a conditional probability that the administered drug is a suspect drug, based on the selected candidate side effect information, the administered drug information, and DB2, under the condition that the candidate side effect is a manifested side effect (S204). A Bayesian estimation model, which will be described later, is used to calculate the conditional probability. The Bayesian estimation model is, for example, stored in advance in the storage unit 12.

[0074] Here, there are k kinds of drugs, D={d1,d2,...,d i ,···d k When a patient receiving a drug "d i If " is the suspected drug, "d i =c" is written as "d" i The conditional probability that " is a suspect drug is expressed by the Bayesian estimation model as follows:

[0075]

number

[0076] Here, equation (1) is expressed as the following equation (2).

[0077]

number

[0078] The conditional probability calculated in this manner functions as an index that probabilistically indicates the possibility that each administered drug will be a suspect drug for a candidate adverse drug reaction. In other words, the conditional probability can be an example of an estimation index in the present invention. Therefore, the probability calculation unit 135 can also function as an index generation unit in the present invention. The conditional probability is, for example, associated with candidate adverse drug reaction information and administered drug information and stored in the storage unit 12. Here, there are cases where the calculation of the conditional probability is theoretically possible but practically impossible (for example, because the probability "P(ADR)" of the occurrence of the adverse drug reaction "ADR" cannot be directly calculated from DB2). Such cases may exist for all administered drugs for the adverse drug reaction "ADR." Therefore, the probability calculation unit 135 calculates the conditional probability using the proportional relationship shown in Equation (2). As a result, the conditional probability calculated by the probability calculation unit 135 differs from the true value of the conditional probability that each administered drug will be a suspect drug, but the magnitude relationship between the conditional probabilities of the administered drugs is maintained.

[0079] Next, the index generating unit 136 calculates a probability score based on the calculated conditional probability. Specifically, the index generating unit 136 calculates the probability score using the following formula (3), which is expressed by taking the logarithm of formula (2) (S205). By taking the logarithm in this way, it becomes possible to calculate the probability score even if, for example, a huge number of drugs are counted as the first number and the second number of cases described later depending on the side effect.

[0080]

number

[0081] The "probability score" is a score indicating the possibility that the administered drug for which the conditional probability is calculated is a suspect drug. In this embodiment, an administered drug with a large probability score is more likely to be a suspect drug. In other words, the probability score functions as an index used to estimate a suspect drug. The probability score is an example of an estimation index in the present invention.

[0082] Next, the ranking unit 137 ranks all the administered drugs based on the calculated probability scores (S206). Specifically, the probability scores for each administered drug are compared, and the administered drugs are ranked in descending order of probability score. The rankings are stored in the storage unit 12 in association with the corresponding administered drug information and the corresponding side effect information.

[0083] Next, the probability calculation unit 135 determines whether there is any candidate side effect information whose conditional probability has not yet been calculated (S207).

[0084] If there is candidate side effect information whose conditional probability has not been calculated ("Y" in S207), the probability calculation process (S2) returns to process S203.

[0085] On the other hand, if there is no candidate side effect information for which the conditional probability has not been calculated ("N" in S207), the output unit 138 outputs the administered drug information, probability score, and ranking for each candidate side effect to the display unit 15 (S208). Here, the conditional probability calculated for an administered drug is the probability that the administered drug is a suspect drug. The probability score is an estimated index generated based on the conditional probability. Therefore, the ranking for each administered drug functions as an index (estimation index) of the possibility that the administered drug is a suspect drug. In other words, the ranking for each administered drug is an example of an estimated index.

[0086] Next, the display unit 15 displays the administered drug and the probability score for each candidate side effect in order of rank (S209).

[0087] 11 is a schematic diagram showing another example of information (administered drugs, probability scores) displayed on the display unit 15. The figure shows that, for each candidate side effect, the administered drugs are displayed in descending order of probability score. For ease of explanation, the figure indicates the candidate side effects with the symbols used in the candidate side effect information.

[0088] As described above, the probability score and ranking function as indices (inference indices) used to infer suspect drugs. Therefore, by displaying the administered drugs in order of ranking for each candidate side effect on the display unit 15, the device 1 can provide the user with meaningful information to specifically infer suspect drugs from the candidate side effects (to assist in inferring suspect drugs). For example, if the user is unable to diagnose the target patient, the user can view the display unit 15 to specifically infer suspect drugs based on the candidate side effects extracted in the candidate side effect extraction process (S1). Furthermore, if the user is a physician, the user can view the display unit 15 to specifically infer suspect drugs based on the results of their own diagnosis (side effects). In this way, the user can use the device 1 to infer suspect drugs.

[0089] In this way, the device 1 executes a candidate side effect extraction process (S1) to extract candidate side effects based on the symptoms manifested by the subject patient, and executes a probability calculation process (S2) to generate (calculate) an inferred index (conditional probability, probability score, ranking) based on the extracted (diagnosed) candidate side effects. That is, the device 1 always uses candidate side effects (candidate side effects extracted by the device 1 or candidate side effects diagnosed by a doctor) to generate (calculate) the inferred index. As a result, the device 1 can provide an inferred index with higher accuracy than conventional devices (hereinafter referred to as "conventional devices") that infer suspect drugs from manifested symptoms without using side effects.

[0090] Furthermore, the device 1 can selectively execute the first process and the second process, for example, at the user's choice. As a result, the device 1 can provide an index (estimation index) used to estimate a suspect drug, regardless of the user's ability (whether or not the user can diagnose). That is, when the user is unable to diagnose the target patient, the user can execute the first process using the device 1 to estimate a suspect drug by extracting candidate side effects from the symptoms manifested by the target patient. On the other hand, when the user is able to diagnose the target patient, the user can execute the second process using the device 1 to estimate a suspect drug from the side effects (candidate side effects) that the user diagnosed.

[0091] In the probability calculation process, the display unit of the present invention may display only the administered drugs in order of rank.

[0092] ●Example● ●Verification of probability calculation process Next, we will explain the results of the probability calculation process (S2) when it is executed on actual reported cases of adverse drug reactions and the verification results. In the following example, the number of reported adverse drug reactions is "72," the number of causative drugs is "72," and the number of concomitant drugs is "1,109."

[0093] FIG. 12 is a graph showing the ranking of the conditional probabilities for each of the causative drug and concomitant medication. The figure shows a box plot of the conditional probability rankings for the causative drug and concomitant medications. The values ​​alongside the plots in the figure are medians. As shown in Figure 12, when a Mann-Whitney test was performed on the conditional probability rankings between the two groups of causative drug and concomitant medication, the probability "P" was "P<0.0001," indicating a significant difference between the two groups.

[0094] Furthermore, according to the ROC curve for estimating the causative drug based on the ranking, the ranking of the conditional probability predicted the causative drug with an accuracy of AUC = 0.86 (0.81-0.91).

[0095] Figure 13(a) is a graph showing the number of times that the causative drug was counted in the ranking assigned by this device 1, and (b) is a graph showing the number of times that the concomitant drug was counted in the ranking assigned by this device 1. In the figure, the horizontal axis indicates the ranking of drugs (causative drugs, concomitant drugs), and the vertical axis indicates the number of causative drugs counted in each ranking. As shown in Figure 13(a), the number of causative drugs ranked "1st" by the device 1 was "41," with a probability of "41 / 72 (56.9%)." On the other hand, as shown in Figure 13(b), the number of concomitant drugs ranked "1st" by the device 1 was "31," with a probability of "31 / 1109 (2.8%)." In this way, the device 1 infers causative drugs as suspect drugs with a relatively high probability, and rarely infers concomitant drugs as suspect drugs.

[0096] Summary According to the embodiment described above, the device 1 includes an information acquisition unit 134, a probability calculation unit 135, and a memory unit 12. The information acquisition unit 134 acquires candidate side-effect information and administered drug information. The probability calculation unit 135 calculates, for each of multiple administered drugs, a conditional probability that the administered drug is a suspect drug, based on the candidate side-effect information and administered drug information, under the condition that the candidate side-effect is a manifested side-effect. The memory unit 12 stores the calculated conditional probabilities. With this configuration, regardless of whether the target patient can be diagnosed, the user can obtain a conditional probability for each administered drug, which functions as an estimation index, by inputting the candidate side-effect information and administered drug information into the device 1. In other words, the device 1 supports the estimation of a suspect drug for a patient's manifested side-effect (manifested symptom).

[0097] Furthermore, according to the embodiment described above, the device 1 includes an index generation unit 136 and an output unit 138. The index generation unit 136 generates an estimation index (probability score, ranking) used to estimate a suspect drug based on the conditional probability. The output unit 138 outputs the estimation index. With this configuration, the user can easily estimate a suspect drug from among multiple administered drugs simply by referring to the output estimation index. In other words, the device 1 supports the estimation of a suspect drug for a patient's observed side effect (presented symptom).

[0098] Furthermore, according to the embodiment described above, the storage unit 12 stores DB2. The probability calculation unit 135 calculates the conditional probability based on DB2. DB2 is information indicating the correspondence between multiple pieces of side effect information and one or more pieces of related drug information. With this configuration, the device 1 can calculate the conditional probability with high accuracy based on the medication information of various patients in the past who experienced side effects due to drug treatment.

[0099] Furthermore, according to the embodiment described above, the probability calculation unit 135 calculates the conditional probability based on the first number of cases, the second number of cases, and the number of side effects stored in DB 2. According to this configuration, the present device 1 calculates the conditional probability using variables based on past actual cases, and therefore can calculate the conditional probability with high accuracy.

[0100] Furthermore, according to the embodiment described above, the device 1 includes a symptom acquisition unit 131 and a side effect extraction unit 132. The memory unit 12 stores DB1. The symptom acquisition unit 131 acquires symptom information of the target patient. The side effect extraction unit 132 extracts candidate side effects corresponding to the symptom based on the symptom information and DB1. With this configuration, even if a user is unable to diagnose the target patient, the user can obtain candidate side effects corresponding to the symptom by inputting the symptom of the target patient into the device 1. In other words, with the device 1, even if the user is unable to diagnose the target patient, candidate side effects can be easily obtained by simply inputting the symptom.

[0101] Furthermore, according to the embodiment described above, the device 1 includes a ratio calculation unit 133 and a ranking unit 137. The ratio calculation unit 133 calculates, for each extracted candidate side effect, the ratio (occurrence ratio) of the manifested symptom to all subjective symptoms stored in DB1 as subjective symptoms corresponding to the co-side effect. The ranking unit 137 ranks the candidate side effects based on the calculated occurrence ratio. With this configuration, even if the user does not have sufficient knowledge about side effects corresponding to the subjective symptoms, the user can extract candidate side effects that are likely to be manifested based on the ranking.

[0102] Furthermore, according to the embodiment described above, the probability calculation unit 135 calculates the conditional probability using a Bayesian estimation model stored in the storage unit 12. With this configuration, the present device 1 can calculate the conditional probability with high accuracy by using a Bayesian estimation model that is generated in advance so as to be suitable for calculating the conditional probability.

[0103] In the embodiment described above, the device can selectively execute the first process and the second process. Alternatively, the device may execute only the second process (probability calculation process). In this case, the device does not need to include a symptom acquisition unit, a side effect extraction unit, and a ratio calculation unit.

[0104] Furthermore, DB1 in the present invention may be any information showing the correspondence between side effect information and subjective symptom information, and is not limited to the present embodiment. That is, for example, DB1 in the present invention may be the PMDA's "side effect-subjective symptom correspondence information" itself, or may be another similar database or information generated from the same database.

[0105] Furthermore, DB2 in the present invention is not limited to the present embodiment as long as it is information showing the correspondence between side effect information and related drug information. That is, for example, DB2 in the present invention may be "JADER" itself, or a similar database or information generated from the same database.

[0106] Furthermore, the storage unit of the present invention may temporarily store DB1 and DB2 when the present method is executed. That is, for example, the present device 1 may obtain DB1 and DB2 via a network each time the present method is executed, and temporarily store DB1 and DB2 in the storage unit. In this case, the RAM constituting the control unit of the present invention may function as the storage unit of the present invention.

[0107] Furthermore, DB1 and DB2 in the present invention may be updated periodically or at any timing. In this case, for example, the present device may periodically search an external device to determine whether the information on which DB1 and DB2 are based has been updated, and may automatically update DB1 and DB2. Also, for example, the present device may update DB1 and DB2 based on user operations.

[0108] Furthermore, the output destination of the output unit in the present invention is not limited to a display unit. For example, the output unit may output information to a printer. Also, for example, the output unit may output information to an external device (e.g., a portable information processing terminal) via a communication unit.

[0109] Furthermore, in the above-described embodiment, the device 1 is configured by one computer. Alternatively, the device may be configured by multiple computers. That is, for example, the device may be configured by a group of multiple computers that function as the device. Specifically, for example, the device (group of computers) may be configured by a computer including a storage unit and a computer including a control unit that executes the method. Also, for example, the multiple computers may have the respective functions of a symptom acquisition unit, a side effect extraction unit, a ratio calculation unit, an information acquisition unit, a probability calculation unit, an index generation unit, a ranking unit, and an output unit in a distributed manner. In this case, the multiple computers that make up the group of computers may send and receive information via a network, or may exchange information using a portable storage medium. [Explanation of symbols]

[0110] 1. Device for supporting the estimation of suspected drugs causing side effects 12 Storage section 131 Symptom Acquisition Department 132 Side effect extraction part 133 Percentage Calculation Section 134 Information Acquisition Department 135 Probability Calculation Unit 136 Index generator 137 Ranking Division 138 Output section

Claims

1. A suspect drug estimation support device for side effect that is used to estimate a suspect drug that may be the causative drug causing a side effect that has occurred in a target patient from among a plurality of drugs administered to the target patient, comprising: an information acquisition unit that acquires candidate side effect information indicating candidate side effects that are candidates for the developed side effects and administered drug information indicating the administered drug; a probability calculation unit that calculates, based on the candidate side effect information and the administered drug information, a conditional probability that the administered drug is the suspect drug for each of the plurality of administered drugs, under the condition that the candidate side effect is the observed side effect; a storage unit that stores the calculated conditional probability; and The calculated conditional probability is used to estimate the suspect drug. A support device for estimating suspected drugs causing side effects, characterized by:

2. an index generating unit that generates an index used to estimate the suspect drug based on the calculated conditional probability; an output unit that outputs the index; consisting of The device for supporting the estimation of suspected drugs for side effects according to claim 1.

3. The storage unit Side effect-related drug report information indicating the relationship between side effect information indicating each of a plurality of side effects including the candidate side effect and related drug information indicating related drugs associated with the side effect information; Remember, The probability calculation unit calculates the conditional probability based on the side effect-related drug report information. The device for supporting the estimation of suspected drugs causing side effects according to claim 1 or 2.

4. The related drug information Specific causative drug information indicating a specific causative drug identified as the cause of the side effect; Including, The probability calculation unit a first number of cases in which the administered drug is stored in the side effect-related drug report information as the specific causative drug of the candidate side effect; and a second number of cases in which the administered drug is stored in the adverse reaction-related drug report information as the related drug of the candidate adverse reaction; The number of side effects for which the side effects are stored in the side effect-related drug report information; and Calculating the conditional probability based on The device for supporting the estimation of suspected drugs for side effects according to claim 3.

5. The storage unit side effect-subjective symptom correspondence information indicating a correspondence between side effect information indicating each of the plurality of side effects including the candidate side effect and subjective symptom information indicating subjective symptoms that may occur due to the side effect; Remember, a symptom acquisition unit that acquires symptom information indicating symptoms occurring in the target patient; a side effect extraction unit that extracts the side effect corresponding to the manifested symptom as the candidate side effect based on the manifested symptom information and the side effect-subjective symptom correspondence information; consisting of The device for supporting the estimation of a suspected drug for a side effect according to any one of claims 2 to 4.

6. a proportion calculation unit that calculates, for each of the extracted candidate side effects, a proportion of the manifested symptom to all subjective symptoms stored in the side effect-subjective symptom correspondence information as the subjective symptoms corresponding to the candidate side effect; a ranking unit that ranks the candidate side effects based on the calculated ratio; consisting of The device for supporting the estimation of suspected drugs for side effects according to claim 5.

7. the model used to calculate the conditional probability is a Bayesian estimation model, the storage unit stores the Bayesian estimation model. The device for supporting the estimation of a suspected drug for a side effect according to any one of claims 1 to 6.

8. causing a computer to function as the device for supporting the estimation of a suspected drug for a side effect according to claim 1; A support program for estimating suspected drugs causing side effects.

9. A method for supporting the estimation of a suspected drug for a side effect, which is executed by a suspected drug estimation support device for supporting the estimation of a suspected drug that may be the causative drug causing a side effect exhibited in a patient from among a plurality of drugs administered to the patient, comprising: The side effect suspect drug estimation support device comprises: storage section, Equipped with The side effect suspect drug estimation support device, acquiring candidate side effect information indicating a candidate side effect that is a candidate for the developed side effect and administered drug information indicating the administered drug; calculating, based on the candidate side effect information and the administered drug information, a conditional probability that the administered drug is the causative drug, for each of the plurality of administered drugs, under the condition that the candidate side effect is the observed side effect; storing the calculated conditional probabilities; and The calculated conditional probability is used to estimate the suspect drug. A method for supporting the estimation of suspected drugs causing side effects.

10. the storage unit stores side effect-subjective symptom correspondence information indicating a relationship between information indicating each of a plurality of side effects including the candidate side effect and subjective symptom information indicating subjective symptoms that may occur due to the side effect; The side effect suspect drug estimation support device, acquiring symptom information indicating symptoms currently occurring in the patient; extracting the side effect corresponding to the symptom as the candidate side effect based on the symptom information and the side effect-subjective symptom correspondence information; consisting of The method for supporting estimation of a suspected drug for a side effect according to claim 9.

Images