JP7451278B2 - Medical information processing device, medical information processing method, and medical information processing program - Google Patents

Description

Embodiments disclosed in this specification and the drawings relate to a medical information processing device, a medical information processing method, and a medical information processing program.

In endovascular treatment, in order to reduce the physical and psychological burden on the patient, it is desirable that the treatment be completed with one insertion of the endovascular treatment device. Therefore, in endovascular treatment, it is important to insert an appropriate endovascular treatment device at an appropriate position in the patient. For example, in thrombus retrieval treatment for acute cerebral infarction, it is desirable that a thrombus retrieval device including a catheter and a stent be inserted once to completely retrieve the thrombus (also referred to as 1PASS). A smaller number of PASSes, including 1 PASS, is a factor that improves the prognosis, including life prognosis and functional prognosis, of patients who have undergone thrombus retrieval treatment.

Generally, there are multiple types of endovascular treatment devices, and multiple types of endovascular treatment devices can be used even in the same surgery. However, the type of endovascular treatment device to use is left to the doctor's discretion, and the therapeutic effect on the patient varies depending on the type of endovascular treatment device used. For example, if an inappropriate endovascular treatment device is used, reinsertion of the endovascular treatment device into the patient may occur, increasing the burden on the patient.

Japanese Patent Application Publication No. 2017-217474

One of the problems that the embodiments disclosed in this specification and drawings attempt to solve is to output appropriate endovascular treatment device information for a patient. However, the problems to be solved by the embodiments disclosed in this specification and the drawings are not limited to the above problems. Problems corresponding to the effects of each configuration shown in the embodiments described later can also be positioned as other problems.

The medical information processing device according to the embodiment includes an acquisition section, an extraction section, and an output section. The acquisition unit acquires a first case related to a patient before treatment. The extraction unit extracts similar cases similar to the first case from among the second cases related to the patient after treatment. The output unit outputs device information related to the device used in the similar case in correspondence with the first case according to the treatment results of the similar case.

FIG. 1 is a block diagram showing an example of the configuration of a medical information processing device according to a first embodiment. FIG. 2 is a flow diagram illustrating an example of the operation of the medical information processing device according to the first embodiment. It is a figure showing an example of patient information before treatment and patient information after treatment in endovascular treatment. FIG. 3 is a diagram showing an example of patient information before treatment in thrombus retrieval treatment. FIG. 3 is a diagram showing an example of patient information after treatment in thrombus retrieval treatment. FIG. 2 is a schematic diagram showing an example of the operation of the medical information processing apparatus according to the first embodiment in thrombus collection treatment. FIG. 3 is a diagram showing an example of a device list in thrombus retrieval treatment. FIG. 2 is a block diagram illustrating an example of the configuration of a medical information processing device according to a second embodiment. FIG. 7 is a flow diagram illustrating an example of the operation of the medical information processing device according to the second embodiment. FIG. 7 is a schematic diagram showing an example of the operation of the medical information processing device according to the second embodiment.

Hereinafter, a medical information processing apparatus, a medical information processing method, and a medical information processing program according to embodiments will be described with reference to the drawings. In the following embodiments, parts with the same reference numerals perform similar operations, and redundant explanations will be omitted as appropriate.

(First embodiment)
An example of the configuration of the medical information processing apparatus according to the first embodiment will be described with reference to FIG. 1.
FIG. 1 is a diagram showing a medical information processing system including a medical information processing apparatus according to a first embodiment. The medical information processing system according to the first embodiment includes a medical information processing device 1, a case database 2, an electronic medical record system 3, and a PACS 4.

The medical information processing device 1 includes a processing circuit 11 , an input interface 12 , a display 13 , a memory 14 , and a communication interface 15 . The processing circuit 11, input interface 12, display 13, memory 14, and communication interface 15 are communicably connected to each other via, for example, a bus (BUS). In addition, the medical information processing device 1 is connected to a case database 2, an electronic medical record system 3, and a PACS 4, which are external devices or external systems provided separately from the device, through a network so as to be able to communicate with each other. There is. At this time, a proxy server may be installed between the medical information processing device 1 and the external device or system as a device having a firewall function.

The processing circuit 11 controls the overall operation of the medical information processing apparatus 1 according to the electrical signal of the input operation output from the input interface 12 . For example, the processing circuit 11 includes a processor including a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit) as hardware resources. The processing circuit 11 implements each function by executing a program developed in the memory 14 via a processor. Note that each function is not limited to the case where it is realized by a processing circuit using a single processor. For example, a processing circuit may be configured by combining a plurality of independent processors, and each function may be realized by each processor executing a program.

The acquisition function 111 acquires various information by accessing the case database 2, electronic medical record system 3, and PACS 4 via the communication interface 15. The acquired various information is stored in the memory 14. For example, the acquisition function 111 acquires a first case regarding a patient before treatment.

The extraction function 112 extracts specific information from among the various information stored in the memory 14. For example, the extraction function 112 extracts similar cases similar to the first case from among the second cases related to the patient after treatment.

The calculation function 113 uses various information stored in the memory 14 to perform mathematical operations and calculate calculation results. For example, the calculation function 113 calculates the usage frequency of an endovascular treatment device used in a similar case.

The display control function 114 displays various information stored in the memory 14, case database 2, electronic medical record system 3, and PACS 4 on the display 13. For example, the display control function 114 outputs device information regarding a device used in a similar case in correspondence with the first case according to the treatment results of the similar case.

The input interface 12 receives various inputs from an operator using the medical information processing device 1 , converts the received inputs into electrical signals, and outputs the electrical signals to the processing circuit 11 . As the input interface 12, physical operating devices including, for example, a mouse, a keyboard, a trackball, a switch, a button, a joystick, a touch pad, and a touch panel display can be used. Not limited to these, the input interface 12 may also be a device that receives input from an external input device provided separately from the medical information processing device 1, converts the received input into an electrical signal, and outputs it to the processing circuit 11. Included in the example. Note that as the types of input interface 12, GUI (Graphical User Interface) and CUI (Character User Interface) can be used.

The display 13 receives electrical signals from the processing circuit 11 and displays various information. The display 13 may be any display including, for example, a cathode ray tube (CRT) display, a liquid crystal display (LCD), a plasma display, an organic electro-luminescence display (OELD), and a tablet terminal. is available.

The memory 14 stores various information. The memory 14 includes, for example, a semiconductor memory element including a RAM (Random Access Memory) as a hardware resource. The memory 14 also includes a hard disk, an optical disk (CD, DVD, Blu-ray (registered trademark), Ultra HD Blu-ray (registered trademark)), a flash memory (USB flash memory, memory card, SSD), and a magnetic tape. It may also be a drive device that reads and writes information to and from an external storage medium that includes. Furthermore, the storage area of the memory 14 may be located in the medical information processing device 1 or may be located in an external storage medium.

The communication interface 15 transmits and receives various information to and from the case database 2, electronic medical record system 3, and PACS 4, which are external devices or systems, by wire or wirelessly. Any communication standard can be used between the communication interface 15 and an external device or system; for example, HL7 (Health Level 7) can be used for communication regarding medical text information, and HL7 (Health Level 7) can be used for communication regarding medical image information. DICOM (Digital Imaging and COmmunications in Medicine) can be used for this.

The case database 2 stores patient information after treatment. Post-treatment patient information includes clinical information, device information, and treatment outcome information regarding the patient after treatment. In the case database 2, clinical information, device information, and treatment outcome information for each of a plurality of patients after treatment are compiled and managed as cases. Here, a case related to a patient after treatment (second case) is also referred to as a post-treatment case. That is, one post-treatment case includes clinical information, device information, and treatment outcome information regarding one patient who underwent endovascular treatment. Further, the entity of the case database 2 is, for example, a server, and patient information after treatment is stored in the memory of the server. Note that the term "patient after treatment" includes, for example, a patient who has undergone endovascular treatment (also referred to as a treated patient). Note that the same patient may have multiple cases after treatment.

The electronic medical record system 3 stores patient information before treatment. Pre-treatment patient information includes clinical information about the patient before treatment. In the electronic medical record system 3, clinical information about each of a plurality of pre-treatment patients is compiled and managed as a case. Here, the case regarding the patient before treatment (first case) is also referred to as the case before treatment. That is, one pre-treatment case includes clinical information regarding one patient undergoing endovascular treatment. Further, the electronic medical record system 3 is actually a system consisting of a plurality of devices including, for example, a server and a display, and patient information before treatment is stored in the memory of the server. Note that the patient before treatment includes, for example, a patient undergoing endovascular treatment (also referred to as a treatment target patient). Note that there may be multiple cases before treatment for the same patient.

PACS4 stores medical images related to patients before treatment and medical images related to patients after treatment. Medical images include, for example, diagnostic images and intraoperative images. In this embodiment, the medical images related to the patient before treatment include diagnostic images, and the medical images related to the patient after treatment include diagnostic images and intraoperative images. In addition, in the PACS 4, medical images related to a patient before treatment are stored in association with a pre-treatment case related to the patient before treatment stored in the electronic medical record system 3, and medical images related to a patient after treatment are stored as a case. It is stored in association with the post-treatment case related to the patient after the treatment stored in the database 2. Furthermore, medical images related to a patient before treatment are included in patient information before treatment, and medical images related to a patient after treatment are included in patient information after treatment. Further, the entity of the PACS 4 is, for example, a system consisting of a plurality of devices including a server and a display, and medical images related to the patient before treatment and medical images related to the patient after treatment are stored in the memory of the server. Note that PACS (Picture Archiving and Communication Systems) is also called a picture archiving and communication system.

Note that the case database 2, electronic medical record system 3, and PACS 4 may be existing ones that have already been constructed. For example, the case database 2 may be a case database operated as a national project such as NCD (National Clinical Database), a case database specific to a medical facility where the medical information processing device 1 is installed, or a case database of multiple medical facilities. It may also be a database. Further, the case database 2 may be a unique case database constructed by collecting arbitrary plural cases from various existing case databases. The unique case database includes, for example, cases stored in the medical facility where the medical information processing device 1 is installed, cases included in performance tests to test the effectiveness of the device, and cases included in published trials. May be added.

An example of the operation of the medical information processing apparatus according to the first embodiment will be described with reference to FIG. 2.
In step S201, the acquisition function 111 acquires patient information before treatment. The acquisition function 111 acquires patient information before treatment from the electronic medical record system 3 and PACS 4, for example. In this embodiment, it is assumed that the acquired pre-treatment patient information includes a plurality of pre-treatment cases.

In step S202, the acquisition function 111 acquires patient information after treatment. The acquisition function 111 acquires patient information after treatment from the case database 2 and PACS 4, for example. In this embodiment, it is assumed that the acquired post-treatment patient information includes a plurality of post-treatment cases.

In step S203, the extraction function 112 extracts similar cases. For example, the extraction function 112 selects one of a plurality of pre-treatment cases included in the obtained pre-treatment patient information, and extracts the clinical information included in the selected pre-treatment case and the obtained treatment. A degree of similarity is calculated between clinical information included in each of a plurality of post-treatment cases included in the subsequent patient information. Note that the extraction function 112 may select cases before treatment according to an input from the operator or a predetermined order stored in the memory 14 in advance. Similarity calculations may use similarity indices used in general natural language processing, including, for example, cosine similarity, JACCARD coefficients, DICE coefficients, and SIMPSON coefficients. That is, the similarity may be calculated using any method. Next, the extraction function 112 creates a similar case group by extracting a plurality of post-treatment cases whose degree of similarity is equal to or higher than a threshold value.

In addition, if the clinical information included in the patient information before treatment and the clinical information included in the patient information after treatment are not similar and have excess or deficiency, the degree of similarity will be determined by the clinical information included in the patient information before treatment. It may be calculated using information existing in both the clinical information included in the patient information after treatment. For example, if the clinical information of patient information before treatment includes the disease name and time of onset, and the clinical information of patient information after treatment includes the name of the disease but not the time of onset, then the common information exists. It suffices if the degree of similarity is calculated for disease names.

Furthermore, by specifying specific information among post-treatment patient information by the operator using the medical information processing device 1, post-treatment cases that do not have the specific information can be excluded from the similarity calculation target. It's okay. For example, if the operator specifies a specific device that exists in the medical facility where the medical information processing device 1 is installed, cases after treatment in which the specific device was not used are excluded from the similarity calculation target. You may. As a result, the medical information processing apparatus 1 can display only the specific devices existing in the medical facility to which the operator belongs in step S206.

In step S204, the extraction function 112 extracts cases with good treatment results. The extraction function 112 creates a good treatment outcome group, for example, by extracting a plurality of post-treatment cases whose treatment outcome is equal to or higher than a threshold value from among a plurality of post-treatment cases included in the similar case group.

In addition, when a plurality of indicators exist as treatment results, weighting may be set for each of the plurality of treatment result indicators. At this time, a good treatment outcome group may be created based on the treatment outcome index with a higher set weight. For example, if there are items such as ``presence of recurrence'' and ``number of PASS'' as indicators of treatment results, the operator can set the item to be considered as a good treatment result if there is no recurrence even if the number of PASS is 3. "Presence or absence of recurrence" may be given greater weight than "PASS number". Alternatively, a good treatment outcome group may be created based on treatment outcomes calculated based on weights set for each of a plurality of treatment outcome indicators.

Furthermore, a group with good treatment results may be created based on a threshold value set for the treatment results. For example, the operator may set the threshold value of the number of PASS to 2 so that a case where the number of PASS, which is an index of treatment results, is 2 or less is regarded as a good treatment result.

In step S205, the calculation function 113 calculates the usage frequency of the device. The calculation function 113 calculates the usage frequency of each device for the plurality of post-treatment cases included in the good treatment outcome group, for example, based on the type of device used in each of the plurality of post-treatment cases. The frequency of use of a device is calculated, for example, using the number of cases after treatment extracted as a group with good treatment results as the denominator, and using the number of cases in which each device was used as the numerator. Here, the frequency of use of the device is not limited to the frequency of use of one device for multiple post-treatment cases as described above. For example, if multiple devices of different types are used in at least one post-treatment case, the frequency of device use with respect to the set of multiple devices for the multiple post-treatment cases. may also be calculated.

In step S206, the display control function 114 displays device information in order of frequency of use. The display control function 114 displays the device types and intraoperative images on the display 13, for example, in the order of calculated frequency of use.

Incidentally, accompanying device information includes accompanying information including the frequency of use of the device calculated in step S205, the treatment results calculated in step S204, and clinical information for which the degree of similarity calculated in step S203 is equal to or higher than the threshold value. may also be displayed.

Further, as long as the post-treatment patient information is acquired from the case database 2 and PACS 4 at the first execution, the process of step S202 does not need to be executed from the second time onwards. Furthermore, if the extraction function 112 is configured to directly access the case database 2 and PACS 4 to extract similar cases in step S203, the process in step S202 may not be executed.

Note that the device information display in step S206 is performed for one of the multiple pre-treatment cases selected in step S203, but the same operation is repeated for other pre-treatment cases. This makes it possible to display device information for all pre-treatment cases in order of frequency of use.

Note that the order of the above steps can be changed as appropriate as long as the device information is finally displayed in order of frequency of use. For example, after the patient information after treatment is acquired first, the patient information before treatment may be acquired (switching S201 and S202). Alternatively, after first extracting a group with good treatment results, a group of similar cases may be extracted (replacing S203 and S204).

An example of patient information before treatment and patient information after treatment in endovascular treatment will be described with reference to FIG. 3.
FIG. 3 is a patient information correspondence table 300 showing a breakdown of patient information before treatment and patient information after treatment in endovascular treatment. According to the patient information correspondence table 300, patient information before treatment includes clinical information, and patient information after treatment includes clinical information, device information, and treatment outcome information.

Clinical information indicates medical information about the patient before or after treatment. The clinical information includes, for example, disease name, onset time, patient's age and sex, location and characteristics of the affected area, diagnostic images, findings from the diagnostic images, and pretreatment. “Disease name” indicates the name of a disease requiring endovascular treatment. Disease names include, for example, acute cerebral infarction, myocardial infarction, cerebral aneurysm, subarachnoid hemorrhage, cerebral arteriovenous malformation, and dural arteriovenous fistula. In addition, for acute cerebral infarction, there are atherothrombotic cerebral infarction and cardiogenic cerebral infarction, and even for the same disease, disease names may be used depending on the disease type. "Onset time" indicates the time when the disease developed. If the exact onset time is unknown, the onset time may be the last time the patient was confirmed to be in a normal state. “Patient age and sex” indicates the age and sex of the patient with the disease. "Location and properties of affected area" indicates the location and properties of the affected area of a patient with a disease. The location and properties of the affected area include, for example, the location of the thrombus and the properties of the blood vessel. "Diagnostic image" indicates a medical image used for diagnosing a disease. Diagnostic images include, for example, endoscopic images and radiographic images of a patient's affected area in the presence or absence of a contrast agent. Specifically, radiological images include X-ray CT (Computed Tomography) images, MR (Magnetic Resonance) images, UL (ULtrasonic) images, and RI (Radio Isotope) images. including. "Findings from diagnostic images" indicates characteristics determined from diagnostic images. "Pretreatment" refers to treatment performed prior to endovascular treatment. Pretreatment includes, for example, administration of t-PA (tissue-plasminogen activator), which is a type of thrombolytic drug.

Note that clinical information is not limited to the above items. Clinical information includes, for example, the patient's allergies, past history, circumstances at the time of onset, oral medications, blood pressure, pulse, respiratory status, body temperature, severity of symptoms, and the severity of symptoms obtained through interviews and tests before endovascular treatment. Contains a range. The severity of symptoms includes, for example, NIHSS (National Institutes of Health Stroke Scale) for cerebral infarction. The range of affected areas includes, for example, ASPECTS (Albetra Stroke Program Early CT Score) for cerebral infarction.

The device information indicates information regarding the device used to treat the patient after treatment. The device information includes, for example, the type of device and an intraoperative image. “Device type” indicates the type of endovascular treatment device used for endovascular treatment. Device types include, for example, thrombus aspiration catheters, ablation catheters, thrombus retrieval stents, flow-occlusion balloons, flow-modifying stents, and aneurysm embolization coils. Device types specifically include Merci Retriever®, Penumbra System®, and Solitaire FR®. "Intraoperative image" indicates a medical image when an endovascular treatment device is being used on an affected area of a patient. Intraoperative images include, for example, endoscopic images and radiographic images taken of the affected area of the patient in the presence or absence of a contrast agent at the time of device insertion.

The treatment outcome information indicates the treatment outcome for the patient after treatment. The treatment outcome information includes, for example, treatment time, evaluation criteria, functional prognosis, and presence or absence of recurrence. "Treatment time" indicates the time when endovascular treatment was completed. Furthermore, as an alternative to the treatment time, the elapsed time from the onset time to the treatment time may be used. "Evaluation criteria" refers to the evaluation criteria used to evaluate the therapeutic results of endovascular treatment. For example, evaluation criteria for thrombus retrieval treatment include TICI (Thrombolysis In Cerebral Infarction), PASS count, and MRS (Modified Rankin Scale). "Functional prognosis" refers to the patient's condition after a certain period of time after endovascular treatment. The functional prognosis of thrombus retrieval therapy includes, for example, the presence or absence of sequelae including motor disorders, language disorders, and cognitive disorders. "Presence or absence of recurrence" indicates whether or not a disease similar to that before the treatment developed after a certain period of time after the endovascular treatment.

An example of patient information before treatment in thrombus retrieval treatment will be described with reference to FIG. 4.
FIG. 4 is a pre-treatment patient information table 400 showing a breakdown of pre-treatment patient information in thrombus retrieval treatment. According to the pre-treatment patient information table 400, the pre-treatment patient information in the thrombus retrieval treatment includes the disease name, onset time, age, sex, thrombus location, blood vessel properties, and pretreatment as clinical information. In the pre-treatment patient information table 400, clinical information for each of a plurality of pre-treatment patients is compiled and managed as a pre-treatment case. Furthermore, each case before treatment is managed by being assigned a case number. For example, the case number is represented by (alphabet A) + (serial number) for cases before treatment. Note that the pre-treatment patient information table 400 is an example of a mode of storing pre-treatment patient information, and is not limited thereto. As described above, pre-treatment patient information may be stored in a format other than a tabular format as long as the clinical information is stored in association with each pre-treatment case.

For example, the first line of the entry in the patient information table 400 before treatment includes "case number" A1, "disease name" atherothrombotic cerebral infarction, "onset time" 16:00, "age" 55, and "gender" male. , "thrombus location" in the frontal lobe of the right brain, "vascular properties" mild stenosis, and "pretreatment" t-PA administration are stored in correspondence. Similarly, for subsequent entries, clinical information is stored in association with each pre-treatment case.

An example of patient information after treatment in thrombus retrieval treatment will be described with reference to FIG. 5.
FIG. 5 is a post-treatment patient information table 500 showing a breakdown of post-treatment patient information in thrombus retrieval treatment. According to the post-treatment patient information table 500, post-treatment patient information in thrombus retrieval treatment includes disease name, onset time, age, sex, thrombus location, blood vessel properties, and pretreatment as clinical information, and device information as device information. The treatment outcome information includes treatment time, TICI, functional prognosis, and presence or absence of recurrence. In the post-treatment patient information table 500, clinical information, device information, and treatment outcome information for each of a plurality of post-treatment patients are compiled and managed as post-treatment cases. Further, each case after treatment is managed by being assigned a case number. For example, the case number is represented by (alphabet B) + (sequential number) for cases after treatment. Note that the post-treatment patient information table 500 is an example of a mode of storing post-treatment patient information, and is not limited thereto. As mentioned above, if clinical information, device information, and treatment outcome information are stored in association with each case after treatment, patient information after treatment can be stored in a format other than tabular format. Good too.

For example, the first line of the entry in the post-treatment patient information table 500 includes "case number" B1, "disease name" atherothrombotic cerebral infarction, "time of onset" 12:00, "age" 60, and "gender" male. , "Location of thrombus" in the frontal lobe of the right brain, "Vascular properties" mild stenosis, "Pretreatment" t-PA administration, "Type of device" catheter X, "Treatment time" 13:00, "TICI" 3, "Functional prognosis" good , and “presence or absence of recurrence” are stored in association with each other. Similarly, for subsequent entries, clinical information, device information, and treatment result information are stored in association with each other for each case after each treatment.

An example of the operation of the medical information processing apparatus according to the first embodiment in thrombus retrieval treatment will be described with reference to FIG. 6.
Hereinafter, a pre-treatment patient information table 400 shown in FIG. 4 is stored in the electronic medical record system 3 as data including pre-treatment patient information, and a post-treatment patient information table 400 shown in FIG. 5 is stored as data including post-treatment patient information. 500 is stored in case database 2. It is assumed that each patient information table in FIGS. 4 and 5 is created in a predetermined data format, such as text format (TXT), tab-separated file format (TSV), and comma-separated file format (CSV). It is created in a text file format containing.

First, the acquisition function 111 acquires patient information before treatment (corresponding to S201). Specifically, the acquisition function 111 acquires the data of the pre-treatment patient information table 400 in FIG. 4 from the electronic medical record system 3 and stores it in the memory 14.

Next, the acquisition function 111 acquires patient information after treatment (corresponding to S202). Specifically, the acquisition function 111 acquires the data of the post-treatment patient information table 500 in FIG. 5 from the case database 2 and stores it in the memory 14.

Next, the extraction function 112 extracts similar cases (corresponding to S203). Specifically, the extraction function 112 selects "A1", which is the case with the youngest case number, from among the plurality of pre-treatment cases included in the pre-treatment patient information table 400 in FIG. 4 . Next, the clinical information included in the pre-treatment case "A1" and each of the multiple post-treatment cases "B1", "B2", "B3", etc. included in the post-treatment patient information table 500 in FIG. Calculate the degree of similarity between clinical information contained in For example, the degree of similarity (J(AP, BP)) between the clinical information of a case before treatment (AP) and the clinical information of a case after treatment (BP) is calculated by equation (1) using the JACCARD coefficient. be done.

J(AP,BP)=|AP∩BP|/|AP∪BP|...(1)
Note that |AP∩BP| refers to the items that are common to each item included in the clinical information (AP) of the case before treatment and the clinical information (BP) of the case after treatment. Represents the number of items. In addition, |AP∪BP| is calculated from the sum of the number of each item included in the clinical information (AP) of the case before treatment and the number of each item included in the clinical information (BP) of the case after treatment. , represents the number obtained by subtracting the number of items common to both.

Here, the clinical information (AP1) included in case "A1" before treatment is "atherothrombotic cerebral infarction", "16:00", "55", "male", "right brain frontal lobe", "mild stenosis". ”, and “t-PA administration”, and the clinical information (BP1) included in case “B1” after treatment is “atherothrombotic cerebral infarction”, “12:00”, “60”, “male” , “right brain frontal lobe,” “mild stenosis,” and “none.” Therefore, the degree of similarity (J(AP1, BP1)) between the clinical information (AP1) of case "A1" before treatment and the clinical information (BP1) of case "B1" after treatment is calculated based on equation (1). It is calculated as follows.

J (AP1, BP1) = 4/10 = 0.40

Similarly, the degree of similarity (J(AP1, BP2), J( AP1, BP3)) are calculated as follows based on equation (1). Note that, in reality, the degree of similarity with the clinical information of the pre-treatment case "A1" may be calculated for the clinical information of each post-treatment case after the treatment "B3".

J (AP1, BP2) = 3/11≒0.27
J (AP1, BP3) = 1/13≒0.077

Subsequently, the extraction function 112 creates a similar case group by extracting cases after treatment whose degree of similarity is equal to or greater than a threshold value. Specifically, when the similarity threshold is set to 0.25, the extraction function 112 extracts post-treatment cases "B1" and "B2" whose similarity is 0.25 or more and forms a group of similar cases. Create. Conversely, the post-treatment case "B3" for which the degree of similarity is less than 0.25 is excluded.

Next, the extraction function 112 extracts cases with good treatment results (corresponding to S204). Specifically, the extraction function 112 extracts post-treatment cases with a "TICI" of 2b or more and no "presence or absence of recurrence" among post-treatment cases "B1" and "B2" included in the similar case group. to create a group with good treatment results. Here, the treatment outcome information included in case "B1" after treatment includes "TICI" 3 and "presence or absence of recurrence", and the treatment outcome information included in case "B2" after treatment is "TICI" 3. 2b, including “no recurrence”. Therefore, the extraction function 112 extracts "B1" and "B2" as post-treatment cases that satisfy the above conditions, and creates a group with good treatment results.

Next, the calculation function 113 calculates the usage frequency of the device (corresponding to S205). Generally, the frequency of use of the device used in each post-treatment case included in the good treatment outcome group is calculated as follows based on equation (2).

(Frequency of device use) = (Number of cases in which the device was used) / (Number of cases after treatment included in the good treatment outcome group)... (2)

Based on formula (2), the frequency of use of "catheter X" used in case "B1" after treatment and the frequency of use of "catheter Y" used in case "B2" after treatment are as follows: It is calculated as follows.

(Frequency of use of catheter X) = 1/2 = 0.50 = 50%
(Frequency of use of catheter Y) = 1/2 = 0.50 = 50%

Finally, the display control function 114 displays the device information in order of frequency of use (corresponding to S206). Specifically, the display control function 114 displays the devices as a device list on the display 13 in the calculated order of frequency of use. An example of the device list for thrombus retrieval treatment created by the above operation is shown in FIG. In fact, by repeating the same operation as above for each of the other acquired cases "A2" and "A3" before treatment, the device list corresponding to case "A2" before treatment is created. , a device list corresponding to the pre-treatment case "A3", and so on, each device list corresponding to each pre-treatment case is created.

An example of a device list for thrombus retrieval treatment will be described with reference to FIG. 7.
FIG. 7 is a device list 700 showing the details of the device list for thrombus retrieval treatment. In the device list 700, the frequency of use is determined based on the device information and treatment result information included in each of a plurality of similar post-treatment cases "B1" and "B2" for the pre-treatment case "A1". Device types are displayed in ascending order. Various information included in the device list 700 includes a case number, a device number, a type of device, and a frequency of use of the device. The case number indicates the case number assigned to the case before treatment. Further, each type of device is managed by assigning a device number to each type. The device number is represented by, for example, (alphabet D) + (sequential number).

For example, in the first row of the entry in the device list 700, case number "A1", device number "D1", "device type" catheter X, and "device usage frequency" 50% are stored in correspondence. . Furthermore, in the second row of the entry in the device list 700, the case number "A1", the device number "D2", the "device type" catheter Y, and the "device usage frequency" 50% are stored in correspondence. . If the devices are used with the same frequency, the devices may be displayed in any order, such as alphabetically. Similarly, for subsequent entries, the case number, device number, device type, and device usage frequency are associated with each other and displayed as a list in the device list 700.

Other modifications of the medical information processing device 1 according to the first embodiment will be described below.

(Warning display)
In step S206, the medical information processing apparatus 1 displays device information in descending order of device usage frequency, but the invention is not limited to this. For example, the medical information processing apparatus 1 can display devices that require careful use by displaying device information in descending order of device usage frequency.

In order to realize the above-mentioned warning display, in step S204, the extraction function 112 extracts a plurality of cases with poor treatment results instead of cases with good treatment results. Next, in step S205, the usage frequency of each device is calculated based on the device information included in the case with poor treatment results. Thereafter, the display control function 114 may display each device as a device that requires caution in use. At this time, the device may be highlighted, or a pop-up window may be displayed to call attention to the device.

(Additional information display)
Furthermore, if metal allergy is included in the clinical information regarding the patient before treatment, in step S203, the operator specifies a device that does not contain a metal allergen, thereby identifying cases after treatment in which a device containing metal was used. It may be excluded from the similarity calculation targets. Alternatively, information regarding metals contained in the device may be displayed as accompanying information of the device information. Further, a user's manual for each device, a link for accessing the home page of the device vendor, and an order button for ordering the device may be displayed as accompanying information of the device information.

(Real-time update of device information)
The operation of the medical information processing device 1 described above is such that device information including the type of device recommended for use on a patient before endovascular treatment is displayed to a doctor performing endovascular treatment. However, it is not limited to this. For example, it is also possible to update the displayed device information in real time in accordance with intraoperative images of a patient before treatment that are sequentially displayed during an endovascular treatment operation.

For example, during surgery for thrombus retrieval treatment, there may be cases where a thrombus cannot be efficiently retrieved even if a particular device is used. At this time, the key image is an intraoperative image of a patient before treatment who is using the specific device, and there is an intraoperative image of a patient after treatment that is similar to the intraoperative image, and the patient after treatment is using another device. can be used to extract cases with a good prognosis after treatment. The degree of similarity between an intraoperative image of a patient before treatment and an intraoperative image of a patient after treatment that is similar to the intraoperative image may be calculated by any method including histogram comparison, feature point matching, and template matching. . Thereafter, the other device used in the treatment that resulted in a good prognosis for the corresponding case may be displayed as an alternative device.

According to the first embodiment described above, by displaying to the doctor endovascular treatment devices that have statistically calculated high treatment results and are frequently used, it is possible to help the doctor decide which device to use. . In turn, the prognosis of patients treated with the device is improved. Conversely, by displaying endovascular treatment devices that have poor treatment results and are used infrequently to the doctor, it is possible to help the doctor determine which devices should be used with caution.

(Second embodiment)
An example of the configuration of a medical information processing apparatus according to the second embodiment will be described with reference to FIG. 8.
FIG. 8 is a diagram showing a medical information processing system including a medical information processing apparatus according to the second embodiment. The medical information processing system according to the second embodiment includes a medical information processing device 1, a case database 2, an electronic medical record system 3, and a PACS 4. The medical information processing device 1 according to the second embodiment includes a learning function 115 in the processing circuit 11. It should be noted that the configuration of the medical information processing system according to the second embodiment other than the learning function 115 is the same as the configuration of the medical information processing system according to the first embodiment, unless otherwise explained, so the explanation will be omitted. .

The learning function 115 generates a learned model by providing a large number of learning samples including input data and correct data to the pre-learning model. Note that the pre-learning model and the trained model are stored in the memory 14 as a program.

The pre-learning model and the trained model are, for example, composite functions that include a program as a network structure and parameters as adjustable numerical values. The network structure of the pre-learning model and the trained model may be, for example, a DNN (Deep Neural Network), which is a multilayer network including an input layer, a middle layer, and an output layer. The network structure is not limited to this, and may be other structures such as a support vector machine, random forest, decision tree, and logistic regression. In this embodiment, the model before being adjusted through learning is referred to as a pre-learning model, and the model after being adjusted through learning is referred to as a trained model.

An example of the operation of the medical information processing apparatus according to the second embodiment will be described with reference to FIGS. 9 and 10. In this embodiment, when pre-treatment patient information is input, the purpose of this embodiment is to generate a trained model for displaying an appropriate device for the pre-treatment patient included in the pre-treatment patient information. The pre-learning model generated under the objective is called the pre-learning device display model, and the trained model is called the trained device display model.
In step S901, the learning function 115 generates a learned model. The learning function 115 performs learning by, for example, providing a large number of learning samples including processed pre-treatment patient information as input data and a device list as correct data to the pre-learning device display model stored in the memory 14. Generate a device display model. Here, the processed pre-treatment patient information includes pre-treatment cases related to pre-treatment patients for which output of the corresponding device list has been completed. Specifically, one learning sample is a set of one pre-treatment case included in the processed pre-treatment patient information and one device list corresponding to the case. That is, a set of a pre-treatment case acquired by the medical image processing apparatus according to the first embodiment as input data and a device list outputted by the apparatus as correct data may be used as the learning sample. In this way, the input data and output data of the medical image processing apparatus according to the first embodiment are respectively converted into the input data and correct answer data of the pre-learning device display model stored in the medical image processing apparatus according to the second embodiment. You can use it as Note that the device list serving as correct data is each device list created for each case before treatment, and may include the device list 700 in FIG. 7 .

Specifically, the learning function 115 performs forward propagation processing by applying the pre-learning device display model to pre-treatment patient information serving as input data, and outputs a provisional device list serving as provisional correct data. Next, backpropagation processing is performed on the difference (also referred to as error) between the provisional device list and the device list serving as correct data, and a gradient vector is calculated. Next, the parameters of the pre-learning device display model are adjusted based on the gradient vector. As described above, a trained device display model is generated by updating parameters by repeating forward propagation processing and back propagation processing for a large number of learning samples including input data and ground truth data.

In step S902, the acquisition function 111 inputs patient information before treatment to the trained model. The acquisition function 111 inputs, for example, pre-treatment patient information to be processed from the electronic medical record system 3 as input data to the learned device display model stored in the memory 14. Here, the pre-treatment patient information to be processed includes pre-treatment cases related to pre-treatment patients for which output of the corresponding device list has not been completed.

In step S903, the display control function 114 displays device information in order of frequency of use. The display control function 114 displays, for example, a device list (also referred to as an estimated device list) output from the learned device display model on the display 13 as output data. That is, the estimated device list is a device list generated by inputting patient information before treatment into a learned device display model.

As described above, the processing result of the medical information processing apparatus according to the first embodiment may be used as the device list that serves as correct data during learning. In addition, the patient information before treatment, which is input data, is the one used in the medical information processing device according to the first embodiment during learning, and the patient information used in the medical information processing device according to the first embodiment during application is different from that during learning. Pre-treatment patient information to be processed, which is not used by the medical information processing device, may be used.

According to the second embodiment described above, it is possible to cause the trained model to perform the same operation as in the first embodiment. Furthermore, the more a set of a pre-treatment case and a device list for the pre-treatment case is given to the trained model as a learning sample, the more accurate the trained model will be. More appropriate device information can be output. Furthermore, even in an offline situation, appropriate device information can be displayed without accessing the case database. Furthermore, by transferring the generated trained model to another device as a program, it is possible to cause the device to which the program is transferred to perform similar operations.

The medical information processing apparatus according to the first embodiment and the medical information processing apparatus according to the second embodiment have been described above, but the embodiment may be a program that implements the above operations. That is, after another system or device stores the program in its own storage area, the processing circuit may read and execute the program.

According to at least one embodiment described above, appropriate endovascular treatment device information for a patient can be output.

Although several embodiments have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, changes, and combinations of embodiments can be made without departing from the gist of the invention. These embodiments and their modifications are included within the scope and gist of the invention as well as within the scope of the invention described in the claims and its equivalents.

1 Medical information processing device 2 Case database 3 Electronic medical record system 4 PACS
11 Processing circuit 12 Input interface 13 Display 14 Memory 15 Communication interface 111 Acquisition function 112 Extraction function 113 Calculation function 114 Display control function 115 Learning function 300 Patient information correspondence table 400 Patient information table before treatment 500 Patient information table after treatment 700 Device list

Claims (11)

an acquisition unit that acquires a first case related to a patient before treatment;
Among at least one second case related to the patient after treatment, at least one second case whose similarity with the first case is a threshold value or more and whose treatment result is a threshold value or more is extracted. an extraction section;
a calculation unit that calculates the frequency of use of each device based on the type of each device used in the extracted second case;
an output unit that outputs device information regarding each device to the first case according to the calculated frequency of use of each device ;
A medical information processing device comprising: The output unit displays the device information in descending order of the calculated frequency of use of each device.
The medical information processing device according to claim 1 . The output unit displays the calculated devices in descending order of usage frequency, to alert the user.
The medical information processing device according to claim 1 or claim 2 . the first case includes clinical information about the patient before treatment;
The second case includes clinical information, device information, and treatment outcome information regarding the patient after the treatment,
The medical information processing device according to any one of claims 1 to 3 . The device information includes at least one of the type of device and an intraoperative image when the device is used.
The medical information processing device according to claim 4 . The treatment outcome information includes at least one of treatment time, evaluation criteria, functional prognosis, and the presence or absence of recurrence.
The medical information processing device according to claim 4 or claim 5 . further comprising a learning unit that generates a trained model in which the first case is used as input data and the device information is trained as correct data;
The medical information processing device according to any one of claims 1 to 6 . a model application unit that inputs a third case related to a patient to be processed into the learned model and outputs device information output from the learned model to the third case; to have ,
The medical information processing device according to claim 7 . The calculation unit calculates the frequency of use of each device by dividing the number of second cases in which each device was used by the number of second cases extracted.
The medical information processing device according to any one of claims 1 to 8. obtaining a first case for a pre-treatment patient;
Among at least one second case related to the patient after treatment, at least one second case whose similarity with the first case is equal to or higher than a threshold and whose treatment result is equal to or higher than the threshold is extracted. ,
Calculating the frequency of use of each device based on the type of each device used in the extracted second case,
outputting device information regarding each device to the first case according to the calculated frequency of use of each device ;
Medical information processing method. to the computer,
an acquisition function for acquiring a first case related to a patient before treatment;
Among the at least one second case related to the patient after treatment, at least one second case whose similarity with the first case is equal to or higher than a threshold value and whose treatment result is equal to or higher than the threshold value is extracted. Extraction function and
a calculation function that calculates the frequency of use of each device based on the type of each device used in the extracted second case;
an output function that outputs device information regarding each device to the first case according to the calculated frequency of use of each device ;
A medical information processing program that realizes.