Quality process management system, program, and quality process management method

Abstract

A quality process management system, program, and quality process management method are provided that can efficiently set quality processes and manage progress for various event information. [Solution] The quality process management system 1 comprises a storage processing unit 102 that stores quality control information and past quality control activity records in a database 3, an event reception unit 101 that receives event information 2 that triggers a quality process, an action generation unit 104 that analyzes the event information 2 received by the event reception unit 101 using a trained model 4 that analyzes the event information 2 based on the information in the database 3, and generates actions required in the quality process caused by the event information 2, a setting unit 105 that sets a workflow for executing the action, and a monitoring processing unit 106 that monitors the progress of the set workflow and displays the progress.

Description

[Technical Field]

[0001] The present invention relates to a quality process management system, a program, and a quality process management method. [Background technology]

[0002] Conventionally, systems relating to document management and business management have been known. For example, Patent Document 1 describes a system for transferring clinical trial management files between pharmaceutical companies and medical institutions. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-95026 Summary of the Invention [Problem to be solved by the invention]

[0004] In QMS (Quality Management System), when an event such as a product complaint occurs, product quality and safety are maintained by implementing various quality processes such as complaint management, CAPA (Corrective Action and Preventive Action), change management, and education management. However, various events occur in the course of daily business activities, and it can become difficult to set up and manage appropriate quality processes because companies are so busy dealing with each and every event.

[0005] An object of the present invention is to provide a quality process management system, program, and quality process management method that can efficiently set quality processes and manage their progress for various types of event information. [Means for solving the problem]

[0006] The quality process management system includes a storage processing unit that stores quality control information and past activity records related to quality control in a database; an event reception unit that receives event information that triggers a quality process related to quality control; an action generation unit that analyzes the event information received by the event reception unit using a trained model that analyzes the event information based on information in the database and generates actions required in the quality process caused by the event information; a setting unit that sets a workflow for executing the action; and a monitoring processing unit that monitors the progress of the set workflow and displays the progress. [Effects of the Invention]

[0007] According to the present invention, it is possible to efficiently set quality processes and manage the progress of various types of event information. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a schematic diagram illustrating a quality process control system according to an embodiment of the present invention. [Figure 2] FIG. 2 is a schematic diagram showing modularized functions provided in a quality control system. [Figure 3] FIG. 1 is a schematic diagram showing types of quality processes. [Figure 4] FIG. 2 is a block diagram showing the hardware configuration of the quality process control system. [Figure 5] FIG. 2 is a block diagram showing the functional configuration of a quality process control system. [Figure 6] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays complaint information that triggers a quality process for complaint management. [Figure 7] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays similar event information and candidates for the root cause of the event information. [Figure 8] 10 is a schematic diagram showing an example of a display screen that displays a summary of the analysis results of complaint information as event information, corrective measures, and preventive measures. FIG. [Figure 9A] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays a workflow of a quality process set by a quality process management system. [Figure 9B] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays a workflow of a quality process set by a quality process management system. [Figure 9C] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays a workflow of a quality process set by a quality process management system. [Figure 10] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays a list of workflows for a plurality of quality processes and their progress statuses. [Figure 11] FIG. 10 is a schematic diagram illustrating an example of a selection screen for quality control documents for generating educational content. [Figure 12] FIG. 10 is a schematic diagram showing an example of an input screen for specific information for generating educational content. [Figure 13] FIG. 10 is a schematic diagram showing an example of a display screen for confirming the details of generated educational content. [Figure 14] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays an analysis result of the respondent's understanding of a test question. [Figure 15] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays additional learning content for a respondent of a test question. [Figure 16] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays a list of education and training and the progress status thereof. [Figure 17] FIG. 10 is a schematic diagram illustrating an example of a display screen that displays the progress status of the quality process of education management. [Figure 18] 1 is a flowchart showing the flow of a quality process management process according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0009] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described with reference to the drawings. Fig. 1 is a schematic diagram showing a quality process control system 1 according to one embodiment of the present invention.

[0010] The quality process control system 1 is a system included in a quality control system (not shown) that supports quality control by a user.

[0011] A quality control system mainly includes a document management function for managing documents, an event management function for managing quality processes, and an education and training function for educating and training employees. FIG. 2 is a schematic diagram showing the modularized functions of a quality control system. In this specification, the term "document" refers to an electronic document in which text information and image information have been digitized. One document refers to, for example, one digitized law, standard, manual, procedure manual, etc.

[0012] As shown in Figure 2, the quality control system is modularized into various functions, including basic functions 61, document and record management functions 62, AI (artificial intelligence) support functions 63, education and training functions 64, and event management functions 65. The basic functions 61 include, for example, user management, authority management, audit logs, notifications, and multilingual settings. The document and record management functions 62 include, for example, document storage, editing, version management, approval workflows, document difference display (displaying differences between documents), access restrictions for stored documents, and electronic signatures. The AI ​​support functions 63 include, for example, presentation of necessary documents and actions, automatic document draft generation, automatic document checking, automatic test question creation, and accumulated data analysis. The education and training functions 64 include, for example, test question creation and distribution, attendance record management, and respondent competency assessment. The event management functions 65 include template and process management functions, alert functions, and other functions. Because each function in the quality control system is modularized, it has a highly scalable functional design.

[0013] The quality process control system 1 is a system included in a quality control system and mainly comprises functions included in an AI assistance function 63, an event management function 65, etc. When the quality process control system 1 receives event information 2, it analyzes the event information 2, sets up a workflow for taking necessary actions for the event indicated by the event information 2, and executes a quality process control process that monitors the progress of the workflow. The event information 2 is information about an event related to quality and is information that triggers the quality process. Examples of the event information 2 include text information contained in digitized electronic documents and emails. The event indicated by the event information 2 is an event that may affect the quality of a product or service. Note that the event indicated by the event information 2 includes all quality-related events and is not limited to adverse events.

[0014] The quality process and event information 2 will be described with reference to Fig. 3. Fig. 3 is a schematic diagram showing the types of quality processes in a QMS.

[0015] A quality process is a process related to quality control. In this embodiment, the quality process in a QMS for medical devices, pharmaceuticals, etc. is used as an example. As shown in FIG. 3, types of quality processes in a QMS include, for example, "complaint management," "deviation / nonconformance management," "internal audit," "CAPA (Corrective Action and Preventive Action)," "change management," and "education management." In this embodiment, the type and content of each quality process are defined in advance.

[0016] In complaint management, for example, dissatisfaction and problem reports (hereinafter simply referred to as complaints) from customers and medical professionals are accepted, the contents are analyzed, and countermeasures are implemented according to the level of importance. Event information 2 that triggers complaint management includes, for example, complaint information about complaints from customers and medical professionals.

[0017] Deviation / non-conformance management involves a series of tasks, for example, identifying and isolating products that are determined to be out of specification during inspection, creating records, and deciding whether to discard them or correct them before using them.Event information 2 that triggers deviation / non-conformance management includes, for example, inspection results that indicate that a product deviates from and does not conform to predetermined criteria such as specifications (hereinafter referred to as deviation / non-conformance).

[0018] In an internal audit, for example, a person in charge within an organization visits each department, checks work records and procedures based on a checklist, evaluates compliance with regulatory requirements, and compiles the results into a report. An example of event information 2 that triggers an internal audit is a report containing an evaluation result that indicates non-compliance with regulatory requirements.

[0019] CAPA involves identifying the root cause of quality problems, developing and implementing corrective and preventive measures for those problems, and verifying the effectiveness of those measures. For example, when a product defect is discovered, CAPA involves conducting an investigation to identify the root cause, and then planning, implementing, and verifying measures to correct the problem and prevent similar problems in the future.

[0020] As shown in Figure 3, CAPA is implemented when, for example, the results of complaint management, deviation / nonconformance management, or internal audits indicate that it is necessary to identify root causes and establish corrective measures. In other words, event information 2 that triggers a CAPA can include, for example, information about the results of complaint management, deviation / nonconformance management, or internal audits.

[0021] Change management involves investigating changes to products, such as manufacturing methods (e.g., product materials and manufacturing procedures), and quality inspection methods, as well as the impact of the changes. For example, change management involves investigating in advance the impact on quality of changes to product materials and manufacturing procedures, and implementing the changes only after obtaining approval from multiple responsible parties. As shown in Figure 3, change management is executed, for example, when there are changes to products, etc. in the corrective and preventive actions created by CAPA. In other words, event information 2 that triggers change management can include, for example, information on corrective or preventive actions created by CAPA.

[0022] In training management, for example, employees in an organization are trained and the effectiveness of the training is confirmed. For example, training management is implemented when it is determined that employees need training based on the contents of changes in manufacturing methods, quality inspection methods, etc. (hereinafter referred to as product changes). Event information 2 that triggers training management includes, for example, the results of a determination of the need for training.

[0023] Each quality process has a predetermined outline of the actions and procedures that must be performed, which are tasks that must be performed according to its type. Furthermore, each type of quality process is interrelated. For example, in the quality processes of complaint management, deviation / nonconformity management, and internal audits, after performing the respective prescribed actions, it is necessary to identify the root cause of Event Information 2 and implement corrective measures, so the actions prescribed in CAPA are executed. Furthermore, if a CAPA results in the need to change the product design, etc., the actions prescribed in change management are executed. Furthermore, if a change management result in a change to the product's manufacturing procedures, etc., and training for manufacturing personnel, etc. is required, the actions prescribed in training management are executed. Therefore, by identifying the type of quality process, it is possible to set the outline of the workflow for the entire quality process triggered by Event Information 2.

[0024] The quality process control system 1 performs quality process control processing using a database 3 storing quality control information, which is information related to quality control, and activity records related to past quality control (hereinafter referred to as quality control activity records), and a trained model 4. Examples of quality control information include rule information such as quality control manuals, laws, regulations, and rules; work procedures related to product manufacturing, work instructions, information recorded on checklists listing inspection items, and complaint response process specifications. Examples of quality control activity records include information related to past quality processes (hereinafter referred to as quality process history), information recorded in manufacturing records and inspection records. Examples of rule information include information recorded in the QMS (Quality Management System) Ministerial Ordinance, ISO (International Organization for Standardization) 13485, ISO 14971, and other related guidelines. Examples of quality process history include past event information 2, root causes of events indicated by the event information 2, and quality process response history. In this embodiment, the quality control information and quality control activity records are text data. The database 3 may also store organizational information related to organizations to which users of the quality process control system 1 belong. The organizational information may include information showing the organizational structure such as an organizational chart, the organization name, the representative's name, location, telephone number, email addresses of the organization and its employees, website, year of establishment, number of employees, business details, etc. The organizational information may also include rules within the organization such as company regulations (hereinafter referred to as "organizational regulations").

[0025] An example of the hardware configuration of the quality process control system 1 will be described with reference to Fig. 4. Fig. 4 is a block diagram showing the hardware configuration of the quality process control system 1.

[0026] 4, the quality process control system 1 includes a computer 18, a storage unit 13, a communication unit 14, an input unit 15, and a display unit 16. A bus 17 and the like connect these units together.

[0027] The computer 18 includes a processor 10 and a read-only memory (ROM) 11 and a random-access memory (RAM) 12 as main storage devices. The processor 10 may be a central processing unit (CPU), a microprocessing unit (MPU), a system on a chip (SoC), a digital signal processor (DSP), a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a field-programmable gate array (FPGA). Alternatively, the processor 10 may be a combination of these. The processor 10 may also be a combination of these with a hardware accelerator or the like. The processor 10 controls each component to realize various functions of the quality process control system 1 based on programs such as firmware, system software, and application software stored in the ROM 11, the RAM 12, or an auxiliary storage device that is part of the storage unit 13. Note that some or all of the programs may be incorporated into the circuitry of the processor 10.

[0028] The storage unit 13 is a storage area for various programs and various data for causing the hardware group to function as the quality process control system 1, and can be configured with a ROM, RAM, flash memory, a solid-state drive (SSD), a hard disk drive (HDD), or the like. Specifically, the storage unit 13 stores programs for causing the computer 18 to execute each function of this embodiment. The storage unit 13 may also include a database 3 for storing quality control information, quality control activity records, and the like. The storage unit 13 may also store a trained model 4. In this embodiment, the database 3 and the trained model 4 are located outside the quality control system, such as on the cloud. The storage unit 13 stores definitions of the type and content of each quality process (hereinafter referred to as process definition information).

[0029] The communication unit 14 executes processing for the quality process control system 1 to communicate with other devices via a network. For example, the quality process control system 1 can connect to the database 3 via the communication unit 14 and refer to or acquire information stored in the database 3. For example, the quality process control system 1 can analyze the event information 2 by communicating with the trained model 4 via the communication unit 14.

[0030] The input unit 15 is a user interface electrically connected to the computer 18. The input unit 15 is composed of buttons, a mouse, a keyboard, a display, etc. The display is composed of, for example, a liquid crystal display (LCD) or an organic electroluminescent (EL) display, and a touch panel that detects the position touched by the user is provided on the image display surface of the display. The user can input information by touching the image display surface of the display.

[0031] The display unit 16 is a user interface electrically connected to the computer 18. The display unit 16 is configured by a display. Images and various information transmitted from the memory unit 13, the communication unit 14, the input unit 15, the computer 18, etc. are displayed on the display unit 16.

[0032] Next, various functions realized by the hardware configuration of the quality process control system 1 will be described with reference to Fig. 5. As shown in Fig. 5, the quality process control system 1 includes an event reception unit 101, a storage processing unit 102, a classification unit 103, an action generation unit 104, a setting unit 105, a monitoring processing unit 106, a similar event search unit 107, a time series analysis unit 108, and an education management unit 110 as functional units operating on the processor 10.

[0033] The event reception unit 101 executes a process of receiving information input by a user. For example, the event reception unit 101 receives event information 2 input via the communication unit 14 or the input unit 15. Furthermore, for example, the event reception unit 101 executes a process of receiving quality control information and quality control activity records via the communication unit 14 or the input unit 15.

[0034] The storage processing unit 102 executes a process of storing the quality control information and quality control activity records acquired via the event receiving unit 101. The storage processing unit 102 stores the quality control information and quality control activity records in the database 3. It is preferable that the quality control information and quality control activity records as documents are divided into multiple semantic chunks and stored in the database 3 in a vectorized state.

[0035] The classifying unit 103 identifies the type of quality process that the event information 2 causes based on the event information 2 received by the event receiving unit 101 and the event definition information stored in the storage unit 13 .

[0036] The classification unit 103 divides the text data of the event information 2 received by the event reception unit 101 into multiple semantic chunks and vectorizes them. In this embodiment, the classification unit 103 interprets the content of the event information 2 using the trained model 4.

[0037] The trained model 4 used by the classification unit 103 in this embodiment is a trained model 4 that has been machine-learned so that, when event information 2 is input, it estimates and outputs the type of quality process corresponding to the input event information 2. The trained model 4 in this embodiment is a large-scale language model that interprets document content, including the context and nuances of the event information 2 and event definition information, by natural language understanding. When event information 2 and a prompt for identifying the quality process classification are input, the trained model 4 compares the content of the event information 2 with the content of the event definition information to identify the classification of the quality process caused by the event information 2.

[0038] The action generation unit 104 analyzes the event information 2 using a trained model 4 that analyzes the event information 2 based on the information in the database 3, and generates an action required for the quality process corresponding to the event information 2.

[0039] The trained model 4 used by the action generation unit 104 is a trained model 4 that has been machine-learned so that, when event information 2 is input, it estimates and outputs candidate root causes for the event information 2 and actions required for the quality process corresponding to the event information 2.

[0040] Machine learning for the trained model 4 includes unsupervised learning, semi-supervised learning, reinforcement learning, etc. In the case of supervised learning, known algorithms for the trained model such as decision trees, random forests, support vector machines, neural networks, deep learning, naive Bayes methods, and K-nearest neighbor methods can be used. Quality control information, quality control activity records, etc. stored in the database 3 are used as data for machine learning for the trained model 4.

[0041] The action generation unit 104 may additionally train (fine-tune) the trained model 4 based on the information in the database 3 at a predetermined timing, and may analyze the event information 2 and generate necessary actions using the additionally trained trained model 4. The predetermined timing may be, for example, the timing when new quality control information, quality control activity records, or other information is stored in the database 3.

[0042] Furthermore, for example, the action generation unit 104 may use Retrieval Augmented Generation (RAG) to identify potential root causes of the problem indicated by the event information 2 and estimate actions required for the quality process corresponding to the event information 2. The trained model 4 used in the Retrieval Augmented Generation may be, for example, a large-scale language model that interprets document content, including the context and nuances of the event information 2 and the information in the database 3, using natural language understanding.

[0043] Specifically, the action generation unit 104 may input the event information 2 and information in the database 3 related to the event information 2 into the trained model 4, and have the trained model 4 estimate the root cause of the problem indicated by the event information 2 and the actions required for the quality process caused by the event information 2.

[0044] An example of a method in which the action generation unit 104 uses the trained model 4 to generate candidates for the root cause of the event information 2 and the necessary actions will be described.

[0045] The action generation unit 104 may generate actions necessary for the quality process based on the type of quality process caused by the event information 2 identified by the classification unit 103, and candidate root causes, corrective measures, preventive measures, etc. for the event information 2 estimated by the trained model 4.

[0046] Specifically, the action generation unit 104 may identify the type of action required for the quality process based on the type of the identified quality process. For example, if the type of quality process is education management, the action generation unit 104 may generate actions related to education management, such as education and training, confirmation of the effectiveness of education and training, etc. For example, if the type of quality process is complaint management, the action generation unit 104 may generate actions related to CAPA actions, such as identification of root cause candidates, corrective measures, preventive measures, and education and training as necessary, based on an analysis of complaint information.

[0047] Next, if the type of quality process is mainly complaint management, deviation / non-conformance management, internal audit, CAPA, etc., the action generation unit 104 may generate specific actions based on the estimated results of the trained model 4, such as potential root causes, corrective measures for event information 2, and preventive measures.

[0048] An example of how the trained model 4 estimates potential root causes of event information 2 and corrective and preventive measures in the quality process is described below.

[0049] For example, the trained model 4 may estimate potential root causes based solely on the document content of the event information 2.

[0050] Furthermore, for example, the trained model 4 may search the database 3 for quality process history of past event information 2 similar to the currently received event information 2 (hereinafter referred to as similar event information), and estimate potential root causes, corrective actions, and preventive actions for the currently received event information 2 based on the searched quality process history. Specifically, the trained model 4 may identify potential root causes of the current event information 2 and the corrective actions and preventive actions for the quality process from the response history of the quality process indicated by the extracted quality process history. The response history of the quality process may include, for example, the root causes, corrective actions, and preventive actions for the event information 2, as well as verification results for the effectiveness of those actions. Note that the extraction of similar event information from the database 3 may be performed by the trained model 4 or the action generation unit 104.

[0051] Furthermore, for example, if the event information 2 is complaint information about a product defect or the like, the trained model 4 may extract the quality manual or work procedure manual corresponding to the defective product from the database 3. Then, the trained model 4 may estimate possible root causes of the current event information 2 and corrective and preventive measures in the quality process based on the document contents of the extracted quality manual or work procedure manual.

[0052] For example, if the event information 2 is complaint information about a product defect or the like, the trained model 4 may identify at least one of the production record and inspection record of the product from the lot of the product for which the complaint occurred. The trained model 4 may then identify potential root causes of the event information based on at least one of the identified production record and inspection record. For example, the trained model 4 may identify the materials and personnel used in manufacturing the product from the document contents of the production record, and estimate potential root causes, corrective measures, preventive measures, etc. based on the identification results. For example, if the person in charge of manufacturing the product for which a defect occurred is a new employee or the like and tends to have low skills, the trained model 4 may identify human error by the person in charge as a potential root cause, and may identify changing the person in charge of manufacturing or providing education and training as corrective and preventive measures.

[0053] The action generation unit 104 may predict the effectiveness of the generated corrective and preventive measures based on the quality process history of the similar event information stored in the database 3. The action generation unit 104 may determine the effectiveness of the generated corrective and preventive measures based on the corrective and preventive measures in the quality process caused by the similar event information and result information related to the results of the corrective and preventive measures. The result information may be, for example, the number of occurrences of similar event information that occurred after the corrective and preventive measures were taken. The action generation unit 104 may also cause the trained model 4 to predict the effectiveness of the corrective and preventive measures.

[0054] 6 is a schematic diagram showing an example of complaint information that is displayed on the display unit 16 and that triggers a complaint management quality process. In the example shown in FIG. 6, a title display unit 21, a text display unit 22, and an analysis start button 23 are displayed on the display unit 16.

[0055] The title display section 21 displays the title of the event information 2. In the example shown in Fig. 6, the title of the event information 2 is displayed in the title display section 21 as "Complaint information from patients (emails, phone memos, etc.)."

[0056] The text display unit 22 displays the contents of the event information 2. In this embodiment, the text display unit 22 displays the entire text of the event information 2. When the analysis start button 23 is pressed, the event information 2 is analyzed using the trained model 4.

[0057] The similar event search unit 107 uses the trained model 4 to search for similar event information from past quality control activity records in the database 3, and displays countermeasures for the problems indicated by the searched similar event information as reference information. The reference information may also include the effects of the countermeasures.

[0058] Fig. 7 is a schematic diagram showing an example of a display screen that displays similar event information and analysis results of the event information (potential root causes of the event information). In the example shown in Fig. 7, an analysis result display section 24 and a similar event information display section 25 are displayed on the display unit 16. Note that the problem indicated by the similar event information in the example shown in Fig. 7 is an "alarm malfunction."

[0059] The analysis result display unit 24 displays the analysis results of the event information 2 by the trained model 4. The analysis result display unit 24 displays information including potential root causes of the event information 2, reasons for estimating the potential root causes, etc. as the analysis results of the event information 2. In the example shown in FIG. 7, the analysis result display unit 24 displays "communication error with battery sensor" as a potential root cause of the alarm malfunction.

[0060] The similar event information display unit 25 displays past similar event information that is similar to the event information 2 that is the subject of analysis by the trained model 4. As shown in Fig. 7 , the similar event information display unit 25 displays, from left to right, a "complaint number" that is the number of the event information 2, a "date" that is the date on which the event information 2 was received, a "content" that is the content of the event information 2, and a "status" that indicates the progress of the quality process for the event information 2.

[0061] 8 is a schematic diagram showing an example of a display screen that displays a summary of complaint information as event information and corrective and preventive measures. In the example shown in FIG. 8, an analysis summary display section 26 and a corrective and preventive measures display section 27 are displayed on the display unit 16.

[0062] The analysis summary display unit 26 displays a summary of the contents of the event information 2. As shown in Fig. 8, the analysis summary display unit 26 displays "product information" of the defective product, "complaint classification" indicating the type and severity of the complaint, "reporter information" indicating the person who reported the complaint and the date and time of the complaint, and "complaint content summary" which shows a summary of the contents of the event information 2.

[0063] The corrective and preventive action display section 27 displays countermeasures, such as corrective and preventive actions, in the quality process that are generated by the event information 2. As shown in FIG. 8, the corrective and preventive action display section 27 displays items such as "Emergency Response Measures," "Root Cause Investigation," "Proposed Corrective Actions," and "Proposed Preventive Actions." The "Emergency Response Measures" item displays items that need to be implemented urgently after receiving the event information 2. The "Root Cause Investigation" item displays items that need to be implemented to investigate the root cause. The "Proposed Corrective Actions" item displays corrective actions in the quality process. The "Proposed Preventive Actions" item displays preventive actions in the quality process. By viewing the display that aggregates this various information, the user can easily understand the event information 2, its root cause, and countermeasures.

[0064] The setting unit 105 executes a process of setting a workflow for executing the action generated by the action generating unit 104. The setting unit 105 may set the workflow based on, for example, the action generated based on the event information 2, the type of quality process classified by the classifying unit 103, and information in the database 3.

[0065] The setting unit 105 may set a person in charge and a deadline for each action based on, for example, the quality process history of the similar event information. For example, the setting unit 105 may reflect the person in charge and the deadline for each action in the workflow indicated by the quality process history of the similar event information in the workflow of the current event information 2. For example, the setting unit 105 may set a person in charge of each action based on organizational information such as an organizational chart of users in the database 3. For example, the setting unit 105 may specify a deadline for each action based on an internal rule. For example, the setting unit 105 may determine the priority and impact of each action and the entire quality process based on the type of quality process and the description of the event information 2, and specify the deadline for each action based on the determination results of the priority and impact. In this case, if the internal rule specifies a deadline according to the priority or impact of the action, the setting unit 105 may specify the deadline for each action based on the determination results of the priority and impact and the internal rule. Examples of deadlines based on the priority and impact of an action include, for example, that the action must be executed within three days if the impact of the action is "high," within ten days if the impact is "medium," and within 30 days if the impact is "low."

[0066] The monitoring processing unit 106 monitors the progress of the workflow set by the setting unit 105 and executes a process of displaying the progress. For example, the monitoring processing unit 106 may manage the deadline for each action in the set workflow. For example, the monitoring processing unit 106 may manage the deadline by displaying the deadline for each action on the display unit 16. Furthermore, for example, when the remaining time until the deadline is shorter than a predetermined time and the action has not been completed, the monitoring processing unit 106 may notify the person in charge of the action to urge them to perform the action. When notifying the person in charge of the action to urge them to perform the action, the monitoring processing unit 106 may, for example, search for the email address of the person in charge stored in the database 3 and notify the person in charge by email.

[0067] 9A to 9C are schematic diagrams showing examples of display screens of workflows of quality processes set by the setting unit 105 and displayed by the monitoring processing unit 106. The type of quality process of the workflows shown in FIGS. 9A to 9C is CAPA.

[0068] In the example shown in Figures 9A to 9C, a basic information display section 30 that displays basic information about the quality process, an overview display section 31 that displays an overview of the workflow in the quality process, and a detail display section 32 that displays details of the workflow are displayed on the display unit 16.

[0069] As shown in FIG. 9A, the basic information display unit 30 displays the following items: "Event Name," which indicates the name of the quality process, and the following items for the quality process: "Responsible Person," "Status," "Priority," "Impact," "Response Deadline," "Created Date," "Last Updated Date," and "Details." The "Event Name" item in the basic information display unit 30 displays "Medical Monitor Battery Life Shortage CAPA" as the quality process. The "Progress" item in the basic information display unit 30 also displays that the workflow of the quality process has progressed 45%. As shown in FIG. 9A, the monitoring processing unit 106 monitors the progress of the workflow of the quality process.

[0070] The overview display section 31 shows an overview of the quality process workflow. In the example shown in FIG. 9A, the overview display section 31 displays the order in which processes 1 to 5 are to be executed, the status of each process, and details of each process. The status of each process may be displayed as "Completed," indicating that the process has already been completed, "In Progress," indicating that the process is in progress, or "Not Started," indicating that the process has not yet been started. In the example shown in FIG. 9A, process 1 is displayed as "Problem Identification and Assessment," process 2 as "Root Cause Analysis," process 3 as "Corrective and Preventive Action Plan," process 4 as "Implementation," and process 5 as "Validation and Completion." By checking the overview display section 31, the user can quickly grasp the progress of the workflow.

[0071] As shown in FIGS. 9A to 9C, the details display section 32 displays, for each process, the actions that are included in each processing of the quality process and that are required for the quality process, and the details of the actions.

[0072] As shown in Fig. 9A to Fig. 9C, the detail display section 32 is composed of a title display section 321 and an action display section 322. The title of each processing step is displayed in the title display section 321. For example, in the example shown in Fig. 9A, "Problem Identification and Evaluation" is displayed as the title of Process 1.

[0073] Details of each action included in each process are displayed in the action display section 322. The action display section 322 displays the title of each action, a status indicating whether the action is completed, in progress, or not yet started, a deadline, a completion date, a person in charge, a priority, related documents, details, etc. In the example of Figures 9A to 9C, as shown in the action display section 322, processes 1 to 3 are each made up of two actions, and processes 4 and 5 are each made up of three actions.

[0074] As shown in FIGS. 9A and 9B, process 1, process 2, and process 3 "Corrective and Preventive Action Plan" up to action 1 "Corrective Action Plan" have been completed, and action 2 "Preventive Action Plan" of process 3 is in progress. The action display section 322 displays that for action 2 "Preventive Action Plan" of process 3, the status is "In Progress," the deadline is "November 15, 2024," the related document is "Preventive Action Plan #PAP-2024-032," and information such as priority and person in charge. By checking the action display section 322 of the workflow, the user can easily understand the person in charge, the deadline, details of the action, as well as related documents.

[0075] Here, when documents and records are scattered across various locations within an organization, searching for the document to be used can take time and effort. In this embodiment, the monitoring processor 106 uses the trained model 4 to extract documents related to each action in the quality process from the documents and records collected in the database 3. Then, as shown in FIGS. 9A to 9C, the monitoring processor 106 displays the extracted related documents in association with the corresponding actions. This allows the person in charge to proceed with each action efficiently and appropriately.

[0076] FIG. 10 shows an event dashboard 33 that displays multiple quality processes monitored by the monitoring processing unit 106 and the progress of their workflows.

[0077] As shown in FIG. 10, the event dashboard 33 displays an event progress display section 34, a search bar 35, a first sort key 36, a second sort key 37, an event list display section 38, and a new event registration button 39.

[0078] The event progress display unit 34 displays the number of quality processes whose workflows are not yet completed and that are being monitored by the monitoring processing unit 106. In the example shown in Fig. 10, there are 11 quality processes whose workflows have not yet started, 7 quality processes that are in progress, and 2 quality processes that have expired.

[0079] The search bar 35 is a bar for inputting keywords to search for quality processes. The first sort key 36 is a key for sorting quality processes according to the progress of the workflow. The second sort key 37 is a key for sorting quality processes according to priority.

[0080] The event list display unit 38 displays a list of quality processes whose workflows are incomplete. The event list display unit 38 displays the name of the quality process, the status indicating whether the workflow has been started or is in progress, the person in charge, the deadline for response, the impact, the priority, the number of processes, the progress expressed as a percentage, and the like.

[0081] The new event registration button 39 is a button for registering a new quality process and its workflow.

[0082] The time series analysis unit 108 determines an increase / decrease trend of the event indicated by the currently received event information 2 through time series analysis of the similar event information, and executes a predetermined process based on the determined increase / decrease trend. The increase / decrease trend of the event identified through time series analysis means, for example, an increase / decrease trend of the number of similar event information items received by the event reception unit 101. The time series analysis unit 108 may tally up the number of received similar event information items for each predetermined period, and determine an increase / decrease trend of the event indicated by the event information 2 based on the tallying result of the number of similar event information items for each predetermined period. For example, the time series analysis unit 108 may tally up the number of similar complaint information items (hereinafter referred to as the number of complaints) received by the event reception unit 101 for each week, and compare the number of complaints this week with the average number of complaints over the past three weeks. If the number of complaints this week exceeds the average number of complaints over the past three weeks, the time series analysis unit 108 may determine that the increase / decrease trend of the event is an "increasing trend."

[0083] The time series analysis unit 108 may output the determined increase / decrease trend of events as a predetermined process. Specifically, the time series analysis unit 108 may display the increase / decrease trend of events on a dashboard generated by the monitoring processing unit 106, or may search for the email address of a person in charge stored in the database 3 and send the result to the searched email address. For example, if the time series analysis unit 108 determines that the increase / decrease trend of events is an increasing trend, the time series analysis unit 108 may send an email to the email address of the person in charge urging them to re-investigate the root cause of the event as a predetermined process. Furthermore, for example, if the increase / decrease trend of events related to complaint information about the same product has been increasing for a predetermined period of time, the time series analysis unit 108 may set the priority of the entire quality process and the action for the complaint information to be high. When the priority of the entire quality process or the action is set to be high, the setting unit 105 sets the deadline for the entire quality process and the action to be short. The time series analysis unit 108 may have the trained model 4 determine the increase / decrease trend of events and execute a predetermined process.

[0084] The time series analysis unit 108 may track temporal changes in the frequency and type of events indicated by the event information 2 and perform trend analysis to identify at least one of seasonality and periodicity. The trend analysis may visualize changes in the occurrence rate of the event before and after the implementation of corrective measures for the event, and evaluate the countermeasures and their effectiveness. For example, the time series analysis unit 108 may graph battery-related malfunctions of implantable cardiac pacemakers on a monthly, quarterly, or annual basis and track associations with specific batches or manufacturing periods. For example, if the results of the tracking show a pattern in which the occurrence rate increases in the summer, the time series analysis unit 108 may instruct the action generation unit 104 to generate an action to revise the battery temperature management protocol. By implementing the action, for example, it may be possible to reduce the occurrence rate of battery-related malfunctions in implantable cardiac pacemakers.

[0085] Next, the education management unit 110 will be described. The education management unit 110 performs processing for education and training (hereinafter referred to as education and training processing). The education management unit 110 of this embodiment executes education and training processing when the quality process is education management or when an action related to education management exists in the workflow.

[0086] The education and training process is a process of generating test questions for education and training, evaluating the respondent's level of understanding, and generating and outputting additional learning content for the respondent. The education and training process will be described with reference to FIGS. 11 to 14.

[0087] As shown in FIG. 5, the education management unit 110 includes a necessity determination unit 111, a specified information reception unit 112, a test generation unit 113, an understanding level assessment unit 114, and a learning optimization unit 115.

[0088] The necessity determining unit 111 executes a process of determining whether or not education and training is necessary based on at least one of the event information 2, the workflow, and the quality control information as a document (hereinafter, referred to as a quality control document).

[0089] For example, the necessity determining unit 111 determines that training processing is necessary when the quality process caused by the event information 2 is training management.

[0090] Furthermore, for example, the necessity determination unit 111 may determine that education and training processing is necessary when an action related to education and training exists in the workflow set by the setting unit 105. Specifically, for example, the necessity determination unit 111 may determine that education and training processing is necessary when education and training exists as an action in a workflow generated based on complaint information as event information.

[0091] Furthermore, for example, if the workflow includes an action that changes the quality control document, the necessity determination unit 111 may determine whether education and training is necessary based on the details of the change in the quality control document. Specifically, for example, if the workflow includes an action that changes the manufacturing method or quality confirmation method for a product such as a medical device, the necessity determination unit 111 determines whether education and training is necessary based on the details of the change in the quality control document, such as a procedure manual, that accompanies the action. For example, the necessity determination unit 111 may determine whether education and training is necessary based on at least one of the degree of change in the quality control document, the importance of the change, and the impact of the change. The necessity determination unit 111 may check the details of the change in the quality control document from the action in the workflow, or may connect to database 3 to check the details of the change in the quality control document.

[0092] The specified information receiving unit 112 executes a process of receiving information for generating test questions for education and training. Specifically, the specified information receiving unit 112 receives the quality control document that will be the subject of the test questions and the specified information input by user operation via the input unit 15 or the display unit 16. The specified information includes at least information regarding the type of test question (hereinafter referred to as question type), the difficulty level of the test question, the number of questions, and additional option information.

[0093] 11 and 12 are schematic diagrams showing a test generation setting screen 40 for setting conditions for generating test questions for education and training. Fig. 11 is a schematic diagram showing an example of the test generation setting screen 40 for selecting quality control documents that will be the subject of test questions. Fig. 12 is a schematic diagram showing an example of the test generation setting screen 40 for inputting specific information for test questions.

[0094] The test generation setting screen 40 is displayed on the display unit 16 and allows the user to input the conditions for generating test questions. The test generation setting screen 40 shown in Fig. 11 displays a screen switching bar 41, a document search bar 42, a document selection section 43, a selected document display section 44, and a test generation start button 45.

[0095] The screen switching bar 41 is a bar for switching the display of the test generation setting screen 40 between a screen for selecting a quality control document and a screen for inputting specified information. The document search bar 42 is a bar for inputting keywords for searching for quality control documents.

[0096] The document selection unit 43 displays a list of quality control documents that can be used as test question topics and accepts the selection of a quality control document to be used as a test question topic. The document selection unit 43 displays the title and version of the quality control document, the department that handles the quality control document, the number of pages, and the registration or update date. By clicking on the display of the desired quality control document, the user can select the clicked quality control document as a test question topic.

[0097] The selected document display section 44 displays the title and version of the quality control document selected as the subject of the test question via the document selection section 43, as well as the department that handles the quality control document. When the test generation start button 45 is pressed, the generation of test questions begins.

[0098] The test generation setting screen 40 shown in Figure 12 displays a screen switching bar 41, a question number selection section 46, a question type selection section 47, a difficulty level selection section 48, an additional option selection section 49, a selected document display section 44, and a test generation start button 45.

[0099] The question number selection unit 46 is composed of a slider 461, a slider bar 462, and a question number display unit 463, which are displayed on the display unit 16. The question number selection unit 46 is configured so that the user can select the number of test questions by moving the slider 461 along the slider bar 462. The number of questions selected by the user's operation is displayed in the question number display unit 463, as shown in FIG.

[0100] The question type selection unit 47 displays multiple types of test questions that the user can select. In the example shown in Fig. 12, the question type selection unit 47 displays check boxes for four types: "multiple choice," which allows the user to select the correct answer from multiple candidates; "true / false," which allows the user to answer whether the statement is correct or incorrect; and "descriptive (short sentence)" and "descriptive (long sentence)," which allow the user to write the answer. The question type selection unit 47 is configured so that the user can select a question type by clicking the check box. In the example shown in Fig. 12, "multiple choice" and "true / false" have been selected.

[0101] The difficulty level selection unit 48 displays check boxes for selecting the difficulty level of the test questions. In the example shown in FIG. 12, the difficulty level selection unit 48 displays three levels of difficulty: "basic," "standard," and "advanced." The difficulty level selection unit 48 is configured so that the user can select the difficulty level of the questions by clicking the check boxes. In the example shown in FIG. 12, the "standard" difficulty level is selected.

[0102] The additional option selection section 49 displays a selection bar for selecting whether or not to include additional options for test questions. In the example shown in FIG. 12, the additional option selection section 49 displays additional options such as "Include explanations," "Randomize questions," and "Set a time limit." The additional option selection section 49 is configured so that the user can select whether or not to include each additional option by clicking the selection bar. In the example shown in FIG. 12, all additional options are selected.

[0103] The test generation unit 113 generates test questions for education and training based on the quality control document and the specified information. The test generation unit 113 generates test questions that are necessary for understanding the contents of the quality control document received by the specified information receiving unit 112 and are in line with the specified information. The test generation unit 113 of this embodiment may generate test questions using a trained model 4. The trained model 4 used by the test generation unit 113 may be a large-scale language model trained by machine learning to generate and output test questions based on the input quality control document and the specified information.

[0104] Fig. 13 is a schematic diagram showing an example of a preview screen 50 for checking the contents of test questions generated by the test generation unit 113. In the example shown in Fig. 13, the subject quality control document is "Product Inspection Procedure Manual PQC-001," and test questions with a difficulty level of "Standard," and question types of "Multiple Choice," "True / False," and "Descriptive (Short Answer)" are displayed.

[0105] 13 displays a copy button 51 for copying the generated test questions, an export button 52, a delete button 53, a regenerate button 56 for causing the test generation unit 113 to regenerate the test questions, and a save button 57 for saving the test questions. Also displayed on the preview screen 50 are a basic information display section 54 for displaying basic information about the test questions, and a question display section 55 for displaying each question.

[0106] The basic information display section 54 displays the title, purpose, difficulty level, time limit, and passing score of the test question. The question display section 55 displays the question text, options, explanation, question type, etc. of each test question. In the example shown in Fig. 13, multiple choice question 1, true / false question 2, and descriptive (short answer) question 3 are displayed.

[0107] The understanding level assessment unit 114 assesses the level of understanding of the respondent of the test questions based on the answers to the test questions generated by the test generation unit 113.

[0108] The comprehension evaluation unit 114 may evaluate the respondent's level of understanding based on the percentage of correct answers to the test questions. The comprehension evaluation unit 114 may also evaluate the respondent's level of understanding based on at least one of the time required to answer the test questions, the type of test question, the difficulty of the test question, and the respondent's answer history for past test questions, in addition to the percentage of correct answers to the test questions. For example, given the same percentage of correct answers, the comprehension evaluation unit 114 may evaluate the respondent's level of understanding higher the shorter the time required to answer the test questions. For example, the comprehension evaluation unit 114 may evaluate the respondent's level of understanding higher if the respondent correctly answers a descriptive test question rather than a multiple-choice or true / false test question. This allows for more accurate evaluation of the respondent's level of understanding. The comprehension evaluation unit 114 may also evaluate the respondent's level of understanding by topic in the test questions. This allows for efficient identification of the respondent's areas of expertise and areas requiring improvement. If the respondent incorrectly answers a question that is the same as a question previously answered incorrectly as indicated by the answer history, the comprehension evaluation unit 114 may evaluate an area related to the question as an area requiring improvement.

[0109] An example of the analysis result of the answerer's level of understanding of the test question output by the understanding assessment unit 114 will be described with reference to Fig. 14. Fig. 14 is a schematic diagram of an understanding display screen 120 showing the analysis result of the answerer's level of understanding of the test question.

[0110] The comprehension level display screen 120 displays a title display section 121, a pass / fail display section 122, a display switching section 123, a total score display section 124, an exam status display section 125, a performance by question type display section 126, a performance by topic display section 127, and a comprehension level analysis display section 128.

[0111] The title display section 121 displays the title of the analysis result of the level of understanding of the test questions. In the example shown in Fig. 6, "Product Inspection Procedure Manual PQC-001 Comprehension Test - Comprehension Analysis" is displayed as the title of the analysis result of the level of understanding in the title display section 121. The pass / fail display section 122 displays the pass / fail result of the respondent for the test questions.

[0112] The display switching unit 123 is a bar for switching between an "Overall Analysis" screen that displays the analysis results of the level of understanding of the entire test questions, a "Detailed Results" screen that displays details of the analysis results of the level of understanding, and an "Improvement Points" screen that displays points for improvement for the respondent. In the example shown in Fig. 14, the "Overall Analysis" screen is displayed.

[0113] The total score display section 124 displays the pass score for the test questions, the respondent's total score, and the respondent's level of understanding. The test status display section 125 displays the number of questions answered by the respondent, the number of correct answers, and the time required to answer (required time). The performance by question type display section 126 displays the correct answer rate for test questions of each question type.

[0114] The performance by topic display section 127 displays the respondent's accuracy rate for each question topic. The comprehension analysis display section 128 displays the areas in which the respondent has strengths and areas that need improvement in the subject of the test question (product inspection procedure manual PQC-1 in FIG. 14). By checking the performance by topic display section 127 and the comprehension analysis display section 128, it is possible to easily grasp the areas in which the respondent has a high level of comprehension and the areas in which the respondent has a low level of comprehension in the subject of the test question.

[0115] The learning optimization unit 115 generates and outputs additional learning content personalized for each respondent based on the comprehension assessment results by the comprehension assessment unit 114. Here, "personalized additional learning content" may mean additional learning content tailored to each respondent, additional learning content tailored to the organization to which the respondent belongs, or additional learning content tailored to each category categorized according to the respondent's attributes. The respondent's attributes may be, for example, gender, age, years of experience in work related to the subject of the test question, etc.

[0116] The learning optimization unit 115 may generate additional learning content based on, for example, the respondent's overall level of comprehension of the test questions. Alternatively, for example, the learning optimization unit 115 may generate additional learning content based on the respondent's level of comprehension for each topic in the test questions. Specifically, the learning optimization unit 115 may extract learning materials, including quality control documents, related to topics for which the respondent tends to have a low level of comprehension, and output the extracted learning content as additional learning content. In this case, the learning optimization unit 115 may identify relevant portions of the quality control documents or learning materials and output the identified portions as additional learning content. Furthermore, when outputting multiple pieces of additional learning content, the learning optimization unit 115 may output a learning path including the order in which the respondent will study the additional learning content based on the respondent's level of comprehension. The learning optimization unit 115 may generate additional learning content using a trained model 4. The trained model 4 used by the learning optimization unit 115 may be a large-scale language model trained by machine learning to generate and output additional learning content in response to input of the test questions answered by the respondent, the evaluation results of the respondent's level of comprehension, and information on the respondent's attributes.

[0117] An example of additional learning content output by learning optimization unit 115 will be described with reference to Fig. 15. Fig. 15 is a schematic diagram showing an example of additional learning content display screen 70 that displays additional learning content for the respondent.

[0118] The additional learning content display screen 70 displays an overview display section 71, an additional learning content display section 72, and a learning path display section 73.

[0119] The summary display section 71 displays a summary of the additional learning content. In the example shown in FIG. 15, the summary display section 71 displays "Supplementary learning materials useful for strengthening weak areas" as a summary of the additional learning content. The additional learning content display section 72 displays the content of the additional learning content proposed to the respondent. Specifically, as shown in FIG. 15, the additional learning content display section 72 displays the title, summary, required time, relevance to the respondent's weak areas, and a start button 721 of the additional learning content. In the example shown in FIG. 15, three additional learning contents are displayed: "Basic flow for processing nonconforming products," "Management and storage of quality records," and "Responsibilities and authority in a QMS system." Of these, it is displayed that the additional learning contents "Basic flow for processing nonconforming products" and "Management and storage of quality records" are highly relevant to the respondent's weak areas. By pressing the start button 722, the additional learning content is displayed on the display section 16, and the respondent can begin studying the additional learning content.

[0120] The learning path display section 73 displays a learning path for the additional learning content based on the respondent's level of understanding. Specifically, the learning path display section 73 displays the required time for each additional learning content and the order in which to start. By checking the learning path display section 73, the respondent can easily understand the content and required time for the additional content. When the learning path start button 74 is pressed, learning of the additional learning content automatically begins according to the learning path.

[0121] FIG. 16 shows an event dashboard 80 that displays the progress of multiple education and training sessions monitored by the monitoring processing unit 106.

[0122] As shown in FIG. 16, an event dashboard 80 displays a progress display section 81, a search bar 82, an education / training list display section 83, and a new education / training registration button 84.

[0123] The number of ongoing trainings, the number of employees whose training has not been completed, and the number of expired trainings are displayed in the progress display section 81. In the example shown in Fig. 16, there are two ongoing trainings, seven employees whose training has not been completed, and zero expired trainings.

[0124] The training list display section 83 displays the training title, start date, deadline, status, and attendance status indicating the number of employees who have completed the training. The search bar 82 is a bar for entering keywords to search for training. The new training registration button 84 is a button for registering new training.

[0125] Next, the progress status of each action in the quality process of education management that is monitored and output by the monitoring processing unit 106 will be described with reference to Fig. 17. Fig. 17 is a schematic diagram showing an example of a progress display screen 90 that displays the progress status of each action in the quality process of education management.

[0126] 17 displays the actions and deadlines for the quality process of training management that will occur when event information for safety training for new employees is received. The progress display screen 90 displays a status display section 91, a basic information display section 92, an overall progress display section 93, and an action list display section 94.

[0127] The basic information display section 92 displays the start date, deadline, alert date, etc. of the training, which is an action of the quality process. The overall progress display section 93 displays the progress of the training of all new employees who need to receive training.

[0128] The name of the material used for each action in the education and training, the type of action, details, and progress rate are displayed in the action list display section 94. In the example shown in Fig. 17, the types of actions displayed are "Test" for solving test questions, "Document" for reading documents such as procedures and manuals, "External Link" for taking an external course, and "Task" for performing a specified task in the workplace.

[0129] Next, an example of the flow of the quality process management process executed by the quality process management system 1 will be described with reference to Fig. 18. Fig. 18 is a flowchart showing the flow of the quality process management process.

[0130] 18, in step S1, the event reception unit 101 receives event information 2. For example, the event reception unit 101 receives the following complaint information as event information 2 via email from a user's in-house system. "Product: RT-500n ventilator Complaint: 'After turning it on, the alarm suddenly stopped working, and I didn't notice it while using it. It was used for about three hours. No patient harm.'"

[0131] In step S2, the classification unit 103 and the action generation unit 104 analyze the event information 2 received in step S1 using the trained model 4. For example, the classification unit 103 divides the text data of the event information 2 into semantic chunks, vectorizes them, and identifies the document content including context and nuances through natural language understanding using the trained model 4. For example, the classification unit 103 identifies the following information from the event information 2: "Complaint information: Alarm malfunction (confidence: 92%), importance rating: high (patient safety risk), urgency: medium-high (no health hazard but potentially serious)"

[0132] In step S3, the similar event search unit 107 searches the database 3 for similar event information related to the event information 2 received in step S1 using the trained model 4. For example, the action generation unit 104 searches for and extracts the following information as similar event information: "There have been seven similar complaints about the same model in the past 12 months. Five of them were about faulty alarm systems. 83% of these occurred within the first 2-4 hours of use."

[0133] In step S4, the action generator 104 estimates root cause candidates for the event information 2 using the trained model 4. For example, the action generator 104 identifies the following root cause candidates: "Predicted causes (in order of probability): Control software bug when switching to battery power (78%), poor contact in the alarm circuit (15%), operational error (7%)."

[0134] In step S5, the action generation unit 104 generates countermeasures as actions for the root cause candidates identified in step S4, using the trained model 4. For example, the action generation unit 104 generates the following corrective measures and preventive measures: "Recommended corrective action: Apply software v2.3.1 update patch, recall affected products. Recommended preventive action: Design redundancy into the alarm system, add precautions to the user manual."

[0135] In step S6, the setting unit 105 sets a workflow for executing the action generated in step S5. The setting unit 105 may set a deadline, a person in charge, and the like for each action based on, for example, the workflow indicated by the quality process history of the similar event information, and the organizational information in the database 3.

[0136] In step S7, if the actions included in the workflow set in step S6 include education and training, the education management unit 110 generates educational content for education and training.

[0137] In step S8, the monitoring processing unit 106 generates a dashboard for managing the progress of the workflow set in step S6, and monitors the progress of the workflow.

[0138] In step S9, the monitoring processing unit 106 evaluates the effectiveness of the action executed after a predetermined period has elapsed since all actions in the workflow generated in step S6 were completed. For example, the monitoring processing unit 106 may evaluate the effectiveness of the action by identifying the number of occurrences of cases similar to the case indicated by the event information 2 received in step S1 and identifying the recurrence rate after a predetermined period has elapsed since all actions in the workflow generated in step S6 were completed. For example, the monitoring processing unit 106 may generate an improvement plan if the effectiveness of the action is lower than a predetermined standard.

[0139] In step S10, the storage processing unit 102 stores the effect of the action evaluated in step S9 together with the event information 2 received in step S1, its root cause candidates, countermeasures, etc. as a quality process history in the database 3. Thereafter, the quality process management processing is terminated.

[0140] According to the embodiment described above, the following effects are achieved.

[0141] (1) The quality process management system 1 of this embodiment includes a storage processing unit 102 that stores quality control information related to quality control and past quality control activity records in a database 3, an event reception unit 101 that receives event information 2 that triggers a quality process related to quality control, an action generation unit 104 that analyzes the event information 2 received by the event reception unit 101 using a trained model 4 that analyzes the event information 2 based on the information in the database 3 and generates actions required in the quality process caused by the event information 2, a setting unit 105 that sets a workflow for executing the action, and a monitoring processing unit 106 that monitors the progress of the set workflow and displays the progress.

[0142] For example, in a QMS, when an event such as a product complaint occurs, product quality and safety are maintained by implementing various quality processes such as complaint management, CAPA, change management, and education management. However, various events occur in the course of daily business activities, and it can become difficult to set up and manage appropriate quality processes because companies are so busy dealing with each and every event.

[0143] According to this embodiment, the actions required in the quality process are generated using a trained model 4 that analyzes event information 2 based on quality control information in a database 3, past quality control activity records, etc., so that an appropriate workflow can be efficiently generated when event information is received. In addition, the progress of the set workflow is monitored and displayed, allowing the user to efficiently manage the progress of the quality process.

[0144] (2) In the quality process management system 1 described in (1), if the event information 2 received by the event receiving unit 101 is information that triggers education management or if an action related to education management exists in the set workflow, the system is provided with an education management unit 110 that performs processing for education and training, and the education management unit 110 has a specified information receiving unit 112 that receives specified information regarding the type and difficulty of test questions for education and training, a test generation unit 113 that generates test questions based on the quality control information and the specified information, a comprehension evaluation unit 114 that evaluates the level of understanding of the respondent of the test questions based on the answers to the test questions, and a learning optimization unit 115 that generates personalized additional learning content based on the comprehension evaluation results.

[0145] The system allows you to set the type and difficulty of test questions, evaluate the respondent's level of understanding, and generate additional learning content appropriate for each respondent, allowing you to educate employees efficiently. Furthermore, if the workflow includes an action related to training management, test questions for training are automatically generated, eliminating the need for laborious training processes.

[0146] (3) The quality process management system 1 described in (1) or (2) includes a classification unit 103 that analyzes the event information 2 using natural language processing and classifies the type of quality process that the event information 2 generates, a setting unit 105 that generates the person in charge of the action and a deadline based on the past quality control activity records in the database 3, and sets a workflow based on the type of quality process, and a monitoring processing unit 106 that notifies the person in charge and manages the deadline for each action.

[0147] For example, it can efficiently generate the person in charge of an action and the deadline based on the records of past quality control activities, and can set an appropriate workflow according to the type of quality process, such as CAPA, defined in the QMS.In addition, notifications to the person in charge and deadline management are also performed automatically, reducing the burden of managing the progress of the quality process.

[0148] (4) The quality process management system 1 described in any one of (1) to (3) includes a similar event search unit 107 that searches past event information 2 similar to the currently accepted event information 2 from past quality control activity records in the database 3 and displays countermeasures for the problem indicated by the searched event information as reference information.

[0149] Past event information 2 similar to the current event information 2 and the corresponding countermeasures are displayed, so that the user can select an action corresponding to the current event information 2 while taking into consideration the past countermeasures.

[0150] (5) In the quality process management system 1 described in any one of (1) to (4), the event information 2 is information described in a document, and the trained model 4 identifies candidate root causes of the problem indicated by the event information 2 based on the description of the event information 2, and generates corrective measures and preventive measures for the identified candidate root causes.

[0151] By using a trained model 4 that analyzes event information 2 based on quality control information, past quality control activity records, etc., it is possible to identify more accurate root cause candidates and generate more appropriate corrective and preventive measures.

[0152] (6) In the quality process control system 1 described in (5), the action generation unit 104 predicts the effectiveness of the generated corrective and preventive measures based on records of similar past quality control activities in the database 3.

[0153] Since the effectiveness of corrective and preventive measures can be predicted based on past quality control activity records, more effective measures can be taken against the problem indicated by the event information 2.

[0154] (7) The quality process management system 1 described in any one of (1) to (6) includes a time series analysis unit 108 that determines the increase / decrease trend of the event indicated by the event information 2 through time series analysis of the event information 2 that is similar to the currently received event information 2, and outputs the determined increase / decrease trend.

[0155] The increase or decrease in the number of events related to the current event information 2 can be easily understood from the accumulated past similar event information 2.

[0156] (8) In the quality process management system 1 according to this embodiment, the training management unit 110 has a necessity determination unit 111 that, when an action that changes a quality control document exists in the workflow, determines whether training is necessary related to the quality control document that will be changed by executing the action, and the action generation unit 104 generates an action for training when the necessity determination unit 111 determines that training is necessary.

[0157] The system automatically determines whether education and training is necessary and generates actions for education and training as necessary, thereby reducing the burden on the quality process and enabling it to be executed efficiently.

[0158] (9) The program causes the computer 18 to execute processes including a storage process for storing quality control information and past activity records related to quality control in a database; an event reception process for receiving event information 2 that triggers a quality process related to quality control; an action generation process for analyzing the event information 2 received in the event reception process using a trained model 4 that analyzes the event information 2 based on information in the database 3 and generating actions required in the quality process triggered by the event information 2; a setting process for setting a workflow for executing the action; and a monitoring process for monitoring the progress of the set workflow and displaying the progress.

[0159] (10) The quality process management method includes a storage process for storing quality control information and past activity records related to quality control in a database 3, an event reception process for receiving event information 2 that triggers a quality process related to quality control, an action generation process for analyzing the event information 2 received in the event reception process using a trained model 4 that analyzes the event information 2 based on the information in the database 3 and generating actions required in the quality process caused by the event information 2, a setting process for setting a workflow for executing the action, and a monitoring process for monitoring the progress of the set workflow and displaying the progress.

[0160] Although the embodiments of the present invention have been described above, the present invention is not limited to the above embodiments and can be modified as appropriate.

[0161] In the above embodiment, the education management unit 110 performed processing for education and training when the event information 2 received by the event reception unit 101 was information that triggered education management or when an action related to education management existed in the set workflow, but the education and training processing may also be performed regardless of the contents of the event information 2 and the workflow.

[0162] Furthermore, for example, the quality process control system 1 may perform root cause analysis, impact analysis, analysis using performance indicators, correlation analysis, geographic analysis, risk analysis, resource allocation analysis, regulatory compliance analysis, predictive analysis, and comparative analysis as processes for analyzing the event information 2 and the quality process history. Details of each analysis will be described below using specific examples.

[0163] Root cause analysis is a technique for identifying and classifying common factors behind an event and focusing resources on the most important causes based on the Pareto principle. For example, the action generation unit 104 may identify the proportion of root causes in each element that constitutes an event and generate an action based on the identified proportion of root causes. For example, the action generation unit 104 may identify the proportion of root causes from a "prosthetic joint device defect report" as 45% for the manufacturing process, 30% for the design, 15% for materials, and 10% for the instruction manual, based on the complaint information. If the action generation unit 104 identifies that sterilization defects account for the largest proportion of root causes in the manufacturing process at 60%, it may generate an action to prioritize resources toward strengthening sterilization validation. Implementing this action may potentially reduce sterilization-related defects.

[0164] Impact analysis is a method for quantifying the severity of an event, its impact on product quality and safety, its impact on regulatory requirements, financial losses, and the like, and identifying issues that require priority attention. For example, the action generation unit 104 may identify information including at least one of the severity of the event indicated by the event information 2, its impact on product quality and safety, and compliance with regulatory requirements, and generate corrective actions as actions based on the identified information. The monitoring processing unit 106 then predicts and compares the financial impact of the corrective actions with the risk of regulatory requirements (hereinafter referred to as regulatory risk) and additional losses due to a delay in the corrective actions, and outputs the comparison results. The comparison results may be displayed on the display unit 16, or the organizational information may be sent to the email address of the organization's management. For example, the action generation unit 104 may stratify the measurement error of a blood glucose meter, which is the event information 2, into error ranges such as ±5%, ±10%, ±20%, and more, and evaluate the clinical impact of each range on patient safety. If the action generation unit 104 determines that an error of ±20% or more poses a risk of serious medication error, the action generation unit 104 may generate a recall of all applicable lots as an action.The monitoring processing unit 106 may then predict and compare the financial impact of market corrective action (e.g., $1.2 million) with the regulatory risk and additional losses (e.g., $4.5 million) due to delayed action, and present the comparison results along with the justification for prompt action to management.

[0165] Analysis using performance indicators is a method for measuring and tracking the efficiency and effectiveness of a problem-solving process, such as the mean time to resolution, the effectiveness of corrective actions, and the recurrence rate. For example, the action generation unit 104 may analyze the quality process history to identify the effectiveness of corrective actions as actions, the mean time to complete each action or quality process, and the recurrence rate of similar events, and set key performance indicators (KPIs) for each action and quality process based on the identified results. For example, if the average number of days from problem detection to root cause identification in a past ventilator software bug fix (hereinafter referred to as a bug fix case) is seven days, the action generation unit 104 may set the KPI related to the number of days from problem detection to root cause identification to four days. Furthermore, if the number of days until fix release in a bug fix case is 30 days and the number of days until validation completion is 14 days, the action generation unit 104 may set the KPI related to the number of days until fix release to 18 days and the KPI related to the number of days until validation completion to 10 days. The action generation unit 104 may then notify the set KPIs to multiple related departments. For example, by optimizing the collaborative process between different departments, such as the development department and the QA (Quality Assurance) department, it is expected that the speed of responding to product defects will improve. This will, for example, improve evaluations from regulatory authorities and improve the effectiveness of CAPA, such as by reducing the recurrence rate of similar problems.

[0166] Correlation analysis is a technique for analyzing the relationships between multiple elements, such as events, products, batches, and suppliers, to reveal hidden patterns and causal relationships. For example, the action generation unit 104 may identify root causes by analyzing the relationships between multiple types of elements related to the events indicated by the event information 2, and generate actions such as corrective measures and preventive measures based on the identified root causes. For example, the action generation unit 104 may analyze the correlation between image artifact defects in MRI systems and multiple elements, such as electronic components from a specific supplier, maintenance work by a specific technician, and power supply stability at a specific facility. Then, for example, if the correlation coefficient between power supply fluctuations and image noise is higher than a predetermined value, the action generation unit 104 may generate an action that prioritizes the introduction of a UPS. This can be expected to reduce the number of image quality complaints at the target facility. Furthermore, the action generation unit 104 may generate an action to switch to an alternative supplier if a common failure pattern exists among a group of products using parts from a certain supplier.

[0167] Geographic analysis is a method for analyzing the distribution of events by region, country, or facility to understand the impact of regional issues and differences in regulatory environments. For example, the action generation unit 104 may analyze identical or similar events occurring in multiple different regions by region, identify the root causes of the events in each region, and generate corrective and preventive measures based on the identified root causes. For example, the action generation unit 104 may analyze hemodialysis machine malfunctions by region. If the water treatment module failure rate in Southeast Asia is higher than in other regions by a predetermined value or more, the action generation unit 104 may generate an action to investigate the local water quality and temperature and humidity conditions. If the cause is microbial growth in a high-temperature, high-humidity environment, the action generation unit 104 may generate an action to develop water treatment filter specifications adapted to the region's unique environment. This is expected to reduce the product failure rate in that region.

[0168] Risk analysis is a method for evaluating the distribution and changes in risk levels of events, verifying the effectiveness of risk mitigation measures, and grasping the overall picture using a risk matrix or heat map. For example, the action generation unit 104 may evaluate the risk of an event indicated by the event information 2 and generate corrective or preventive measures for the event based on the evaluation results. For example, in a risk assessment of a surgical robot system, the action generation unit 104 may generate an action that maps incidents that have occurred by severity and occurrence frequency. The action generation unit 104 may perform a Failure Mode and Effects Analysis (FMEA) for both position control errors (high frequency, medium impact) and communication errors (low frequency, severe impact) and generate an action that implements countermeasures based on a Risk Priority Number (RPN) that also takes detectability into account. Implementing this action is expected to reduce the number of high-risk items (RPN>100) and the incidence rate of serious incidents after launch. The action generation unit 104 may also generate an action to revise a training program for surgeons. By implementing this action, it is expected that the number of occurrences of risks related to operational errors will be reduced.

[0169] Resource allocation analysis is a method for achieving optimal resource allocation by analyzing departmental response status, resource utilization efficiency, and quality cost structure. For example, the action generation unit 104 may refer to quality control activity records and generate an action for an event indicated by event information 2 based on the effectiveness of actions such as corrective and preventive measures for the event indicated by similar event information and resource information such as human resources and funds used to execute those actions. For example, the action generation unit 104 may analyze the input resources and quality improvement effects for each department in responding to a quality issue with a biological monitoring device. For example, the action generation unit 104 may determine that strengthening user training is the most cost-effective option based on the analysis results that show a 30% reduction in defects in the design department with 40 employees, a 25% reduction in defects in the manufacturing department with 60 employees, a 10% reduction in defects in the quality department with 30 employees, and a 35% reduction in defects in the training department with 20 employees. Then, the action generation unit 104 may generate an action to increase investment in a department or activity if the cost-effectiveness is high. This is expected to reduce costs associated with product recalls and complaint handling due to product defects.

[0170] Regulatory compliance analysis is a method for reducing compliance risk by evaluating the rate of reportable events, compliance with regulatory timelines, and the quality of submitted documents. For example, the monitoring processor 106 may evaluate compliance with rule information for past events based on quality control activity records and output the results. For example, for orthopedic implant defect reports, the monitoring processor 106 may determine that 12% of EUDAMED (European Database on Medical Devices) reports are late based on the regional reporting deadline compliance rate and output the determination result. Based on the determination result, the action generator 104 may determine that the root cause is the absence of a responsible person for the reporting process and unclear judgment criteria, and generate actions such as redesigning the reporting flow and introducing an automatic alert system. Executing such actions can be expected to improve the reporting deadline compliance rate, for example. For example, the monitoring processor 106 may monitor the progress of corrective measures for items identified in inspections by the Food and Drug Administration (FDA) or other authorities.

[0171] Predictive analysis is an advanced analytical technique that identifies early warning indicators from historical data and uses statistical techniques to predict future event occurrences and risks. For example, the action generation unit 104 may have the trained model 4 analyze past events from quality control activity records and predict patterns that are precursors to those events. For example, the action generation unit 104 may build a machine learning model from defibrillator battery consumption data to identify patterns that are precursors to failure. The action generation unit 104 may generate a trained model to predict patterns that are precursors to failure using variables such as frequency of use, number of charging cycles, and temperature history. This enables highly accurate detection of precursors to failure. Introducing a preventive maintenance program using this trained model can reduce the incident rate due to sudden battery failure, potentially reducing both patient risks and maintenance costs.

[0172] Comparative analysis is a method for evaluating relative performance and identifying best practices through comparisons between product lines, processes, facilities, and industry standards. For example, the quality process control system 1 may include a product improvement department that compares the types and frequencies of defects between different generations of products that have experienced defects based on quality control activity records and identifies areas for improvement. For example, the product improvement department may compare and analyze the types and frequencies of defects between different generations of endoscope systems, quantitatively evaluate the advantages of new designs, and output the results of the quantitative evaluation. A quantitative evaluation result may show that optical system problems were a major issue in the previous generation, but that optical system problems in the new generation have decreased by 12% compared to the previous generation, while software-related defects have increased from 9% to 23%. For example, the product improvement department may compare quality control activity records with industry standards and output the comparison results. A comparison result may show that the average recall rate for a company's products is 0.7%, compared to the industry average of 1.2%, but that a specific product category within the company's products is inferior at 1.8%. [Explanation of symbols]

[0173] 1. Quality Process Control System 2. Event Information 3 Database 4. Pre-trained model 18 Computer 101 Event Reception 102 Memory Processing Unit 104 Action Generation Unit 105 Settings 106 Monitoring processing unit

Claims

1. A quality process control system for medical devices or pharmaceuticals, comprising: a storage processing unit that stores quality control information and past activity records related to quality control of the medical device or the pharmaceutical in a database; an event reception unit that receives event information regarding a complaint about the medical device or the pharmaceutical, which is a trigger for a quality process related to the quality control; an action generation unit that analyzes the event information received by the event reception unit using a trained model that analyzes the event information based on information in the database, and generates an action required in a quality process caused by the event information; a setting unit that sets a workflow for executing the action; a monitoring processing unit that monitors the progress of the set workflow and displays the progress, The monitoring processing unit obtaining the number of occurrences of cases similar to the case indicated by the event information during a predetermined period after the action in the workflow is completed; A quality process management system that evaluates the effectiveness of the action based on the acquired number of occurrences, and the effectiveness of the action is compared with a predetermined reference value to provide an indicator of whether or not an improvement plan will be generated for the action.

2. an education management unit that performs processing for education and training when the event information received by the event reception unit is information that triggers a quality process for education management or when an action related to the education management exists in the set workflow; The education management department a designated information receiving unit that receives designated information regarding the type and difficulty of test questions for education and training; a test generation unit that generates the test questions based on the quality control information and the specification information; a comprehension evaluation unit that evaluates the comprehension of the respondent of the test question based on the answer to the test question; The quality process management system according to claim 1 , further comprising: a learning optimization unit that generates personalized additional learning content based on the results of the assessment of the level of understanding.

3. a classification unit that analyzes the event information by natural language processing and classifies the type of quality process caused by the event information; the setting unit generates a person in charge and a deadline for the action based on the information in the database, and sets the workflow based on the type of the quality process; The quality process control system according to claim 1 , wherein the monitoring processing unit notifies the person in charge and manages deadlines for each action.

4. 2. The quality process control system according to claim 1, further comprising a similar event search unit that searches the past activity records in the database for past event information similar to currently received event information, and displays countermeasures for problems indicated by the searched event information as reference information.

5. Event information is information contained in a document, The trained model is Identifying potential root causes of the problem indicated by the event information based on a description of the event information; The quality process control system of claim 1 , wherein corrective and preventative actions are generated for the identified potential root causes.

6. The action generation unit 6. The quality process control system of claim 5, wherein the effectiveness of the generated corrective and preventive actions is predicted based on the past activity records in the database.

7. 2. The quality process control system according to claim 1, further comprising a time series analysis unit that determines an increase or decrease trend of an event indicated by the event information through time series analysis of past event information similar to the event information, and sets a priority of the action based on the determined increase or decrease trend.

8. the training management unit includes a necessity determination unit that, when an action that changes the quality control information exists in the workflow, determines whether or not training is necessary related to the quality control information that is changed by executing the action; The quality process management system according to claim 2 , wherein the action generation unit generates an action for education and training when the necessity determination unit determines that education and training is necessary.

9. A storage process step of storing quality control information and past activity records related to quality control of medical devices or pharmaceuticals in a database; an event reception step of receiving event information related to a complaint about the medical device or the pharmaceutical product, which will trigger a quality process related to the quality control; an action generation step of analyzing the event information received in the event reception step using a trained model that analyzes the event information based on information in the database, and generating an action required in a quality process caused by the event information; a setting step of setting a workflow for executing the action; a monitoring process step of monitoring the progress of the set workflow and displaying the progress, acquiring the number of occurrences of cases similar to the case indicated by the event information during a predetermined period after the action in the workflow is completed; a step of evaluating the effectiveness of the action based on the acquired number of occurrences, the effectiveness of the action being an index indicating whether or not an improvement plan for the action is generated by comparing it with a predetermined reference value; and a monitoring process including:

10. A storage process step of storing quality control information and past activity records related to quality control of medical devices or pharmaceuticals in a database; an event reception step of receiving event information related to a complaint about the medical device or the pharmaceutical product, which will trigger a quality process related to the quality control; an action generation step of analyzing the event information received in the event reception step using a trained model that analyzes the event information based on information in the database, and generating an action required in a quality process caused by the event information; a setting step of setting a workflow for executing the action; a monitoring process step of monitoring the progress of the set workflow and displaying the progress, acquiring the number of occurrences of cases similar to the case indicated by the event information during a predetermined period after the action in the workflow is completed; a step of evaluating the effectiveness of the action based on the acquired number of occurrences, the effectiveness of the action being an index indicating whether or not an improvement plan for the action is generated by comparing it with a predetermined reference value; and and a computer-implemented quality process control method comprising:

Images