Information processing device and computer program

Abstract

To improve the ability to maintain appropriate inventory levels of pharmaceuticals in medical institutions. [Solution] The server device 12 is capable of communicating with a client device 11 associated with a medical institution. The communication interface 121 receives patient information PT from the client device 11, which originates from the prescription of medicines for the patient. The processor 122 uses a machine learning-trained inference model 123 to output a predicted return visit time when the patient will revisit the medical institution to receive the medicines in response to the input of patient information PT, and obtains predicted values ​​of predetermined parameters based on the predicted return visit time. If the patient information PT satisfies predetermined exception conditions associated with the risk of the predicted return visit time, the processor 122 causes the client device 11 to provide an indicator that draws attention to information corresponding to at least one of the predicted return visit time and the predicted value obtained based on the patient information PT.

Description

Technical Field

[0001] The present disclosure relates to an information processing apparatus capable of communicating with a user interface apparatus associated with a medical institution. The present disclosure also relates to a computer program executable by a processor installed in the information processing apparatus.

Background Art

[0002] Patent Document 1 discloses an apparatus for predicting a time series of the number of patients who visit a medical institution to receive a specific medicine dispensed, based on prescription information corresponding to a plurality of prescriptions accepted at the medical institution, in order to maintain an appropriate inventory amount of medicines at the medical institution. For the prediction, an inference model generated through machine learning is used.

Prior Art Literature

Patent Literature

[0003]

Patent Document 1

Summary of the Invention

Problem to be Solved by the Invention

[0004] There is a demand for improving the maintainability of appropriate inventory amounts of pharmaceuticals in medical institutions.

Means for Solving the Problem

[0005] A first exemplary embodiment that can be provided by the present disclosure is an information processing apparatus capable of communicating with a user interface apparatus associated with a medical institution, an interface that receives patient information originating from the prescription of a pharmaceutical to a patient from the user interface apparatus; A processor that, in response to the input of patient information, uses a machine learning-trained inference model to output a predicted return visit time when the patient will revisit the medical institution to receive the prescribed medication, and obtains predicted values ​​for predetermined parameters based on the predicted return visit time. It is equipped with, The processor causes the user interface device to provide an indicator to draw attention to information corresponding to at least one of the predicted return visit time and the predicted value obtained based on the patient information, if the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time.

[0006] A second example of the embodiments that this disclosure may provide is a computer program executable by a processor installed in an information processing device capable of communicating with a user interface device associated with a medical institution, By being executed, the information processing device will Patient information originating from the prescription of medications to patients is received from the user interface device. Using a machine learning-trained inference model that outputs a predicted return visit time for the patient to receive the prescribed medication in response to the input of the patient information, predictive values ​​of predetermined parameters based on the predicted return visit time are obtained. If the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time, the user interface device is instructed to provide an indicator to draw attention to the information corresponding to at least one of the predicted return visit time and the predicted value obtained based on the patient information.

[0007] According to the configurations of the first and second embodiments, patient information originating from the prescription of medicines to patients can be input into an inference model from a user interface device associated with a medical institution, thereby obtaining a predicted time when the patient will revisit the medical institution to receive the medicines. In addition, predicted values ​​for predetermined parameters can also be obtained based on this predicted time. On the other hand, under certain conditions, relying on the inference results of the inference model may carry risks in medicine inventory management. According to the configurations of the first and second embodiments, if the patient information satisfies the exception conditions associated with such risks, a warning indicator is provided to draw attention to the information corresponding to at least one of the predicted revisit time and predicted value obtained based on the patient information. This allows the user of the user interface device to take appropriate measures rather than blindly accepting the inference results of the inference model. Therefore, it is possible to improve the ability to maintain appropriate levels of medicine inventory in medical institutions.

[0008] A third example of the embodiments that this disclosure may provide is an information processing device capable of communicating with a user interface device associated with a medical institution, An interface that receives patient information originating from the prescription of medications to patients from the user interface device, A processor that, in response to the input of patient information, uses a machine learning-trained inference model to output a predicted return visit time when the patient will revisit the medical institution to receive the prescribed medication, and obtains predicted values ​​for predetermined parameters based on the predicted return visit time. It is equipped with, The processor causes the user interface device to provide information corresponding to at least one of the corrected predicted return visit time and the predicted value, by determining whether the patient will return to the medical institution, if the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time.

[0009] A fourth example of the embodiments that this disclosure may provide is a computer program executable by a processor installed in an information processing device capable of communicating with a user interface device associated with a medical institution, By being executed, the information processing device will Patient information originating from the prescription of medications to patients is received from the user interface device. Using a machine learning-trained inference model that outputs a predicted return visit time for the patient to receive the prescribed medication in response to the input of the patient information, predictive values ​​of predetermined parameters based on the predicted return visit time are obtained. The processor causes the user interface device to provide information corresponding to at least one of the corrected predicted return visit time and the predicted value, by determining whether the patient will return to the medical institution, if the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time.

[0010] According to the configurations of the third and fourth embodiments, patient information originating from the prescription of medicines to patients can be input into an inference model from a user interface device associated with a medical institution, thereby obtaining a predicted time when the patient will revisit the medical institution to receive the medicines. In addition, predicted values ​​for predetermined parameters can also be obtained based on this predicted time. On the other hand, under certain conditions, relying on the inference results (probability of revisit) from the inference model may carry risks in medicine inventory management. According to the configurations of the third and fourth embodiments, if the patient information satisfies the exceptional conditions associated with such risks, the user of the user interface device can be provided with information corresponding to at least one of the predicted revisit time and predicted value, which has been modified to reduce the risk by determining whether or not a revisit will occur according to the exceptional conditions. Therefore, it is possible to improve the ability to maintain an appropriate inventory level of medicines in medical institutions. [Brief explanation of the drawing]

[0011] [Figure 1]The configuration of an inventory management system according to an example embodiment is illustrated. [Figure 2] An example functional configuration of the inventory management system of FIG. 1 is shown. [Figure 3] An example flow of processing executed by the server device of FIG. 1 is shown. [Figure 4] Another example flow of processing executed by the server device of FIG. 1 is shown. [Figure 5] An example of information provided in the client device of FIG. 1 is shown. [Figure 6] Another example of information provided in the client device of FIG. 1 is shown. [Figure 7] Another example of information provided in the client device of FIG. 1 is shown. [Figure 8] Another example of information provided in the client device of FIG. 1 is shown. [Figure 9] Another example flow of processing executed by the server device of FIG. 1 is shown. [Figure 10] Another example of information provided in the client device of FIG. 1 is shown. [Figure 11] Another example of information provided in the client device of FIG. 1 is shown. DETAILED DESCRIPTION OF EMBODIMENTS

[0012] Example embodiments will be described in detail below with reference to the accompanying drawings. In the respective drawings referred to in the following description, as necessary the scale is changed to make each illustrated element recognizable in size.

[0013] As used in this disclosure, the term "medical institution" refers to an institution where medical professionals provide patients with information about pharmaceuticals. Examples of "medical institutions" include hospitals, clinics, medical offices, pharmacies, drugstores, nursing homes, care homes, and elderly care facilities. Examples of "medical professionals" include doctors, nurses, pharmacists, care managers, and registered dietitians.

[0014] Figure 1 illustrates the configuration of an inventory management system 10 according to one embodiment. The inventory management system 10 includes a client device 11 and a server device 12.

[0015] Client device 11 is the device on which information is provided. Client device 11 is used in a medical institution. Therefore, there may be multiple client devices 11. Furthermore, the same client device used by multiple users with different accounts may be considered multiple client devices. Client device 11 may be a stationary device installed in a specific location, or it may be a portable device that can be carried by the user.

[0016] The server device 12 is installed at a location separate from the client device 11. The client device 11 and the server device 12 are configured to communicate data bidirectionally via the communication network 20.

[0017] As illustrated in Figure 2, the client device 11 is equipped with a communication interface 111. On the other hand, the server device 12 is equipped with a communication interface 121. Both the communication interface 111 and the communication interface 121 are hardware interfaces configured to enable bidirectional data communication between the client device 11 and the server device 12.

[0018] The client device 11 comprises a processor 112 and a user interface 113. The user interface 113 is a hardware interface that mediates the exchange of information between the user and the processor 112 and receives user instructions to cause the client device 11 to perform predetermined operations. In this embodiment, the user interface 113 includes a display. The client device 11 is an example of a user interface device.

[0019] The processor 112 and the user interface 113 may be mounted in a common chassis, or they may be provided in a manner distributed across separate chassis.

[0020] A predetermined operation performed by the client device 11 includes the transmission of patient information PT to the server device 12. The patient information PT is transmitted from the communication interface 111 and received by the communication interface 121.

[0021] Patient information PT includes information originating from the prescription of medications to patients. Examples of information that may be included in patient information PT include the following items. However, not all of the listed items need to be included in patient information PT. • Patient identification information • Patient age, sex, and measurable biometric parameters • Patient's medical history, medication history, and usage of over-the-counter (OTC) medications and health supplements. • Number of patient visits to medical institutions that own client device 11 • Prescription identification information • Name, ingredients, therapeutic classification, dosage form, administration method, duration of use, and efficacy of the prescribed medication. • Date the prescription was issued • Identification information of the prescribing physician • Identification information of the medical institution to which the prescribing physician belongs • Results of dispensing based on prescriptions • Date the medication was dispensed • Identification information of the healthcare professional who dispensed the medication • Identification information of the medical institution to which the healthcare professional who dispensed the medication belongs. • Medication history information

[0022] The server device 12 includes a processor 122. The processor 122 is configured to store patient information PT in a storage device (not shown) in a state that can be referenced later. The storage device can be a memory device that can be implemented using semiconductor memory, a hard disk drive, a magnetic tape drive, etc. The patient information PT stored in the storage device forms part of the patient information set. The server device 12 is an example of an information processing device.

[0023] The server device 12 includes an inference model 123. The processor 122 is configured to use the inference model 123 to predict when a patient identified by patient information PT will revisit a medical institution that uses the client device 11 to receive medication.

[0024] The inference model 123 is a machine learning-trained computer algorithm that, in response to input of patient information PT (which identifies patients who have visited a medical institution), outputs the predicted timing of when a patient will revisit that medical institution. The machine learning method may be rule-based or deep learning-based.

[0025] The inference results from inference model 123 are output as the probability that the patient will revisit the medical institution on a specific day included in the predicted revisit period. If this probability is above a predetermined threshold, it is considered that a revisit will occur. If the probability is below the threshold, it is considered that no revisit will occur. This threshold may be constant or variable.

[0026] By performing inference using inference model 123 for at least one patient, a predicted number of patients who will visit the medical institution on a specific day to receive a specific medication (predicted number of returning patients) can be obtained. The number of returning patients is an example of a predetermined parameter. The predicted number of returning patients obtained based on the predicted return visit timing is an example of a predicted value.

[0027] The processor 122 is configured to transmit the first prediction information PD1, which includes the acquired predicted return visit time and the predicted number of patients returning, to the client device 11 via the communication interface 121.

[0028] The processor 112 of the client device 11 is configured to display information on the user interface 113's display that corresponds to at least one of the acquired predicted return visit time and predicted number of patients returning, based on the first predicted information PD1 received via the communication interface 111. In addition to or instead of this, the information may be provided through projection by a projector or printing by a printer.

[0029] In healthcare facilities using the client device 11, inventory management of pharmaceuticals is performed based on information corresponding to at least one of the provided predicted return visit timing and predicted number of returning patients. For example, if a large quantity of a particular pharmaceutical is predicted to be dispensed on a specific day, an order for the necessary or sufficient quantity of that pharmaceutical may be placed prior to that specific day.

[0030] Figure 3 shows an example of the processing flow executed by the processor 122 of the server device 12 configured as described above.

[0031] First, the processor 122 determines whether the patient information PT transmitted from the client device 11 has been received by the communication interface 121 (STEP 11). This process is repeated until it is determined that the patient information PT has been received (NO in STEP 11).

[0032] If it is determined that patient information PT has been received (YES in STEP 11), the processor 122 inputs the patient information PT into the inference model 123 and obtains the predicted return visit time and the predicted number of patients to return (STEP 12).

[0033] Next, the processor 122 refers to the patient information PT and determines whether the predetermined exception conditions are met (STEP 13). The exception conditions define the conditions associated with the risk of the inference result by the inference model 123. Details of the exception conditions will be described later.

[0034] If it is determined that the exception conditions are not met (NO in STEP 13), the processor 122 transmits first prediction information PD1, which includes at least one of the acquired predicted return visit time and predicted number of return patients, to the client device 11 via the communication interface 121 (STEP 14). The first prediction information PD1 is received by the communication interface 111 of the client device 11.

[0035] The processor 112 of the client device 11 provides information corresponding to the first prediction information PD1 through the user interface 113.

[0036] On the other hand, if it is determined that the exception condition is satisfied (YES in STEP 13), the processor 122 generates warning sign information corresponding to a warning sign that draws attention to the information based on the predicted revisit timing output from the inference model 123 (STEP 15).

[0037] The warning sign is configured to have an appropriate form that allows it to identify information based on the predicted return visit time, which is obtained based on patient information PT that satisfies the exception conditions, from the information provided through the user interface 113 of the client device 11. Specifically, the warning sign may be composed of at least one of the following: letters, symbols, figures, colors, patterns, etc.

[0038] In this case, the processor 122 transmits the first prediction information PD1 and the warning sign information from the communication interface 121 to the client device 11 (STEP 16). The first prediction information PD1 and the warning sign information are received by the communication interface 111 of the client device 11.

[0039] The processor 112 of the client device 11 provides information corresponding to the first prediction information PD1 and warning sign information through the user interface 113. That is, while information based on the predicted return visit time obtained based on patient information PT that is judged to satisfy the exceptional conditions is identified, information corresponding to the first prediction information PD1 is provided through the user interface 113.

[0040] Figure 4 illustrates the flow of processing performed by processor 122 to determine whether the exception condition is met in STEP 13.

[0041] First, the processor 122 determines whether the number of patients who received medication identified by patient information PT at the medical institution using the client device 11 that transmitted patient information PT is below a threshold (STEP 21). This determination can be made by referring to the patient information set stored in storage.

[0042] The period for counting patient numbers can be determined as appropriate. Examples of such periods include specific days, weeks, months, seasons, or years. The medical institutions whose patient numbers are counted may include other medical institutions belonging to the same group as the medical institution using the client device 11. This group may be defined by corporate unit, regional unit, etc. The threshold may be changed depending on the medical institution, day of the week, time of year, season, etc.

[0043] If the number of patients is determined to be below the threshold (YES in STEP 21), the processor 122 determines whether the unit price of the drug identified by the patient information PT is below the threshold (STEP 22). This threshold may be changed depending on the medical institution.

[0044] If the unit price of the medication identified by the patient information PT is determined to be above the threshold (NO in STEP 22), processor 122 determines that the exception condition has been met and concludes that the patient will not revisit the medical institution (STEP 23). In other words, the probability of the patient revisiting is forcibly set to 0%.

[0045] If the unit price of the drug identified by the patient information PT is determined to be below a threshold (YES in STEP 22), the processor 122 determines whether the patient identified by the patient information PT has a chronic disease (STEP 24). For example, a determination that the patient has a chronic disease may be made in the following cases. • If the medication identified by the patient information PT is prescribed to a patient with a chronic disease. • When a chronic disease is suspected from a combination of multiple medications included in the patient information PT. • When a chronic disease is identified based on the medical history and medication history information included in the patient information PT.

[0046] If the patient identified by the patient information PT is determined not to have a chronic disease (NO in STEP 24), the processor 122 determines that the exception condition has been met and concludes that the patient will not revisit the medical institution (STEP 23). In other words, the probability of the patient revisiting is forcibly set to 0%.

[0047] If the patient is determined to have a chronic disease (YES in STEP 24), processor 122 determines that the exception condition has been met and concludes that the patient will revisit the medical institution (STEP 25). In other words, the probability of the patient revisiting is forcibly set to 100%.

[0048] In other words, processor 122 determines that the exception condition has been satisfied if any of the following situations are met. Note that either condition 1 or condition 2 may be omitted. Condition 1: At least one medical institution using the client device 11 has a number of patients who have received prescriptions for medications identified by patient information PT that are below a threshold, and the unit price of the medications identified by patient information PT is above a threshold. Condition 2: At least one medical institution using the client device 11 has a number of patients who have received prescriptions for medications identified by patient information PT that is below a threshold, and the patients identified by patient information PT are associated with chronic diseases.

[0049] If the number of patients who have received a particular medication at a specific medical institution is small, the inference accuracy of the inference model 123 may be insufficient due to insufficient samples. In such circumstances, if the unit price of the particular medication is above a threshold, and the predicted number of returning patients exceeds the actual number of visiting patients, the relatively expensive medication will remain in inventory, creating a risk of worsening the financial condition of the medical institution. Here, by forcibly setting the probability of returning to 0% at the predicted return visit time for patients related to patient information PT, the possibility of avoiding such a risk can be increased.

[0050] When a patient with a chronic disease is associated with a patient information PT, it becomes relatively easy to determine whether or not a return visit is necessary. If the patient has a chronic disease, the probability of a return visit is high, and if they do not, the probability of a return visit is low. Therefore, by forcing the return visit probability at the predicted return visit time to 100% in the former case and forcing it to 0% in the latter case, it is possible to increase the possibility of avoiding the risks that arise from relying on inaccurate inference results.

[0051] If it is relatively easy to determine whether or not a patient will return for a follow-up visit, then alternative conditions may be adopted in addition to, or instead of, the presence or absence of chronic disease in STEP 24.

[0052] For example, if the presence or absence of a return visit is determined based on a conversation between a healthcare professional and a patient, and information indicating this fact is included in the patient information PT, that information may be referenced. If a return visit is clear, the probability of a return visit at the predicted return visit time is forcibly set to 100%, and if it is clear that there will be no return visit, the probability of a return visit at the predicted return visit time is forcibly set to 0%.

[0053] As an alternative example, if the frequency of past visits by a patient identified by patient information PT exceeds a threshold, the probability of a return visit at the predicted return visit time may be forcibly set to 100%. As an alternative example, if the interval between past visits by a patient identified by patient information PT exceeds a threshold, the probability of a return visit at the predicted return visit time may be forcibly set to 0%.

[0054] On the other hand, if the number of patients who have received prescriptions for the medication identified by the patient information PT at a medical institution using the client device 11 that transmitted the patient information PT is above a threshold (NO in STEP 21), the processor 122 determines whether the interval between visits to the medical institution by the patient identified by the patient information PT is irregular (STEP 26). For example, in the following cases, it may be determined that the visit interval is irregular. • When a drug identified by patient information PT is associated with a disease in which the revisit rate within a specified period is below a threshold. • When an acute disease is suspected from a combination of multiple medications included in the patient information PT. • If the patient information PT contains information entered by a healthcare professional indicating that continued medication is not required.

[0055] If it is determined that the visit intervals are not irregular (NO in STEP 26), the processor 122 determines that the exception condition is not met (STEP 27). That is, if the number of patients who have been dispensed the prescribed medication at a particular medical institution is relatively large, and it is determined that the intervals between visits of those patients to that medical institution are not irregular, the processor 122 decides to rely on the inference results from the inference model 123.

[0056] If the visit interval is determined to be irregular (YES in STEP 26), the processor 122 executes special processing (STEP 28). Specifically, it executes a process to provide information to the user interface 113 of the client device 11 prompting it to place a purchase order so that the inventory level does not fall below a threshold. The inventory level threshold may be changed as appropriate depending on the type of drug, season, time of year, etc.

[0057] In other words, even if the visit intervals of a particular patient are irregular, if the number of patients visiting the medical institution is relatively large, then, regardless of the inference results of inference model 123, the priority is to avoid situations where dispensing is impossible due to insufficient stock.

[0058] Figure 5 shows an example of the information provided by the user interface 113 of the client device 11 as a result of the processing shown in Figures 3 and 4.

[0059] In this example, the time series TS1 of the daily predicted return visit patient count, aggregated based on the predicted return visit timing inferred by the inference model 123 of the server device 12, is shown. Among the multiple predicted return visit patient counts included in the time series TS1, the plots showing the predicted return visit patient counts for "February 13th" and "February 14th" are marked with a cautionary label CT. The cautionary label CT indicates that the patient information PT that contributed to the calculation of these predicted return visit patient counts includes cases that satisfy the aforementioned exception information, and draws attention to the risks associated with the inference results by the inference model 123.

[0060] Figure 6 shows another example of information provided by the user interface 113. In this example, the predicted return visit dates estimated for each patient based on patient information PT are displayed in calendar format. For example, on "February 13th," return visits are predicted for "Patient A," "Patient B," and "Patient C." On "February 14th," a warning sign CT is formed by differentiating the display of "Patient Y" from the other patients. In this example, attention is drawn to the predicted return visit date inferred for "Patient Y." That is, it is suggested that the predicted return visit date of "Patient Y" contributes to the reason why attention was deemed necessary for the number of predicted return visits on "February 14th" in Figure 5.

[0061] Figure 7 shows another example of the information provided by the user interface 113. In this example, a cautionary marker CT corresponding to the result of the exception condition judgment process, as explained with reference to Figure 4, is displayed for each patient.

[0062] This example shows that inference model 123 inferred that "Patient Y" would revisit on "February 14th" to receive prescriptions for "ABC" and "KHS". On the other hand, it is shown that the exception conditions regarding the likelihood of a revisit and the unit price of the medication were satisfied based on the information contained in "Patient Y's" patient information PT. Specifically, a cautionary marker CT is shown indicating that there is a high probability that the revisit will occur on "February 13th" rather than "February 14th", and a cautionary marker CT is shown indicating that "KHS" is an expensive medication.

[0063] Users of client device 11 who have access to this information can take appropriate action. For example, they can take measures such as ordering the purchase of "KHS" in a timely manner in preparation for the possibility that "Patient Y" may revisit on "February 13th". The information provided through user interface 113 may also be modified as appropriate.

[0064] As described above, according to the configuration of this embodiment, by inputting patient information originating from the prescription of medicines to patients from the client device 11 associated with the medical institution into the inference model 123, it is possible to obtain a predicted time when the patient will revisit the medical institution to receive the medicines. In addition, the predicted number of patients who will revisit the medical institution based on the predicted time can also be obtained. On the other hand, under certain conditions, relying on the inference results of the inference model 123 may involve risks in medicine inventory management. According to the configuration of this embodiment, if the patient information satisfies the exception conditions associated with such risks, a warning sign CT is provided to draw attention to the information corresponding to at least one of the predicted revisit time and the predicted number of revisiting patients obtained based on the patient information. Therefore, the user of the client device 11 can take appropriate measures instead of blindly accepting the inference results of the inference model 123. Thus, it is possible to improve the ability to maintain an appropriate amount of medicine inventory at the medical institution.

[0065] Furthermore, if, as a result of the judgment process related to the exceptional conditions explained with reference to Figure 4, it is determined whether or not the patient will revisit the medical institution (STEP 23 or STEP 25), at least one of the predicted revisit time obtained using the inference model 123 and the predicted number of patients who will revisit, calculated based on that predicted revisit time, may be modified.

[0066] As illustrated in Figure 8, instead of providing information corresponding to the predicted number of returning patients based on the inference results of the inference model 123, the user interface 113 may provide information corresponding to the corrected predicted number of returning patients as described above. In this example, the time series TS2 of the predicted number of returning patients with the correction reflected is shown. Specifically, in the example explained with reference to Figures 5 to 7, the correction is made after it is determined that "Patient Y" will return on "February 13th" instead of "February 14th". As a result, as can be seen from the comparison with the time series TS1 illustrated in Figure 5, the predicted number of returning patients on "February 13th" has increased, and the predicted number of returning patients on "February 14th" has decreased.

[0067] Although not shown in the diagrams, the above modifications may be reflected in the information corresponding to the predicted return visit period in the example information described with reference to Figures 6 and 7.

[0068] Figure 9 shows another example of the processing flow performed by the processor 122 of the server device 12 to enable such information provision. Processing elements that are substantially the same as those described with reference to Figure 3 are given the same reference numerals, and redundant explanations are omitted.

[0069] In this example, if it is determined that the exception condition is satisfied as a result of the process illustrated in Figure 4 (YES in STEP 13), the processor 122 modifies at least one of the predicted return visit timing and the predicted number of returning patients based on the inference results of the inference model 123 according to the exception condition (STEP 31).

[0070] Next, the processor 122 transmits the second prediction information PD2, which corresponds to the corrected information, to the client device 11 via the communication interface 121 (STEP 32). The second prediction information PD2 is received by the communication interface 111 of the client device 11.

[0071] The processor 112 of the client device 11 provides information corresponding to the second prediction information PD2 through the user interface 113. Depending on the specifications of the information provided through the user interface 113, either the revised predicted return visit time or the predicted number of returning patients may not be included in the second prediction information PD2.

[0072] As mentioned above, under certain conditions, relying on the inference results (probability of return visit) from the inference model 123 may carry risks in terms of drug inventory management. According to the configuration of this processing example, if the patient information satisfies the exception conditions associated with such risks, the user of the client device 11 can be provided with information corresponding to at least one of the predicted return visit time and the predicted number of returning patients, which has been modified to reduce the risk by determining whether or not a return visit will occur according to the exception conditions. Therefore, it is possible to improve the ability to maintain appropriate drug inventory levels in medical institutions.

[0073] The process for generating warning sign information (STEP 15), as explained with reference to Figure 3, can also be applied to the processing example in Figure 9. In this case, as illustrated in Figure 9, a warning sign CT may be added to information that has been modified in at least one of the predicted return visit time and the predicted number of returning patients.

[0074] This configuration provides the user of the client device 11 with an opportunity to verify the appropriateness of information that has been automatically corrected based on exceptional conditions. If, as a result of the verification, it is determined that the correction is inappropriate, the user can make the necessary corrections through the user interface 113.

[0075] The patient information PT contains information related to the quantity of medication prescribed to the patient. Therefore, based on the predicted return visit timing of the patient inferred by the inference model 123, the demand for the medication at that predicted return visit time can be predicted. The demand for the medication is an example of a predetermined parameter. The predicted demand (predicted demand) is an example of a predicted value.

[0076] Figure 10 shows another example of information provided in the user interface 113 of the client device 11. In this example, the time series TS3 of the predicted demand for the drug "KHS" is shown. In this case, information more directly related to drug inventory management at medical institutions can be provided to the user of the client device 11.

[0077] As mentioned above, the inference results from inference model 123 are output as the probability that a patient with patient information PT will revisit a medical institution at a specific time, and therefore inherently involve a distribution. In each of the configuration examples described so far, representative values ​​of this distribution (mean, median, mode, etc.) are used to obtain predicted revisit timing, predicted number of revisiting patients, predicted demand, etc.

[0078] However, as illustrated in Figure 10, a distribution indicator DS showing the distribution of predicted demand, which can be obtained by reflecting the distribution of predicted return visit timings, may be provided in the user interface 113 of the client device 11. Examples of distributions include 95% confidence intervals and standard deviations. The distribution indicator DS is an example of information corresponding to the distribution. This explanation can also be applied to the predicted number of returning patients, as explained with reference to Figures 5 and 8.

[0079] Figure 11 shows another example of information provided in the user interface 113 of the client device 11. In this example, the return visit date for "Patient Y," who will receive the drug "KHS," is predicted to be "February 14th." In this example, a distribution indicator DS is provided that shows the distribution of predicted return visit dates. In the calendar-format display configuration, the distribution indicator DS, which extends horizontally, indicates the intervals before and after the predicted date in which "Patient Y" is likely to return. If the day of the week on which the patient visits is fixed, the distribution indicator DS may also extend vertically.

[0080] With the configuration described above, the degree of freedom in placing purchase orders based on the inference results of the inference model 123 can be increased. For example, depending on the healthcare institution's policy regarding inventory levels, an appropriate value within the range indicated by the distribution indicator DS can be adopted. In the case of a healthcare institution with a policy of maintaining a sufficient inventory level, a purchase order may be placed based on the upper value of the distribution indicator DS in Figure 10. Similarly, a purchase order may be placed based on an earlier date in the distribution indicator DS in Figure 11.

[0081] Each of the processors 112 of the client device 11 and 122 of the server device 12, which have the various functions described above, can be realized by at least one general-purpose microprocessor operating in cooperation with at least one general-purpose memory. Examples of general-purpose microprocessors include CPUs, MPUs, and GPUs. Examples of general-purpose memory include ROM and RAM. In this case, the ROM may store a computer program that performs the above-described processing. ROM is an example of a non-temporary computer-readable medium in which a computer program is stored. The general-purpose microprocessor selects at least a portion of the program stored in the ROM and loads it onto the RAM, and then performs the above-described processing in cooperation with the RAM. The computer program may be pre-installed in the general-purpose memory, or it may be downloaded from an external server device via a communication network and then installed in the general-purpose memory. In this case, the external server device is an example of a non-temporary computer-readable medium in which a computer program is stored.

[0082] Each of processors 112 and 122 may be implemented by at least one dedicated integrated circuit capable of executing the above-described computer program. Examples of dedicated integrated circuits include microcontrollers, ASICs, FPGAs, etc. In this case, the above-described computer program is pre-installed in a memory element included in the dedicated integrated circuit. This memory element is an example of a computer-readable medium in which the computer program is stored. Each of processors 112 and 122 can also be implemented by a combination of a general-purpose microprocessor and a dedicated integrated circuit.

[0083] The configurations described herein are merely examples to facilitate understanding of this disclosure. Each configuration example may be modified or combined with other configuration examples as appropriate, as long as it does not deviate from the intent of this disclosure.

[0084] In the above embodiment, the inference model 123 is implemented in the server device 12. However, the inference model 123 may be provided by another service provider that can communicate with the server device 12 via the communication network 20. In this case, the processor 122 of the server device 12 inputs patient information PT to the inference model 123 through the communication interface 121 and obtains information corresponding to the inference result through the communication interface 121.

[0085] The various processes described as being performed by the processor 122 of the server device 12 may also be performed by the processor 112 of the client device 11. That is, information can be exchanged between the client device 11 and other service-providing devices as described above. In this case, the processor 112 inputs patient information PT to the inference model 123 through the communication interface 111 and obtains information corresponding to the inference result through the communication interface 111. Information based on the inference result is provided through the user interface 113. In this example, the processor 112 is an example of an information processing device, and the user interface 113 is an example of a user interface device.

[0086] The configurations listed below also constitute part of this disclosure. Item 1: An information processing device capable of communicating with a user interface device associated with a medical institution, An interface that receives patient information originating from the prescription of medications to patients from the user interface device, A processor that, in response to the input of patient information, uses a machine learning-trained inference model to output a predicted return visit time when the patient will revisit the medical institution to receive the prescribed medication, and obtains predicted values ​​for predetermined parameters based on the predicted return visit time. It is equipped with, The processor causes the user interface device to provide an indicator to draw attention to information corresponding to at least one of the predicted return visit time and the predicted value obtained based on the patient information, if the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time. Information processing device. Item 2: An information processing device capable of communicating with a user interface device associated with a medical institution, An interface that receives patient information originating from the prescription of medications to patients from the user interface device, A processor that, in response to the input of patient information, uses a machine learning-trained inference model to output a predicted return visit time when the patient will revisit the medical institution to receive the prescribed medication, and obtains predicted values ​​for predetermined parameters based on the predicted return visit time. It is equipped with, The processor causes the user interface device to provide information corresponding to at least one of the corrected predicted return visit time and the predicted value, by determining whether the patient has returned to the medical institution, if the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time. Information processing device. Item 3: The processor causes the user interface device to provide an indicator that draws attention to information corresponding to at least one of the corrected predicted revisit time and the predicted value. The information processing device described in item 2. Item 4: The aforementioned exception condition is that the number of patients who received the drug at the medical institution is less than the threshold, and the unit price of the drug is determined to be equal to or greater than the threshold. An information processing device as described in any one of items 1 to 3. Item 5: The aforementioned exception conditions are that the number of patients who have received the drug at the medical institution is below a threshold, and that the patients are associated with a chronic disease. An information processing device as described in any one of items 1 through 4. Item 6: The processor, when it is determined that the number of patients who have received the drug at the medical institution is above a threshold and that the timing of the patients' return visits to the medical institution is irregular, causes the user interface device to provide information prompting it to place a purchase order so that the inventory of the drug does not fall below a threshold. An information processing device as described in any one of items 1 through 5. Item 7: The processor causes the user interface device to provide information corresponding to the distribution of at least one of the predicted revisit time and the predicted value. An information processing device as described in any one of items 1 through 6. Item 8: The predicted return visit timing is determined by the predetermined parameters, which are at least one of the number of patients visiting the medical institution and the demand for the pharmaceutical product. An information processing device as described in any one of items 1 through 7. [Explanation of Symbols]

[0087] 11: Client device, 111: Communication interface, 112: Processor, 113: User interface, 12: Server device, 121: Communication interface, 122: Processor, 123: Inference model, CT: Warning sign, DS: Distribution sign, PT: Patient information

Claims

1. An information processing device capable of communicating with a user interface device associated with a medical institution, An interface that receives patient information originating from the prescription of medications to patients from the user interface device, A processor that, in response to the input of patient information, uses a machine learning-trained inference model to output a predicted return visit time when the patient will revisit the medical institution to receive the prescribed medication, and obtains predicted values ​​for predetermined parameters based on the predicted return visit time. It is equipped with, The processor causes the user interface device to provide an indicator to draw attention to information corresponding to at least one of the predicted return visit time and the predicted value obtained based on the patient information, if the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time. Information processing device.

2. An information processing device capable of communicating with a user interface device associated with a medical institution, An interface that receives patient information originating from the prescription of medications to patients from the user interface device, A processor that, in response to the input of patient information, uses a machine learning-trained inference model to output a predicted return visit time when the patient will revisit the medical institution to receive the prescribed medication, and obtains predicted values ​​for predetermined parameters based on the predicted return visit time. It is equipped with, The processor causes the user interface device to provide information corresponding to at least one of the corrected predicted return visit time and the predicted value, by determining whether the patient has returned to the medical institution, if the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time. Information processing device.

3. The processor causes the user interface device to provide an indicator that draws attention to information corresponding to at least one of the corrected predicted revisit time and the predicted value. The information processing apparatus according to claim 2.

4. The aforementioned exception condition is that the number of patients who received the drug at the medical institution is less than the threshold, and the unit price of the drug is determined to be equal to or greater than the threshold. The information processing apparatus according to claim 1 or 2.

5. The aforementioned exception conditions are that the number of patients who have received the drug at the medical institution is below a threshold, and that the patients are associated with a chronic disease. The information processing apparatus according to claim 1 or 2.

6. The processor, when it is determined that the number of patients who have received the drug at the medical institution is above a threshold and that the timing of the patients' return visits to the medical institution is irregular, causes the user interface device to provide information prompting it to place a purchase order so that the inventory of the drug does not fall below a threshold. The information processing apparatus according to claim 1 or 2.

7. The processor causes the user interface device to provide information corresponding to the distribution of at least one of the predicted revisit time and the predicted value. The information processing apparatus according to claim 1 or 2.

8. The predicted return visit timing is determined by the predetermined parameters, which are at least one of the number of patients visiting the medical institution and the demand for the pharmaceutical product. The information processing apparatus according to claim 1 or 2.

9. A computer program executable by a processor installed in an information processing device capable of communicating with a user interface device associated with a medical institution, By being executed, the information processing device will Patient information originating from the prescription of medications to patients is received from the user interface device. Using a machine learning-trained inference model that outputs a predicted return visit time for the patient to receive the prescribed medication in response to the input of the patient information, predictive values ​​of predetermined parameters based on the predicted return visit time are obtained. If the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time, the user interface device is instructed to provide an indicator to draw attention to the information corresponding to at least one of the predicted return visit time and the predicted value obtained based on the patient information. Computer program.

10. A computer program executable by a processor installed in an information processing device capable of communicating with a user interface device associated with a medical institution, By being executed, the information processing device will Patient information originating from the prescription of medications to patients is received from the user interface device. Using a machine learning-trained inference model that outputs a predicted return visit time for the patient to receive the prescribed medication in response to the input of the patient information, predictive values ​​of predetermined parameters based on the predicted return visit time are obtained. The processor causes the user interface device to provide information corresponding to at least one of the corrected predicted return visit time and the predicted value, by determining whether the patient has returned to the medical institution, if the patient information satisfies predetermined exception conditions associated with the risk of the predicted return visit time. Computer program.

Images