Information processing system

Abstract

The application provides an information processing system. An information processing system, characterized by comprising: a processor; wherein the processor is configured to: receive input of health information and emotional state of a person; analyze the health information and emotional state of the person, generate prompt information indicating generation of diet and drug recommendations from a preventive medicine perspective; provide the generated recommendations to the person and link with an external food delivery service to make ordering possible; generate prompt information indicating adjustment of alcohol intake limit according to the emotional state of the person, and adjust the alcohol intake limit using the prompt information.

Description

Technical Field

[0001] The technology disclosed herein relates to an information processing system. Background Technology

[0002] Japanese Patent Application Publication No. 2022-180282 discloses a method for controlling a role-based chatbot executed by at least one processor. The method includes the following steps: receiving a user's speech; adding the user's speech to a prompt word, the prompt word containing instruction statements associated with an explanation of the chatbot's role; encoding the prompt word; and inputting the encoded prompt word into a language model to generate a chatbot response to the user's speech.

[0003] Existing health management and preventive medicine support systems often rely solely on an individual's physiological or medical-related health information for analysis and recommendations, neglecting the significant impact of an individual's emotional state on eating habits, medication adherence, and alcohol consumption. Emotional fluctuations (such as anxiety, depression, excitement, or excessive stress) can easily lead to overeating, inappropriate alcohol consumption, or unauthorized adjustments to medication dosage, thereby weakening the effectiveness of preventive medicine recommendations and even triggering new health risks.

[0004] On the other hand, existing systems typically only provide static dietary or medication recommendations, lacking effective integration with external food delivery services. After receiving the recommendations, users still need to search, select, and place orders themselves, which is cumbersome and costly, resulting in a low actual implementation rate of the recommendations.

[0005] Furthermore, current technologies for managing alcohol intake often involve fixed restrictions or broad guidelines, failing to dynamically adjust based on the user's current emotional state. This can lead to either excessive restrictions on the user's quality of life when their emotions are relatively stable, or failure to tighten restrictions in a timely manner when the user is depressed or under excessive stress, thereby increasing the risk of alcohol abuse.

[0006] Therefore, it is necessary to provide a system that can comprehensively consider individual health information and emotional state, automatically generate dietary and medication recommendations from a preventive medicine perspective, and can be linked with external food delivery services. At the same time, it can dynamically adjust alcohol intake restrictions based on emotional state to improve the personalization and feasibility of recommendations, thereby more effectively supporting individuals' long-term health management. Summary of the Invention

[0007] To address the aforementioned technical problems, the present invention provides an information processing system comprising a processor; the processor is configured to receive input of an individual's health information and emotional state, comprehensively analyze the individual's health information and emotional state, generate personalized recommendations for diet and medication from a preventive medicine perspective, and provide the recommendations to the individual in a form that is easy for the user to understand and implement.

[0008] To reduce the difficulty for users to follow dietary recommendations and improve their feasibility, the processor is also configured to work in conjunction with external food delivery services: after generating dietary recommendations or suitable food information that are appropriate for the individual's current health and emotional state, it automatically maps the food information to the corresponding dishes or products in the external food delivery service, generating options for direct ordering, so that users can complete the entire process from receiving recommendations to placing orders within the same system.

[0009] To address the influence of emotional state on alcohol consumption behavior, the processor is further configured to generate prompts based on the individual's emotional state to instruct adjustments to alcohol intake limits, and to dynamically adjust the individual's alcohol intake cap or related control strategies using these prompts. For example, when the system detects that the individual is under high stress, experiencing low mood, or is prone to impulsivity, it guides the tightening of alcohol intake limits through prompts; when the system detects that the individual's emotions are relatively stable and there are no obvious high-risk signals, the limits can be appropriately relaxed under safe conditions, thereby achieving a balance between safety and quality of life.

[0010] Furthermore, to further enhance the overall health management effect, the processor can also generate stress management suggestions based on individual health information and emotional state, such as relaxation training, sleep hygiene guidance, or appropriate exercise arrangements, and combine them with dietary, medication, and alcohol intake control strategies to form a closed-loop, dynamically adjusted personalized health management plan. Through the above technical means, this invention overcomes the problems of neglecting emotional factors, broken execution chains, and inadequate alcohol intake control in existing technologies, and more effectively supports individual preventive medicine practices.

[0011] "System" refers to an overall device or collection of devices consisting of one or more hardware and / or software components, used to perform comprehensive processing functions such as receiving information, parsing information, generating suggestions, and interacting with external services.

[0012] A "processor" is a computing unit that can execute program instructions to complete processing tasks such as data reception, data parsing, logical judgment, prompt message generation, suggestion generation, and interaction with external services. It can be a single physical processor, a combination of multiple processors, or any processing resource including CPU, GPU, application-specific integrated circuits, etc.

[0013] "Personal health information" refers to information related to a specific individual's physical condition, medical condition, or lifestyle, including but not limited to age, gender, weight, height, past medical history, medication use, allergy history, family genetic information, lifestyle habits, and physiological indicators.

[0014] "Emotional state" refers to information indicating an individual's psychological or emotional state during a specific period of time, including but not limited to states such as anxiety, depression, pleasure, anger, tension, stress level, and excitement level. It can be obtained through self-report, scale assessment, physiological signal analysis, or other detection methods.

[0015] "Analysis" refers to the process of processing and analyzing received personal health information and emotional state, including but not limited to data cleaning, feature extraction, pattern recognition, risk assessment, rule matching, and model reasoning, in order to generate subsequent suggestions and prompts.

[0016] "Preventive medicine" refers to the medical field aimed at preventing the occurrence and development of diseases, promoting health, and slowing the progression of diseases. It includes disease risk assessment, lifestyle intervention, nutrition management, preventive drug use, and health education.

[0017] "Dietary recommendations" refer to recommendations given from a preventive medicine perspective based on individual health information and emotional state regarding food types, nutritional structure, intake, meal times, and eating habits, including foods that should be prioritized or restricted and their combinations.

[0018] "Drug advice" refers to drug-related guidance provided from the perspectives of preventive medicine and rational drug use, based on individual health information and emotional state. This includes, but is not limited to, medication reminders, precautions, tips to avoid self-adjustment of dosage, and recommendations for medical consultation.

[0019] "Prompt messages" refer to messages or instructions generated by the processor to instruct or guide subsequent processing or user behavior, including internal prompts to instruct the generation of dietary and medication recommendations, and control prompts to instruct the adjustment of alcohol intake limits.

[0020] "Generating suggestions" refers to the process by which a processor, after analyzing an individual's health information and emotional state, and based on preventive medicine knowledge, rule bases, and / or model outputs, generates specific and actionable dietary, medication, stress management, or other health intervention plans.

[0021] "Provided to individuals" means that the suggestions generated by the processor are displayed or conveyed to the corresponding individual through a user terminal interface, notification, report or other information presentation methods, so that the individual can know about them and choose whether to adopt and implement them.

[0022] "External food delivery services" refers to third-party service platforms or systems that are independent of this system and can provide functions such as online food selection, ordering, payment and delivery, including but not limited to food delivery platforms, online catering ordering systems and related interface services.

[0023] "Linkage" refers to the process by which this system works collaboratively with external food delivery services through interfaces, protocols, or data interaction. This includes converting generated food information into orderable product options, sending order requests to external services, and receiving order status feedback.

[0024] "Placing an order" refers to the process of generating a formal order for a specific food or meal in an external food delivery service, which includes steps such as selecting the product, determining the quantity, confirming delivery information, and submitting the order request.

[0025] "Alcohol intake restriction" refers to the constraints or control strategies that restrict an individual's total alcohol intake, frequency of drinking, occasions of drinking, or related behaviors within a certain period of time. It can take the form of quantitative restrictions, frequency restrictions, or prohibition of drinking.

[0026] "Adjusting alcohol intake limits" refers to the process of increasing, decreasing, relaxing, or tightening established alcohol intake limits, drinking frequency, or related control rules based on an individual's current emotional state and health condition.

[0027] "Stress management advice" refers to guidance provided based on an individual's health information and emotional state to reduce or alleviate psychological stress, including but not limited to relaxation training, breathing exercises, sleep improvement suggestions, exercise stress reduction programs, and tips for seeking professional psychological help.

[0028] "Suitable food information" refers to a list of foods or dishes that meet preventive medicine and current health goals, as well as their related descriptions, which are filtered or generated by the processor based on personal health information and emotional state. This list includes the food name, type, main nutritional components, and applicable scenarios, and can be used for matching and ordering in external food delivery services. Attached Figure Description

[0029] Figure 1 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the first embodiment.

[0030] Figure 2 This is a conceptual diagram illustrating an example of the main functions of the data processing apparatus and smart device according to the first embodiment.

[0031] Figure 3 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the second embodiment.

[0032] Figure 4 This is a conceptual diagram illustrating an example of the main functions of the data processing device and smart glasses according to the second embodiment.

[0033] Figure 5 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the third embodiment.

[0034] Figure 6 This is a conceptual diagram illustrating an example of the main functions of the data processing apparatus and head-mounted terminal according to the third embodiment.

[0035] Figure 7 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the fourth embodiment.

[0036] Figure 8 This is a conceptual diagram illustrating an example of the main functions of the data processing device and robot according to the fourth embodiment.

[0037] Figure 9 This represents an emotion map that maps multiple emotions.

[0038] Figure 10 This represents an emotion map that maps multiple emotions.

[0039] Figure 11 This is a sequence diagram illustrating the processing flow of the data processing system of the first embodiment.

[0040] Figure 12 This is a sequence diagram illustrating the processing flow of the data processing system in Application Example 1.

[0041] Figure 13 This is a sequence diagram illustrating the processing flow of the data processing system of the second embodiment.

[0042] Figure 14 This is a sequence diagram illustrating the processing flow of the data processing system in Application Example 2. Detailed Implementation

[0043] Hereinafter, an example of an implementation of the system to which the technology of this disclosure relates will be described with reference to the accompanying drawings.

[0044] First, let me explain the terminology used in the following instructions.

[0045] In the following embodiments, the processor (hereinafter referred to as "processor") with reference numerals may be a single computing device or a combination of multiple computing devices. Furthermore, the processor may be a single computing device or a combination of multiple computing devices. Examples of computing devices include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), APU (Accelerated Processing Unit), etc.

[0046] In the following embodiments, RAM (Random Access Memory), as indicated in the figures, is a memory that temporarily stores information and is used as working memory by the processor.

[0047] In the following embodiments, the memory, as indicated by the reference numerals, is one or more non-volatile storage devices that store various programs and parameters. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), disks (e.g., hard disks), or magnetic tapes.

[0048] In the following embodiments, the communication I / F (Interface) with reference numerals is an interface that includes a communication processor and an antenna, etc. The communication I / F is responsible for communication between multiple computers. As an example of a communication specification applicable to the communication I / F, wireless communication specifications such as 5G (5th Generation Mobile Communication System), Wi-Fi (wireless fidelity) (registered trademark), or Bluetooth (registered trademark) can be listed.

[0049] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B". That is, "A and / or B" means that it can be only A, only B, or a combination of A and B. Furthermore, in this specification, when "and / or" connects to express more than three items, the same interpretation as "A and / or B" applies.

[0050] First Implementation Method Figure 1 An example of the configuration of the data processing system 10 according to the first embodiment is shown.

[0051] like Figure 1 As shown, the data processing system 10 includes a data processing device 12 and an intelligent device 14. A server can be cited as an example of the data processing device 12.

[0052] The data processing apparatus 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" as understood in this disclosure. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, the database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0053] The smart device 14 includes a computer 36, a receiving device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. In addition, the receiving device 38, output device 40, camera 42, and communication I / F 44 are also connected to the bus 52.

[0054] The receiving device 38 includes a touchscreen 38A and a microphone 38B, and receives user input. The touchscreen 38A receives user input via touch by detecting contact with an indicator (e.g., a pen or finger). The microphone 38B receives user input via sound by detecting the user's voice. The control unit 46A in the processor 46 sends data representing the user input received by the touchscreen 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data representing the user input.

[0055] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting data in a form perceptible to the user 20 (e.g., sound and / or text). The display 40A displays visual information such as text and images according to instructions from the processor 46. The speaker 40B outputs sound according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0056] Communication I / F44 is connected to network 54. Communication I / F44 and 26 are responsible for sending and receiving various information between processor 46 and processor 28 via network 54.

[0057] Figure 2 The diagram shows an example of the main functions of the data processing device 12 and the smart device 14.

[0058] like Figure 2 As shown, in the data processing apparatus 12, specific processing is performed by the processor 28. A specific processing program 56 is stored in the memory 32. The specific processing program 56 is an example of a "program" as understood in this disclosure. The processor 28 reads the specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. Specific processing is implemented by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0059] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. In the emotion inference function (emotion-specific function) using the emotion-specific model 59, various inferences and predictions related to the user's emotions are performed, including inferences and predictions of the user's emotions, but this is not limited to this example. Furthermore, emotion inference and prediction may also include, for example, emotion analysis (parsing).

[0060] In the smart device 14, the processor 46 performs the acceptance output processing. The memory 50 stores the acceptance output program 60. The acceptance output program 60 is used in conjunction with the data processing system 10 and the specific processing program 56. The processor 46 reads the acceptance output program 60 from the memory 50 and executes the read acceptance output program 60 on the RAM 48. The acceptance output processing is implemented by the processor 46 acting as the control unit 46A according to the acceptance output program 60 executed on the RAM 48. Furthermore, the smart device 14 has the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and these models can also be used to perform the same processing as the specific processing unit 290. The acceptance output processing is implemented by the processor 46 acting as the control unit 46A according to the acceptance output program 60 executed on the RAM 48.

[0061] Alternatively, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device (e.g., a generation server) may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. Furthermore, the data processing device 12 may be a server device or a user-held terminal device (e.g., a mobile phone, robot, home appliance, etc.). Next, an example of the processing of the data processing system 10 of the first embodiment will be described.

[0062] Example 1 The flow of a specific process in Example 1 will be described. Each part of the system described below is implemented by the data processing device 12 and the smart device 14. Furthermore, the data processing device 12 is referred to as the "server," and the smart device 14 is referred to as the "terminal."

[0063] In recent years, with the increasing demand for individual health management and the development of generative artificial intelligence technology, systems that utilize computing devices to analyze individual health information and generate personalized health recommendations have emerged. However, existing technologies typically only directly input user-generated health data as natural language prompts into generative AI models, lacking structured data preprocessing, statistical analysis, and risk assessment steps, leading to the following technical problems: First, existing systems primarily handle health data on the server side with simple verification and storage, failing to fully utilize data processing components for refined calculation and structured representation of health indicators such as body mass index and risk level. Consequently, generative AI models lack key quantitative features and risk labels in their prompts, resulting in unstable and poorly interpretable outputs, making it difficult to guarantee the relevance and consistency of the recommendations.

[0064] Second, existing systems often use direct splicing of users' original text or simple templates to construct prompts, without systematically designing the structure and constraining the output format of the prompts based on statistical analysis results, risk assessment results, body mass index classification, and quantitative information on lifestyle habits. As a result, the input quality of generative artificial intelligence models cannot be guaranteed, and the health advice output by the models is significantly inadequate in terms of structure, content coverage, and matching with the user's status.

[0065] Third, when existing technologies are linked with external services (such as goods providing services), they mostly just provide simple redirects or recommendation lists. They do not automatically generate linkage information that can be directly used by goods services based on dietary suggestions output by generative artificial intelligence models on the server side. Nor do they organically combine health analysis results, generative artificial intelligence responses, and external ordering operations within the same technical system. This results in a disconnect between the process of "health suggestion generation" and "specific action execution," leading to a poor user experience and insufficient overall system efficiency.

[0066] Fourth, existing systems typically treat emotional state information as secondary information and fail to deeply integrate emotional state with health risk assessment and prompt generation logic on the server side. In particular, they lack dynamic adjustment mechanisms for sensitive behavioral suggestions such as alcohol intake restrictions, and cannot weaken or strengthen relevant suggestions in a timely manner according to the user's emotional state. This poses a risk that the generated content may not match the user's psychological state or even cause negative experiences.

[0067] Fifth, existing technologies lack a long-term closed-loop optimization mechanism for prompt statements and generated results. Most systems do not systematically record the correspondence between historical prompt statements, model outputs, and user evaluation information. They cannot automatically adjust the structure and content of prompt statements through statistical processing on the server side, thus failing to continuously improve the performance of generative AI models in specific health scenarios at the computational level. The technical solutions remain in a one-time configuration state and lack adaptive evolution capabilities.

[0068] Therefore, it is necessary to provide a new technical solution to improve the traditional health management and generative artificial intelligence combination model on the server side from aspects such as computer data structure design, statistical analysis process, prompt statement construction logic, and external service linkage mechanism. This will solve computer technology problems such as low prompt input quality, unstable generation results, weak linkage capability, and lack of self-learning capability, thereby significantly improving the overall performance and user experience of the health management system based on generative artificial intelligence.

[0069] The specific processing performed by the specific processing unit 290 of the data processing apparatus 12 in Embodiment 1 is achieved by the following means.

[0070] In this invention, the server includes: a receiving unit for receiving individual attribute information, medical history information, lifestyle habit information, and emotional state information from a terminal having a display device and an input device via a communication network; a storage management unit for standardizing the attribute information, medical history information, lifestyle habit information, and emotional state information according to an information management structure stored in an information storage device and storing it in a structured form; an analysis unit for calculating body mass index and other health indicators based on the stored information using statistical processing programs and data processing components, and generating an individual's health status analysis result through a risk assessment that combines past disease information, family medical history information, and lifestyle habit information; a prompting construction unit for constructing prompt statements for inputting into a generative artificial intelligence model based on the health status analysis result and the emotional state information, using character sequence generation logic or template processing logic, and embedding risk assessment results, body mass index classification, and quantitative information related to lifestyle habits in the prompt statements, while also including constraint conditions and output format specification information to constrain the output of the generative artificial intelligence model; and a prompting construction unit for sending the prompt statements to an external information processing device that provides the generative artificial intelligence model, so that the external information processing device can use preventive medicine knowledge... The system includes: a model interaction unit that generates response text containing dietary advice, medication-related advice, stress management advice, alcohol intake restriction guidelines, and exercise guidelines, and obtains the response text from an external information processing device; a service linkage unit that generates linkage information indicating food selection candidates and order information based on the dietary advice contained in the response text, and sends the linkage information to an external information processing device providing the item service, thereby enabling the individual to place an order through the item service; an output control unit that outputs the response text and the health status analysis results to the terminal in a visual form, generates additional prompts for adjusting the intensity or frequency of alcohol intake restrictions based on the emotional state information, and dynamically modifies the alcohol-related advice in the response text based on the additional prompts; and a prompt optimization unit that stores historical prompts corresponding to the health status analysis results and the emotional state information, as well as evaluation information provided by the individual for the prompts, in the form of historical records in an information storage device, and updates the content or components of the prompts based on the statistical processing results of the historical records, thereby gradually optimizing the suggestion content generated by the generative artificial intelligence model.This allows for structured processing of health and emotion data on the server side, generating high-quality prompts through refined statistical analysis and risk assessment. This significantly improves the relevance, consistency, and structure of the output of generative AI models, while also enabling automatic interaction with external services and closed-loop optimization of prompts and generated results based on user feedback. This substantially improves the performance of generative AI-based health management systems from the perspectives of computer data processing and system architecture.

[0071] "Terminal" refers to an information processing device that has a display device and an input device and is able to send and receive data with a server through a communication network, including but not limited to smartphones, tablet computing devices, desktop computing devices or other electronic devices with human-computer interaction interfaces.

[0072] "Communication network" refers to wired or wireless network infrastructure used to transmit data between terminals and servers, and between servers and external information processing devices, including but not limited to the Internet, mobile communication networks, local area networks, or combinations thereof.

[0073] "Individual attribute information" refers to a set of data used to represent the basic characteristics of a specific user, including but not limited to basic demographic information related to health status such as age, gender, height, and weight.

[0074] "Medical history information" refers to data related to an individual's past or present medical conditions, including but not limited to information on chronic diseases, acute diseases, surgical history, medication history, and family medical history.

[0075] "Lifestyle information" refers to data that reflects an individual's behavioral patterns in daily life, including but not limited to eating habits, smoking, drinking, exercise frequency, sleep habits, and work and leisure activities.

[0076] "Emotional state information" refers to data used to represent an individual's psychological or emotional state within a specific time period, including but not limited to anxiety level, depressive tendency, stress level, and pleasure level, which can be obtained through self-assessment questionnaires, scales, or other perceptual methods.

[0077] "Information storage device" refers to a storage medium and its control components used to store programs, data structures and data records related to individuals, including but not limited to magnetic storage media, solid-state storage media, optical storage media or semiconductor memory.

[0078] "Information management structure" refers to the logical structure or pattern used in information storage devices to organize and manage data, including but not limited to database table structure, field definition, index structure, and the relationships between data.

[0079] "Standardization processing" refers to the processing steps that unify the format, convert units, standardize encoding, handle missing values ​​and outliers of the raw data received from the terminal, so that it can be stored in a structured manner and statistically analyzed later.

[0080] "Structured storage" refers to a storage method that stores processed data in an information storage device in a field-based, record-based, and indexable manner according to a predefined data model, in order to support efficient retrieval and calculation.

[0081] "Statistical processing program" refers to a program or software module that runs on a server and is used to perform statistical calculations, aggregation operations, trend analysis, and indicator calculations on health-related data.

[0082] "Data processing components" refer to the collection of hardware and software resources in a server used to perform data reading, writing, transformation, calculation and analysis, including but not limited to data processing algorithm modules, data framework libraries and processor resources that work with them.

[0083] "Body mass index" is a calculated indicator used to reflect an individual's body shape or degree of obesity. It includes at least an index calculated using the relationship between weight and height, and can be expanded to include other body shape-related indicators such as waist-to-height ratio.

[0084] "Health indicators" refer to quantitative or qualitative parameters used to characterize health status, calculated based on an individual's attribute information, medical history information, and lifestyle information, including but not limited to body mass index, risk level, and habit score.

[0085] "Risk assessment" refers to the calculation process of inferring and classifying the probability of a specific health event or health problem based on past medical information, family medical history information, and lifestyle information, and outputting results including risk level or risk interpretation.

[0086] "Health status analysis results" refer to the analytical output obtained after statistical processing programs and data processing components comprehensively calculate an individual's attribute information, medical history information, lifestyle information, and emotional state information. This output includes health indicators, risk assessment results, and corresponding explanations.

[0087] "Generative artificial intelligence models" refer to artificial intelligence models that are trained on a large amount of text data based on machine learning or deep learning techniques and can automatically generate new natural language text content based on input text.

[0088] "Prompt statements" refer to the input text constructed to drive generative artificial intelligence models to generate the desired content. They include information describing the individual's state, task instructions, output requirements, and constraints.

[0089] "Character sequence generation logic" refers to the processing logic on the server side that converts various structured data into natural language text strings according to pre-defined rules, including string concatenation, placeholder replacement, and rule-based text generation.

[0090] "Template processing logic" refers to the process of generating a complete prompt statement by filling the placeholder positions in the template with the analysis result data based on a pre-designed text template.

[0091] "Constraints" refer to the instructive descriptions in the prompt statements used to constrain the output range, style, or content type of the generative artificial intelligence model, including but not limited to prohibited output content, guidelines to be followed, and role settings.

[0092] "Output format specification information" refers to the descriptive information in the prompt statement used to specify the output structure of the generative artificial intelligence model, including but not limited to requirements such as output in a list format, segmented output, or output by category.

[0093] "Response text" refers to the natural language output content generated by generative artificial intelligence models based on prompts, including dietary advice, medication-related advice, stress management advice, alcohol intake restriction guidelines, and exercise guidelines.

[0094] "Dietary recommendations" refer to the suggestions for adjustment or optimization given in the response text regarding an individual's dietary structure, food choices, intake, and eating time.

[0095] "Medication-related advice" refers to general, non-prescription advice provided in the response text regarding the timing of medication use, lifestyle-related precautions, etc., without including specific prescriptions or dosage instructions.

[0096] "Stress management suggestions" refer to behavioral recommendations provided in the response text to reduce psychological stress or improve emotional state, including but not limited to relaxation training, adjusting work and rest schedules, and social activities.

[0097] "Alcohol intake restriction guidelines" are binding recommendations given in the response text regarding the frequency, quantity, or avoidance of alcoholic beverages, in order to reduce alcohol-related health risks.

[0098] "Exercise guidelines" refer to the suggestions made in the response text regarding the type, frequency, duration, and intensity of physical activity.

[0099] "External information processing device" refers to an information processing system that operates outside of a server and is used to provide generative artificial intelligence services or product services, including but not limited to cloud service platforms and third-party service servers.

[0100] "Goods services" refers to online services related to physical goods provided to individuals through external information processing devices, including but not limited to food delivery services, goods ordering services, or services for other goods.

[0101] "Linkage information" refers to structured information generated by the server based on the response text and used for data interaction with external goods and services, including but not limited to food selection candidates, specifications, quantity information, and order information.

[0102] "Visualized output" refers to converting the response text and health status analysis results into a format suitable for presentation on a terminal display device, and displaying them in the form of text, icons, charts, or a combination thereof.

[0103] "Additional prompts" refer to extra prompts that the server actively generates based on emotional state information after receiving the response text. These prompts are used to strengthen, weaken, or supplement specific suggestions (such as suggestions about alcohol consumption).

[0104] "Dynamic modification" refers to the server adjusting the existing response text content based on additional prompts without re-requesting the generative artificial intelligence model. This includes adding or deleting certain suggestions, changing the strength of suggestions, or adding reminder information.

[0105] "Historical prompts" refer to prompts that were built by the server and sent to the generative artificial intelligence model in past interactions. These prompts, along with the corresponding response text and user evaluations, are recorded and saved.

[0106] "Evaluation information" refers to feedback data from individuals on the output of generative artificial intelligence models, including but not limited to usefulness evaluations, comprehension difficulty evaluations, willingness to perform evaluations, and textual comments.

[0107] "Historical records" refer to a collection of related data stored in an information storage device in chronological order, including prompts, responses, health status analysis results, emotional state information, and evaluation information.

[0108] "Statistical processing results" refers to the indicators and inferences obtained after statistical analysis of historical records, including but not limited to the satisfaction distribution corresponding to a certain type of prompt statement, execution rate statistics, and the correlation between risk level and feedback.

[0109] The "Prompt Optimization Unit" refers to a functional module in the server that adjusts the structure, content, or constraints of prompt statements based on statistical processing results, thereby gradually improving the quality of health advice output by generative artificial intelligence models.

[0110] In this embodiment of the invention, the server, terminal, and user work collaboratively to achieve a series of technical processes, including the collection, structured processing, statistical analysis, construction of prompt statements, invocation of generative artificial intelligence models, and linkage with external items and services. This embodiment focuses on server-side data structure design, algorithm flow, and model interaction, improving the data processing capabilities of computers and the utilization of generative artificial intelligence in health management scenarios.

[0111] The server can utilize general-purpose computer equipment in its hardware, including at least one central processing unit (CPU), an optional graphics processing unit (GPU), main memory, non-volatile memory, and a network interface. In terms of software, the server can run a general-purpose operating system (such as a Unix-like operating system), web server software (such as a reverse proxy server), application server frameworks (such as a Python-based web application framework), database management systems (such as a relational database management system), and data analysis libraries (such as the Pandas data analysis library). Generative artificial intelligence models can be deployed on the same physical server or on a separate inference server. The inference server may include GPUs and run deep learning frameworks (such as frameworks based on tensor operations).

[0112] The terminal can be a smartphone, tablet, or desktop computer in terms of hardware, and has input devices such as a display device, touch device, or keyboard, as well as a network communication module. In terms of software, the terminal can run a general-purpose browser application or a dedicated health management application to present a graphical user interface and interact with the server. Users input health information through the terminal and view the analysis results and personalized suggestions returned by the server.

[0113] The server maintains multiple logical data structures in the information storage device, including at least: an attribute table for storing basic user information, a health record table for storing medical history and lifestyle information, an emotion record table for storing emotional state information, an interaction record table for storing prompts and corresponding response texts, and a feedback record table for storing user evaluation information. The server uses foreign keys or indexes between these tables to efficiently perform joint queries by user, time, and session. Through this structured data management approach, the server achieves unified management of multidimensional health data and interaction data, reduces redundancy, and improves access efficiency during retrieval and analysis.

[0114] In terms of data processing, the server uses a Python interpreter to execute data processing programs, calling the Pandas library to convert data read from the database into DataFrame structures. The server performs normalization processing on the DataFrames, including unit standardization, missing value imputation, and outlier detection. For example, the server converts user input such as "170cm" and "80kg" into floating-point numbers and maps text fields such as gender and disease type to integer codes for subsequent efficient arithmetic operations and conditional filtering. By storing data in memory in a columnar structure, the server significantly reduces the overhead of repeated parsing and conversion, enabling batch parallel computation of multi-user data and improving the throughput of statistical processing.

[0115] The server employs a well-defined algorithm for calculating health indicators. It reads the height and weight fields from the data frame, converts the height from centimeters to meters, and performs floating-point calculations using the formula Body Mass Index (BMI) = weight (kg) / height (m)^2. The server adds the calculation results to a new column and categorizes the BMI into different classes (e.g., underweight, normal, overweight, obese) based on predefined threshold ranges, storing the classification results in another column. When performing batch calculations, the server utilizes vectorized computation, avoiding individual loop processing for each record. This leverages the underlying linear algebra library for acceleration, achieving higher computational efficiency than traditional record-by-record processing.

[0116] The server combines medical history, family medical history, and lifestyle information for risk assessment. It can store a rule configuration table in its information storage device. This table defines conditions and weights in row records, such as "basic cardiovascular risk score +3 for hypertension," "extra cardiovascular risk score +2 for overweight BMI," "extra score +4 for a first-degree relative with a history of premature heart attack," and "extra score +3 for smoking more than 10 cigarettes per day." During analysis, the server reads the rule configuration table, connects it to a data frame, or maps the conditions to vectorized Boolean judgments within the program. It calculates a weighted total score for each user record and then maps the score range to low, medium, or high risk levels. The server thus constructs a unified risk scoring algorithm framework. When rules or weights are adjusted, no program logic needs to be modified; only the configuration data needs to be updated. Compared to traditional hard-coded conditions, this method is easier to maintain and extend, and the vectorized condition calculation reduces computational complexity.

[0117] The server employs a combination of template-based and data-driven methods to construct prompt statements. Multiple text templates are stored in the information storage device, each containing placeholder symbols such as "{age}", "{gender}", "{BMI value}", "{BMI classification}", "{major chronic diseases}", "{emotional state}", and "{cardiovascular risk level}". When generating prompt statements, the server reads the corresponding fields from the health status analysis results object and replaces them with the placeholders in the templates. The server also embeds explicit output constraints and formatting requirements into the prompt statements. For example, it requires generated responses to be listed in sections such as "Overview Recommendations", "Dietary Recommendations", "Exercise Recommendations", "Stress Management Recommendations", and "Alcohol Intake Recommendations", and requires the use of concise language and avoidance of specific drug dosages. This template-based construction method ensures that the input received by the generative AI model contains highly structured information and clear task descriptions, which is beneficial for the model to generate stable and predictable output formats.

[0118] For example, the server can generate the following prompt and send it to the generative artificial intelligence model: "You are a health management consultant who specializes in lifestyle interventions, not a clinical diagnostic physician."

[0119] Based on the following user health information and system analysis results, please provide specific and actionable health recommendations.

[0120] User Information: - Age: 45, Male - Height: 170cm - Weight: 80kg, BMI approximately 27.7, falls within the overweight range. - Past medical history: hypertension - Family history: Father has a history of heart disease. - Allergies: No known allergies - Lifestyle habits: Exercise once a week, smoke about 10 cigarettes a day, and drink alcohol about 3 times a week. - Emotional state: Recently reported experiencing significant stress and poor sleep quality. System analysis results: - Cardiovascular risk level: High - Metabolic-related risks: Moderate Please output: 1. Overall health risk description (plain, non-diagnostic statement); 2. Lifestyle recommendations for weight management and blood pressure control; 3. Specific suggestions and reasons for quitting smoking and limiting alcohol consumption (no medication prescriptions provided); 4. Suggestions for stress management and sleep improvement; 5. It is recommended to output in item form, and divide it into sections: 'General Description', 'Diet', 'Exercise', 'Stress Management', and 'Alcohol Intake'. The server can employ a Transformer-based language model on the generative AI model side. This model includes a multi-layered self-attention encoder and decoder, with each layer containing a multi-head attention sublayer and a feedforward network sublayer. When calling the inference service, the server converts the prompt into a tokenized sequence via a tokenizer, further encodes it into integer indices, and then feeds it into the model's embedding layer. The model uses pre-trained weights and can be fine-tuned in advance for health scenarios. The server sets generation parameters, such as maximum generation length, temperature, and top-k or top-p sampling strategies, to control the diversity and stability of the output.

[0121] During model training, the server uses a large amount of anonymized health consultation question-and-answer corpus and generated structured prompt-response pairs as training data. The server uses cross-entropy as the loss function, comparing the differences between the model-generated sequences and the target sequences. The server updates model weights through backpropagation and iteratively trains using stochastic gradient descent or its variants (such as the Adam optimization algorithm). Data augmentation techniques can be employed during training, such as synonym replacement and sequence fine-tuning in health information descriptions, to improve the model's robustness to different expressions. After training, the server freezes the model parameters and deploys it as an inference service interface.

[0122] Instead of retraining the model during runtime, the server records historical prompts, model outputs, and user reviews to statistically analyze the impact of different prompt structures and content on user satisfaction. The server can employ clustering or regression analysis to calculate the correlation between different elements in the prompts (e.g., whether risk grading is included, whether the output structure is explicitly defined) and user feedback metrics (e.g., comprehension and helpability ratings). When the server finds that a certain prompt pattern is significantly better than another, it updates the prompt template or constraint configuration, thereby improving the quality of generated content through input layer structure optimization without altering the model's internal weights. This server-side prompt optimization is an adaptive configuration mechanism of the computer system, which is beneficial for continuously improving overall system performance.

[0123] In utilizing emotional state information, the server embeds emotional scores as control variables into the health analysis results and prompts. For example, when the emotional state reflects high stress or depressive tendencies, the server adds prompts such as "Please use gentle, encouraging language when giving advice, and avoid expressions that may induce guilt," and requires stricter restrictions on alcohol consumption recommendations. The server can also configure local rules to reduce the intensity of suggestions regarding vigorous exercise and increase the emphasis on stress management and sleep improvement when emotional stress is high. In this way, the server guides the direction and focus of the model's output through specific logic and rules, achieving suggestion generation that aligns with emotional states and reducing the maladaptive effects of mechanical, uniform output.

[0124] In terms of integration with external goods services, the server parses the dietary recommendations output by the generative AI model into a set of features that can be matched with a food database. For example, the server extracts keywords such as "low salt," "high fiber," and "rich in high-quality protein" from the response text, combines them with the user's preferred tastes and contraindications, and searches for matching foods or meal sets in the local or external goods service's food database. The server constructs a linked information record for each candidate food, including the food identifier, main nutritional characteristics, reasons for recommendation, recommended intake frequency, and available order specifications. The server packages this linked information into structured data and sends it to the external goods service, enabling the service to directly generate an order candidate list. Users only need to perform a few confirmation operations on their terminals to place an order. Through this structured linkage mechanism, the server connects health recommendations with the real-world food delivery process, achieving a closed loop from "analysis—generation—execution," reducing the burden of manual querying and matching for users.

[0125] In terms of output control, the server combines the generated response text with the health status analysis results into a page data structure, labeling each module type (e.g., "Risk Statement," "Dietary Recommendations," "Exercise Recommendations," "Alcohol Restriction," "Stress Management"). The server adjusts the content segment length and display order based on the terminal type (small-screen mobile or large-screen desktop) to reduce scrolling and cognitive load. The server can also add concise icons or risk level indicators next to the text, such as using color and "low / medium / high" labels to represent risk levels, to improve comprehension efficiency.

[0126] After reading the suggestions on the terminal, users can provide feedback through rating controls or text boxes. The terminal sends this feedback to the server, which records it in a feedback log table. The server aggregates and analyzes a large amount of user feedback to identify patterns and content characteristics that generate negative feedback, such as suggestions that are too general or overemphasize limitations. Based on this, the server adjusts the description of the suggestion template, for example, by adding specific suggestions or more detailed explanatory text in high-risk situations. This feedback-driven suggestion optimization process allows the system to gradually learn more effective suggestion configurations during operation, thereby optimizing the entire calculation process without interfering with the internal parameters of the model's black box.

[0127] Through the aforementioned implementation, the server goes beyond simply automating human consultation. It introduces specific algorithms and control logic into multiple internal computer processes, including data structure design, statistical calculation, prompt statement construction, model interaction, and external service linkage. This results in the following technical effects: First, by separating vectorized statistical calculation and rule configuration tables, the processing speed and scalability of multi-user health data analysis are significantly improved. Second, by using structured prompt statements and output format constraints, the randomness and noise of generative AI model output are reduced, improving the stability and parsability of the results. Furthermore, by dynamically adjusting the content and intensity of suggestions based on emotional state information, the risk of inappropriate output is reduced, and the system's sensitivity to changes in user status is improved. Moreover, by constructing linkage information and connecting with external services, abstract health suggestions are transformed into actionable behaviors, enhancing the system's technical application value. Overall, this implementation achieves a highly efficient data processing and model utilization architecture for generative health management on the server side, representing an improvement in computer technology itself in terms of data representation, computational processes, and model interaction.

[0128] use Figure 11 The processing flow is explained.

[0129] Step 1: Users access the health management interface using a browser or dedicated application on their terminal. The input consists of the user's current unstructured health information and emotional state information. Users sequentially fill in their age, gender, height, weight, past medical history, family medical history, allergies, dietary habits, exercise frequency, smoking and drinking habits, and self-assessment of their emotional state (e.g., stress level, mood). The terminal performs preliminary validations, checking for empty fields and numeric fields, and encapsulates all fields into an HTTP request (e.g., form data in an HTTPS POST request or JSON). The output is a data message containing the original health and emotional state information, which is then sent to the server via the communication network.

[0130] Step 2: The server receives request messages from the terminal via a network interface. The input is the HTTP request data sent by the terminal. The server uses the application framework's routing module to parse the request header and request body, extracting the values ​​of various fields, such as age, gender, height_raw, weight_raw, disease_text, lifestyle_text, and emotion_score. The server performs data validity checks according to preset validation rules (numerical range, string length, enumeration validity), generating error messages for fields that do not conform to the rules. The server performs format conversions on the validated data, for example, parsing "170cm" as 170.0, "80kg" as 80.0, mapping gender text to integer encoding, and mapping emotion level text to numerical scores. The server's output is a standardized data structure of health and emotional state information that has undergone basic validation and format uniformity.

[0131] Step 3: The server uses database access components to write standardized data to the information storage device. The input consists of a standardized data structure generated by the server and the current user's identifier. The server constructs insert statements or data records, storing attribute information, medical history information, lifestyle information, emotional state information, and timestamps into their respective logical tables (such as attribute tables, health record tables, and emotional record tables), and maintaining primary key and foreign key relationships. The server executes a transaction commit to ensure data persistence. The output is a set of newly added structured records in the database, along with record identifiers available for subsequent queries.

[0132] Step 4: The server reads health and mood records from the database for analysis. Input is a user identifier or query criteria. The server executes the query, retrieving the user's most recent or multiple health and mood data, and loads the results into a DataFrame using the Pandas library. The server performs further normalization on the DataFrame, including missing value imputation (e.g., filling with the mean or marking it as "unknown"), outlier detection (e.g., marking height below a reasonable lower limit as an outlier), and unit standardization. The output is a cleaned and normalized DataFrame object containing all the fields needed for subsequent calculations.

[0133] Step 5: The server calculates basic health indicators such as Body Mass Index (BMI) within a DataFrame. The input is a DataFrame containing fields such as height and weight. The server converts the `height_cm` field to a `height_m` column (in meters) and calculates BMI using vectorized operations: BMI = weight_kg / (height_m^2), writing the result to a new `bmi` column. The server categorizes BMI according to preset ranges and writes category labels to the `bmi_category` column, such as "underweight," "normal," "overweight," and "obese." If necessary, the server can also calculate additional indicators such as waist-to-height ratio. The output is an expanded DataFrame with added BMI values ​​and category labels, used for subsequent risk assessment and prompt generation.

[0134] Step 6: The server performs risk assessments based on medical history, family medical history, and lifestyle information. Inputs include disease fields, family history fields, smoking and drinking fields, exercise frequency fields, and BMI-related fields from an expanded data frame. The server reads rule configuration data (e.g., conditions and weights stored in a configuration table), uses Boolean logic to determine if a record meets specific conditions, and adds the weights of those meeting the conditions to the corresponding risk dimension scores, such as cardiovascular risk score and metabolic risk score. The server maps the accumulated score range to risk levels such as "low," "medium," and "high," and stores the results in new fields such as `cardio_risk_level` and `metabolic_risk_level`. The server also generates brief risk description text, such as "Presents hypertension and is overweight, with a high cardiovascular risk." The output is a health status analysis result object containing multiple risk levels and description text.

[0135] Step 7: The server integrates health status analysis results with emotional state information to generate prompts for input into a generative AI model. Input includes health status analysis results (including BMI, risk level, major diseases, and a summary of lifestyle habits), emotional scores, and predefined prompt templates. The server selects an appropriate prompt template (e.g., emphasizing weight loss and exercise or chronic disease risk descriptions), fills the data fields into placeholder positions in the template, and adds output structure requirements and constraints to the prompt (e.g., prohibiting the output of specific drug dosages, requiring segmented output by module). The server adjusts the tone and emphasis of the prompts based on the emotional state; for example, when under high stress, instructions such as "Please use gentle and encouraging language" and "Please increase the length of stress management suggestions" are added to the template. The output is a complete natural language prompt text suitable as input to the generative AI model.

[0136] Step 8: The server calls the generative AI model interface and obtains the response text. The input consists of the prompt statement generated in step 7 and the generation parameters (such as maximum length, temperature, top-p, etc.). The server uses an HTTP client library to send a request to the external information processing device deploying the language model, encoding the prompt statement into a request body and attaching authentication information. The generative AI model in the external device, based on a Transformer architecture, segments and encodes the prompt statement, progressively outputting the response text through multi-head self-attention and feedforward network layers. The server waits for the response and receives the generation result within the timeout period. The server parses the returned JSON response and extracts the response text fields. The output is a natural language response text containing dietary recommendations, exercise suggestions, stress management advice, and alcohol restriction guidelines.

[0137] Step 9: The server performs structured parsing and filtering on the generated response text. The input is the raw response text returned by the generative AI model. The server segments and categorizes the text according to a pre-defined output structure (such as paragraph headings or bullet points), mapping the content to modules such as "General Instructions," "Dietary Recommendations," "Exercise Recommendations," "Stress Management," and "Alcohol Intake." The server checks for content exceeding system limits, such as specific drug names and dosage instructions, and performs deletion or replacement prompts for such content to ensure compliance with constraints. The server may further adjust the strength of alcohol-related recommendations based on emotional state information, for example, increasing the restrictive tone in high-risk and emotionally unstable situations. The output is a modular and filtered structured response object, with each module containing processed natural language text.

[0138] Step 10: The server generates linkage information with external item services based on dietary recommendations in the response text. Input includes the dietary recommendation module text from the structured response object, along with the user's preferences and allergy information. The server uses keyword extraction rules or a simplified classification model to identify nutritional characteristic tags such as "low salt," "high fiber," "low fat," and "rich in protein" from the dietary recommendations, and combines these with prohibited ingredients and taste preferences to form food search criteria. The server queries the item service's food database for matching food entries, generating a linkage record for each entry, including the food ID, recommendation reason, expected consumption frequency, and available order specifications. The server encapsulates these linkage records into structured linkage information and sends it as output to the external item service interface so that candidate orders can be directly presented in its system.

[0139] Step 11: After integrating health status analysis results, structured response objects, and linkage information, the server generates display data for the terminal. Inputs include health status analysis result objects, modular response objects, and food linkage information. The server designs the display structure based on the terminal type and screen size, presenting risk levels and key health indicators using icons or brief text, organizing various suggestion modules into segments, and adding interactive elements (such as "order button" labels) to orderable food candidates. The server packages this content into JSON format response data and outputs a response message containing all necessary display fields, which is then sent to the terminal.

[0140] Step 12: The terminal receives display data from the server and presents it in a graphical interface. The input is a JSON response returned by the server. The terminal parses the JSON in its front-end program and generates different display areas on the screen according to module type, such as a risk overview area, various suggestion areas, and food recommendation areas. In the food recommendation area, the terminal generates a clickable order control for each candidate food item; when the user clicks it, an order request is sent to an external item service. The terminal converts the text suggestions provided by the server into a readable layout (headings, lists, paragraphs) and uses colors or icons to emphasize high-risk warnings as needed. The output is a visual health analysis and suggestion interface, allowing users to directly read the suggestions and perform corresponding actions.

[0141] Step 13: Users read the suggestions and provide feedback on the terminal. The input consists of the health advice interface and food recommendation interface displayed on the terminal. Users rate the usefulness and ease of understanding of the suggestions based on their own experience, or enter comments and supplementary explanations in the text boxes. The terminal packages this feedback into an HTTP request and sends it to the server. After receiving the feedback, the server correlates it with the corresponding prompt statement, response text, and health status analysis results, writing it into a feedback record table. In subsequent batch processing analysis, the server uses this feedback as input to perform statistical calculations on the relationship between different prompt statement configurations and output quality, outputting optimized parameters for updating prompt templates and constraints, thereby improving the structure and content in the next round of prompt statement generation.

[0142] Application Example 1 The process flow corresponding to the specific processing in Use Case 1 will be described below. The various parts of the system described below are implemented by the data processing device 12 and the intelligent device 14. Furthermore, the data processing device 12 is referred to as the "server" and the intelligent device 14 is referred to as the "terminal".

[0143] In existing technologies, computer implementations for individual health management and incentives often employ simple rule engines or static threshold judgments, providing fixed suggestions based solely on limited health parameters. This presents the following technical problems: (1) At the information processing level, servers usually only perform coarse-grained statistics or single-dimensional risk scoring on individual health information. They lack unified modeling and feature extraction of multi-source data (medical information, lifestyle habits, emotional state, etc.), resulting in insufficient accuracy of health status assessment and failure to make full use of available data.

[0144] (2) At the level of intelligent computing, traditional systems rely on pre-set rule bases for decision-making. They lack the comprehensive reasoning ability of generative artificial intelligence models for complex health situations and cannot dynamically generate personalized health advice and incentive strategies based on individual differences, which limits the expressive power and scalability of computing models.

[0145] (3) At the interaction and control level, existing systems usually decouple health advice from external services (such as ordering systems for goods or services, electronic payment systems). The server cannot automatically generate executable control instructions and prompts in the end-to-end data flow, making it difficult to accurately map health assessment results into order control, payment discount applications or intake restriction adjustment instructions that can be directly processed by machines, thereby reducing the overall automation level and computational efficiency of the system.

[0146] (4) In terms of economic incentives and behavioral intervention, even if there are incentives such as discounts or points in the existing technology, they are mostly static benefits that are unrelated to health status. The server cannot calculate and adjust the individual-level economic benefit parameters (such as discount rate) in real time based on the fine-grained health status assessment results and explanatory information output by the generative artificial intelligence model, thus failing to form a closed-loop incentive mechanism based on health performance.

[0147] (5) In terms of the linkage between emotion and intake control, traditional systems often rely on manual prompts or simple on / off controls when dealing with the relationship between an individual's emotional state and the restriction of their addiction intake. The server lacks a unified computing framework that integrates emotional state with health assessment results and dynamically drives the update of intake restriction strategies through prompt statements, resulting in insufficient refinement and automation of behavioral intervention.

[0148] The purpose of this invention is to provide a health management system that utilizes generative artificial intelligence models and prompt statements. By introducing a unified preprocessing, feature calculation, generative reasoning, and control instruction generation mechanism for individual medical information, emotional state, and behavioral data on the server side, this system improves the processing methods of health assessment and incentive control from the perspectives of computer architecture and data processing flow. This enhances the accuracy and personalization of health status assessment, strengthens the server's automatic linkage capabilities with external services and electronic payments, and enables dynamic economic incentive control based on health performance and emotional state. As a result, the overall processing performance and automation level of computer technology in health management and incentive scenarios are improved.

[0149] The specific processing performed by the specific processing unit 290 of the data processing apparatus 12 in Application Example 1 is achieved by the following means.

[0150] In this invention, the server includes a control unit for acquiring individual medical information and emotional state from an input device connected to an information processing device; a control unit for performing numerical conversion, feature calculation, and statistical processing on the acquired individual medical information and emotional state to generate physical state indicators; a control unit for constructing structured data or prompts for inputting generative artificial intelligence models based on individual medical information and physical state indicators and obtaining multi-level health status assessment results from the generative artificial intelligence models; and a control unit for generating dietary plans, medication plans, behavioral guidelines, and economic incentive information related to preventive medicine based on health status assessment results and emotional state, and transmitting the economic incentives... The system comprises several control units: one for identifying discounts or service offerings used in electronic payment processing; another for linking order processing with external service providers to correspond to dietary and medication plans, and applying economic incentives to these devices or the electronic payment process; a third for generating prompts based on emotional and health status assessments to dynamically adjust the amount or frequency of addictive substances consumed and updating intake restriction strategies accordingly; and a fourth for presenting health status assessments, dietary and medication plans, and economic incentives to individuals, and controlling the execution of external service providers and electronic payment processes based on individual choices. This allows for the formation of an integrated data processing and control chain within the server, encompassing multi-source individual data collection, feature preprocessing, generative AI model inference, external service control command generation, and automatic application of electronic payment discounts. This enables high-precision, multi-dimensional health status assessment and real-time conversion of assessment results into individualized order control, economic incentives, and intake restriction adjustments. Ultimately, this enhances the intelligent decision-making capabilities, automated linkage capabilities, and overall processing efficiency of the health management system at the computer technology level.

[0151] "Individual medical information" refers to various types of data used to reflect the physical condition of a specific individual, including but not limited to age, gender, height, weight, past medical history, allergy information, medication history, examination results, and health-related lifestyle information.

[0152] "Emotional state" refers to data used to represent an individual's psychological or emotional state over a specific period of time, including but not limited to the categories, intensity, and characteristics of changes over time of emotions such as tension, anxiety, pleasure, depression, and irritability.

[0153] "Information processing device" refers to a computing device that has the functions of receiving, storing, processing and outputting data, including but not limited to servers, computer terminals, virtual computing nodes or combinations thereof.

[0154] "Input device" refers to a device or component used to acquire individual input data and provide it to an information processing device, including but not limited to terminal devices with graphical user interfaces, sensor devices, wearable devices, and data acquisition interfaces.

[0155] A “control unit” refers to a logical module or circuit structure in an information processing device that performs a specific function through program execution. It can be implemented individually or in combination by software modules, firmware modules, or hardware circuits executed by a processor.

[0156] "Preprocessing" refers to the formatting, cleaning, and transformation operations performed on raw data before individual medical information and emotional states are input into generative artificial intelligence models or other computing modules. This includes numerical transformation, handling of missing values, calculation of features, normalization, and statistical analysis.

[0157] "Body condition indicators" refer to a set of parameters used to quantify an individual's physical condition, obtained after preprocessing and feature calculation based on individual medical information and related data. These include, but are not limited to, body mass index, blood pressure risk level, metabolic risk score, and comprehensive health score.

[0158] "Generative AI models" refer to AI models that can automatically generate output content based on input data or prompts, including but not limited to deep learning-based language models, generative models, or hybrid models, used to generate health status assessment results, explanatory information, or suggestions.

[0159] "Structured data" refers to data forms that are organized according to a predetermined format and can be directly parsed and processed by programs, including but not limited to vector, matrix, tabular, or key-value pair data, used as input for generative artificial intelligence models or other algorithmic models.

[0160] "Prompt statements" refer to text content written in natural language or a form similar to natural language, used to interact with generative artificial intelligence models. By embedding individual medical information, emotional state, and lifestyle information, these prompts guide the model to generate health status assessment results or suggestions.

[0161] "Health status assessment results" refer to the conclusions about an individual's health status output by generative artificial intelligence models or other algorithmic models based on individual medical information, emotional state, and physical state indicators, including classification results that divide health status into multiple levels and their related descriptions.

[0162] A “dietary plan” refers to recommendations generated for an individual based on health status assessment results and preventive medicine knowledge, which relate to the types of food, intake, frequency of intake, and dietary structure.

[0163] "Medication use plan" refers to the recommended information generated for an individual based on health status assessment results, past medical conditions, and medication information, which includes recommendations related to the type, dosage, frequency, and precautions for use of medication.

[0164] "Behavioral guidelines" refer to suggestions and information related to daily behavior, exercise habits, rest arrangements, stress management, and lifestyle adjustments generated to guide individuals in improving their health.

[0165] "Economic incentive information" refers to parameters and content that are associated with an individual's health status assessment results and behavioral performance and are used to provide economic benefits in transactions or services, including but not limited to discount rates, reduction amounts, points rewards, and free services.

[0166] "Electronic payment processing" refers to the process of settling funds between the payer and the payee through the internet or electronic communication, including steps such as order amount calculation, discount application, authorization request, deduction and result return.

[0167] "Fee discount" refers to a price adjustment method in electronic payment processing that reduces the original payment amount based on economic incentive information. It can take the form of a percentage discount or a fixed amount reduction.

[0168] "Service provision content" refers to the additional rights provided to individuals based on economic incentives in external services or transactions, including but not limited to free services, value-added services, upgraded services, and the right to use additional functions.

[0169] "External service providing device" refers to a system or device that communicates with an information processing device via a network or interface for providing goods or services to individuals, including but not limited to order processing systems, delivery systems, health service platforms, and third-party business systems.

[0170] "Addiction products" refer to consumer goods that individuals consume out of preference and that have a potential impact on health, including but not limited to alcoholic beverages, sugary drinks, high-fat foods, and other non-essential foods or drinks.

[0171] "Intake limit" refers to the maximum quantity or total amount of a particular habit that is allowed to be consumed within a given time period, and is used to constrain an individual's intake behavior.

[0172] "Intake frequency" refers to the number of times or time intervals within which an individual is allowed to consume a craving product, and is used to control the frequency of intake behavior.

[0173] "Lifestyle information" refers to data used to describe an individual's behavioral patterns in daily life, including but not limited to exercise frequency, sleep duration, drinking habits, smoking status, and dietary preferences.

[0174] "Behavioral history information" refers to individual behavior-related data collected and recorded over a period of time, including but not limited to past exercise records, diet records, addiction intake records, sleep records, and historical health intervention implementation status.

[0175] "Explanatory information" refers to natural language or structured descriptions generated by generative artificial intelligence models or other algorithms to explain the reasons or basis for the formation of health status assessment results.

[0176] "Improvement suggestions" refer to specific adjustment plans or action recommendations generated based on the health status assessment results and explanatory information, in order to improve an individual's health level or emotional state.

[0177] "Guidance information" refers to the instructive or suggestive content output by the system, compiled based on improvement suggestions, and used to guide individuals in carrying out stress management or behavior change activities.

[0178] "Individual identification information" refers to identification data used to uniquely identify a specific individual in the system, including but not limited to user ID, account ID, terminal ID, or a combination thereof.

[0179] In one embodiment of the present invention, a server, a terminal, and a user collaboratively constitute a health management and incentive system. The server, as the core information processing device, is equipped with a processor, memory, a network interface, and an optional graphics processing unit. At the software level, the server can use a general-purpose operating system-based runtime environment to run multiple modules, including data processing programs, generative artificial intelligence model inference programs, and network service programs. In one embodiment, the server can use a scripting language environment to execute program logic and load and execute generative artificial intelligence models through a machine learning framework (such as a deep learning framework).

[0180] At the hardware level, servers can use multi-core central processing units to handle tasks such as request scheduling, data preprocessing, and database access, and use graphics processing units to accelerate forward inference operations of deep neural networks, such as matrix multiplication, convolution operations, and non-linear activation operations. At the software level, servers can use data analysis libraries to perform numerical calculations and statistical processing to extract and transform features of individual medical information and emotional states.

[0181] In one embodiment, the terminal can be a smart mobile terminal, including a touchscreen, a wireless communication module, a sensor module, and local storage. The terminal runs a mobile application at the software level, which provides a graphical user interface to display input forms, health status assessment results, and economic incentive information, and communicates with the server encrypted via a network communication library. In this invention, the terminal primarily performs the functions of data collection, result presentation, and user interaction.

[0182] Users input their individual medical information and emotional state through the terminal's graphical user interface. Users can enter their age, gender, height, weight, past medical conditions, allergies, and lifestyle habits (including exercise frequency, alcohol consumption, and smoking status). They can also input their current emotional state, such as level of tension or pleasure, through a scale or selection controls. After completing the input, users submit the data to the server via the terminal's submission button.

[0183] The server receives data requests from terminals using a network service program and parses them into an internal data structure. In a preferred embodiment, the server organizes the input data into a key-value pair record structure and temporarily stores it in memory as a feature dictionary or feature vector. The server performs numerical conversion on individual medical information, converting string-based values ​​into numeric types and mapping category information such as gender, past medical conditions, and emotional state to integer codes. The server further calculates derived features, such as calculating body mass index (BMI) based on weight and height, and calculating a comprehensive risk score based on age, BMI, and past medical conditions.

[0184] During feature processing, the server uses statistical parameters (such as the mean and variance pre-computed on the training dataset) to standardize or normalize continuous features. The server also vectorizes the emotional state data, for example, mapping different emotional dimensions to a multi-dimensional vector space, and can calculate moving averages or fluctuations of emotional states based on time windows to reflect the stability or abnormality of emotions. These processed features are combined into fixed-length feature vectors, which serve as structured inputs to generative artificial intelligence models.

[0185] In one embodiment, the server uses two types of models to work in parallel or selectively: a discriminative deep neural network based on feature vectors to output health status levels; and a generative artificial intelligence model based on natural language to output natural language descriptions, behavioral suggestions, and prompts. The server inputs the feature vectors into a multi-layer feedforward neural network, which may include an input layer, several hidden layers, and an output layer. The layers are connected by fully connected weight matrices and non-linear activation functions. The hidden layers may use rectified linear unit activation functions, and the output layer uses a normalization function suitable for multi-class classification, outputting probability values ​​corresponding to three or more health levels.

[0186] During model training, the server uses labeled historical medical data and emotional state data as the training set. The server calculates the model output through forward propagation and then uses the cross-entropy loss function to calculate the error between the predicted result and the true label. The server uses gradient descent or its variants (such as adaptive learning rate algorithms) for backpropagation to update the neural network's weight parameters. During training, the server can introduce data augmentation strategies, such as adding small noise to numerical features or perturbing age or weight data within a reasonable range, to improve the model's robustness. The server can also employ regularization or dropout techniques to prevent overfitting, thereby improving generalization ability on unknown user data.

[0187] During deployment, the server loads the pre-trained model weights and fixes them in inference mode. After receiving the feature vectors sent by the terminal, the server performs forward propagation on the graphics processing unit or central processing unit to obtain probability vectors for each health status level. The server determines the health status assessment result based on the index corresponding to the highest probability, such as "good," "attention," or "needs improvement." When needed, the server can also output the second highest probability as a confidence interval reference for further adjusting the upper and lower limits of economic incentives.

[0188] The server simultaneously constructs natural language prompts based on the same individual medical information, physical condition indicators, and emotional state, and uses these as input to a generative artificial intelligence model. In one embodiment, the server organizes the prompts into continuous text, for example: "The user is 30 years old, male, weighs 70 kg, is 175 cm tall, has no medical history, runs 3 times a week for 30 minutes each time, does not smoke, and occasionally drinks alcohol. Based on this personal medical information, please assess the user's health status and choose one of the three levels: 'Good,' 'Pay attention,' or 'Need improvement,' and give a conclusion, explaining your reasoning in one sentence." In another embodiment, the server uses a message prompting the user with an indication of an economic incentive policy, for example: "You are a generative AI model in a health management and incentive system. Below is a user's personal medical information: age 28, male, weight 68 kg, height 178 cm, no past medical history, lifestyle habits include running 3 times a week for 30 minutes each time, non-smoker, and occasional alcohol consumption. Please first determine whether the user's health status falls into the 'good,' 'attention,' or 'needs improvement' category. Then, design a suitable electronic payment discount ratio for the user (e.g., 3%–10%), and explain in two sentences why this discount ratio is given. Please answer in a structured manner, for example: 'Health status: good; Recommended discount: 5%; Reason: ……'." The server inputs the aforementioned prompts into a generative artificial intelligence model. In one embodiment, this model may employ a multi-layer encoder and decoder structure based on a self-attention mechanism, internally utilizing components such as multi-head attention, feedforward networks, and layer normalization. During the model inference phase, the server performs embedding transformation, positional encoding, attention weight calculation, and decoding steps, thereby outputting natural language text such as health status conclusions, explanations of reasons, and recommended discounts. The server then parses the output text, extracting health status labels, discount ratios, and explanations of reasons based on predefined keywords or regular expressions, thus transforming the natural language results into a machine-processable data structure.

[0189] The server performs a consistency check between the health status probabilities output by the discriminative model and the natural language conclusions output by the generative AI model. In one embodiment, the server specifies that when the two health levels are inconsistent, the following rule applies: if the highest probability of the discriminative model exceeds a preset threshold, the result of the discriminative model takes precedence; otherwise, the level parsed from the text of the generative AI model takes precedence. This unconventional decision fusion rule combines, to some extent, the stability of the numerical model with the interpretability of the generative model, which can improve the reliability of the overall evaluation results. The server writes the final health status level and model confidence score together into the database to support subsequent strategy adjustments and statistical analysis.

[0190] After receiving the health status assessment results and explanatory information, the server generates diet plans, medication usage plans, and behavioral guidelines based on preset rules. At the algorithmic level, the server can define several rule mapping tables, mapping combinations of health status levels, body mass index ranges, and emotional state levels to dietary control levels, exercise suggestion intensity levels, and medication usage attention levels. The server obtains personalized suggestion parameters through table lookups and numerical calculations, and then a generative artificial intelligence model converts these parameters into readable natural language descriptions, thus balancing machine processing efficiency and user understandability.

[0191] The server generates economic incentive information based on the health status assessment results and explanatory information. The server can preset several candidate economic incentive tiers, such as 3%, 5%, 8% discounts, and complimentary specific services. The server selects the appropriate incentive tier based on multiple features, including health status level, emotional stability, and historical behavior and fulfillment records, using a rule engine or lightweight regression model. In one embodiment, the server defines a calculation formula that maps health scores and behavioral scores to a continuous interval of discount rates, then truncates it to a legal range based on business constraints. The server binds the final discount rate to user identification information, records it in a database, and associates it with the validity period and applicable scenarios.

[0192] In terms of technical implementation, the server achieves several technical effects through the aforementioned data structures and algorithmic processes. Because the server performs feature-based and standardized processing of raw medical information and emotional states, the model can more effectively distinguish between high-risk and low-risk users, thereby improving the accuracy of health status assessment. By performing vectorized operations and batch inference within the server, the system can maintain a high processing speed even under high-concurrency requests, reducing the response time of a single assessment. By uniformly managing economic incentive parameters on the server side and automatically applying them during electronic payments, the system reduces the number of round-trip interactions between the terminal and the external payment system, thereby reducing communication load and the probability of synchronization errors.

[0193] In this invention, the terminal displays the health status assessment results, diet and medication usage plans, and economic discount information generated by the server. Upon receiving the server's response, the terminal displays the user's health status (e.g., "Good," "Note," "Needs Improvement"), the explanatory text output by the generative artificial intelligence model, and information such as the discount percentage and validity period. The terminal can temporarily store discount information locally so that it can automatically prompt for available discounts when the user initiates electronic payment. When linked with the payment system, the terminal sends the order amount and user identification information to the server, which then executes the discount application logic internally, thereby avoiding the repeated implementation of complex calculation rules on the terminal side and reducing the terminal's processing burden.

[0194] Users can view health assessment details and recommendations on the terminal, and choose whether to order specific diet or medication plans, and whether to use health-related discounts at checkout. When using addictive substances, users can also view their current intake limit and remaining allowed intake times on the terminal. The server dynamically adjusts the addictive substance intake restriction strategy based on emotional state and health assessment results, and returns this information to the terminal as a prompt, such as: "Based on your current emotional state and health condition, it is recommended that you drink no more than one alcoholic beverage today. Please avoid exceeding this limit for two consecutive days." Users receive this prompt on the terminal, thus guiding their behavior.

[0195] Through the specific model structure, feature processing methods, and control logic described above, the server not only implements the health management business itself but also improves the accuracy, efficiency, and resource utilization of data processing from a computer technology perspective. Utilizing generative artificial intelligence models and prompting mechanisms, the server transforms natural language descriptions, originally intended only for human reading, into parameters that can be parsed and controlled by machines, thereby bridging the technical link from multi-source data to external device control (including discounts applied in ordering systems and electronic payment systems). This architecture, which combines discriminative and generative models and deeply couples health status assessment with economic incentive control, differs from traditional systems implemented solely with rule engines or static templates. It significantly improves automation and scalability while ensuring computational accuracy.

[0196] use Figure 12 The processing flow is explained.

[0197] Step 1: Users enter their personal information through the terminal and confirm submission.

[0198] Users enter their age, gender, height, weight, past medical conditions, allergies, lifestyle habits, and current emotional state score in the graphical user interface of the terminal, and then click the "Submit" button.

[0199] Input: Raw text and numerical data entered or selected by the user in the interface components.

[0200] The terminal reads the content from each interface component as internal variables, checks the required fields, and performs format validation on fields such as age, height, and weight (e.g., whether they are numbers or within a reasonable range).

[0201] Data processing: The terminal provides error messages for invalid inputs according to predefined rules and prevents submission; when all inputs are valid, these fields are assembled into a record with a key-value pair structure.

[0202] Output: A structured record of an individual's medical information and emotional state, ready to be sent to the server.

[0203] Step 2: The terminal packages individual information and sends it to the server.

[0204] After the terminal passes the verification, it serializes the structured record into the request body content and adds metadata such as user identifier and timestamp.

[0205] Input: The structured record generated in step 1 and the user identification information stored locally.

[0206] Data processing: The terminal uses network communication components to construct request messages, writes authentication information in the message header, and embeds serialized individual medical information and emotional state data in the message body.

[0207] Output: A remote request sent via a network interface, carrying individual medical information and emotional state, targeting the server's health assessment interface.

[0208] Step 3: The server receives the request and parses the individual information.

[0209] After receiving a request from the terminal, the server's network service program forwards it to the backend processing module.

[0210] Input: A request message containing individual medical information and emotional state data.

[0211] Data processing: The server parses the user identifier from the message header, parses each data field from the message body, and maps the text-based fields to internal data structures (such as dictionaries or record objects); at the same time, it checks the integrity of the fields and the validity of the data types.

[0212] Output: A structured record of individual information in the server's memory, ready for preprocessing; if parsing fails, an error response is generated and returned to the terminal.

[0213] Step 4: The server preprocesses and performs feature calculations on individual medical information and emotional states.

[0214] The server performs numerical conversion, encoding, and derived feature calculation on the parsed individual information in the data processing module.

[0215] Input: The structured individual information record obtained in step 3.

[0216] Data processing: The server converts strings such as age, height, and weight into numerical types; maps category fields such as gender, past medical conditions, and emotion category to integer codes; calculates the body mass index based on height and weight, and generates a comprehensive risk score based on age, body mass index, and past medical conditions; applies standardization transformations (subtracting the mean and dividing by the standard deviation) to continuous features, and constructs an emotion feature vector from the emotion scores.

[0217] Output: A set of fixed-length feature vectors containing standardized medical features, physical condition indicators, and emotional features, for use in subsequent model calculations.

[0218] Step 5: The server constructs structured input and prompt statements for generative artificial intelligence models.

[0219] The server dynamically generates two types of model inputs based on feature vectors and raw text fields: numerical vector inputs and natural language prompts.

[0220] Input: The feature vector obtained in step 4, as well as the original individual information and emotional state text.

[0221] Data processing: The server arranges the feature values ​​into a one-dimensional numerical array according to a predefined order for use in the discriminative model; simultaneously, it generates prompt statements using string concatenation, for example: "The user is 30 years old, male, weighs 70 kg, is 175 cm tall, has no medical history, runs 3 times a week for 30 minutes each time, does not smoke, and occasionally drinks alcohol. Based on this personal medical information, please assess the user's health status and choose one of the three levels: 'Good,' 'Pay attention,' or 'Need improvement,' and give a conclusion, explaining your reasoning in one sentence." Output: A feature vector input for the numerical model and a prompt statement for the generative artificial intelligence model.

[0222] Step 6: The server invokes a discriminative neural network to generate health level probabilities.

[0223] The server loads a pre-trained multi-layer feedforward neural network into the inference module and inputs the feature vectors into the network to perform forward propagation.

[0224] Input: Feature vector from step 5.

[0225] Data processing: The server performs matrix multiplication and nonlinear activation in each network layer to transform the input features into hidden representations layer by layer. Finally, the output layer obtains the probability distribution of the corresponding health levels such as "good", "attention", and "needs improvement". The server then determines the preliminary health level through maximum index operation and retains the complete probability vector as confidence information.

[0226] Output: A preliminary health level label and the corresponding multidimensional probability vector.

[0227] Step 7: The server invokes a generative artificial intelligence model and parses the natural language to arrive at the conclusions.

[0228] The server takes the prompt generated in step 5 as input and passes it to the generative artificial intelligence model to perform text generation inference.

[0229] Input: Prompt text and model inference parameters (e.g., maximum output length, temperature coefficient).

[0230] Data Processing: The server performs embedding transformation, self-attention calculation, and decoding steps within the model to obtain a complete natural language output, such as: "The user's health status is 'good' because their weight matches their height, their BMI is within the normal range, and they have a regular exercise habit." The server then performs keyword matching and pattern parsing on the output text to extract health level terms and explanatory sentences.

[0231] Output: A health rating text conclusion from a generative AI model and a text explaining the reasoning behind it.

[0232] Step 8: The server combines the outputs of the two models to obtain the final health status assessment result.

[0233] The server performs consistency assessments and priority decisions on the results of discriminative neural networks and the textual conclusions of generative artificial intelligence models.

[0234] Input: The probability vector and health level label obtained in step 6, and the health level words and explanations parsed in step 7.

[0235] Data processing: The server compares whether the two health levels are consistent; if they are inconsistent, it checks whether the highest probability of the discriminative model exceeds a preset threshold. If it does, the numerical model result is used; otherwise, the generative artificial intelligence model result is used. The server combines the final adopted health level with the corresponding reason to form a unified health status assessment data record.

[0236] Output: A comprehensive health status assessment result that includes the final health level, probability information, and natural language explanation.

[0237] Step 9: The server generates diet plans, medication usage plans, and behavioral guidelines.

[0238] The server, in its rules engine or lightweight decision module, formulates personalized recommendations based on the final health status assessment results and individual characteristics.

[0239] Input: The comprehensive health status assessment results generated in step 8, and the physical status indicators and emotional characteristics in step 4.

[0240] Data processing: Based on health level, body mass index range, past disease type, and emotional stability, the server selects the applicable dietary control level, recommended food categories, medication usage precautions, and exercise frequency from a predefined rule mapping table; the server combines these discrete options and generates multiple guidance statements through templates, such as the daily recommended energy intake range and suitable exercise duration.

[0241] Output: A set of structured suggested parameters and corresponding natural language text for diet plans, medication use plans, and behavioral guidelines.

[0242] Step 10: The server calculates and determines the economic incentive information.

[0243] The server provides a matching economic incentive plan based on health status, behavioral suggestions, and explanatory information.

[0244] Input: Health level and reason explanation in step 8, behavioral guidelines in step 9, and historical behavioral performance (if any).

[0245] Data processing: The server uses preset discount calculation rules to map health level to a basic discount range, and adjusts the upper and lower limits based on emotional stability and historical performance. The server obtains the specific discount rate through numerical calculation and selects whether to add free services. The server generates a discount record and binds the discount rate, validity period, and applicable scope to the user's identification information.

[0246] Output: A record of economic discount information, including discount rate, validity period, applicable service type and user ID.

[0247] Step 11: The server generates prompts related to addiction intake control.

[0248] The server uses health and emotional states to generate natural language prompts for limiting or adjusting addiction intake.

[0249] Input: Health level and mood characteristics in step 8, user's favorite products usage preferences and historical intake records.

[0250] Data processing: The server calculates the current recommended intake limit and allowable frequency based on preset strategies. For example, it reduces the alcohol intake limit based on the high-risk level and increases the strictness of the restriction based on emotional fluctuations. The server fills the calculation results into the prompt template and generates a prompt statement, such as: "Based on your current health and emotional state, it is recommended that you drink alcohol no more than 3 times this week, no more than 1 drink each time. Please avoid drinking alcohol on two consecutive days." Output: A prompt statement regarding the maximum intake and frequency of the addictive substance.

[0251] Step 12: The server returns the overall results and control parameters to the terminal.

[0252] The server packages the health assessment results, various recommendations, and economic incentive information into a unified response data and returns it to the terminal.

[0253] Input: the comprehensive health status assessment results from step 8, the suggested text from step 9, the preferential information from step 10, and the prompt statement from step 11.

[0254] Data processing: The server organizes this content into a structured response, which includes natural language text for display and parameter values ​​for subsequent payment control; the server sends the response to the terminal via a network interface.

[0255] Output: A comprehensive response data set for the end user, including health status, explanation of reasons, suggested text, economic benefits parameters, and addiction control prompts.

[0256] Step 13: The terminal displays the results and prepares for subsequent payment and behavior control.

[0257] After receiving the server's response, the terminal displays the relevant content on the interface, preparing for future electronic payments and intake control.

[0258] Input: The comprehensive response data returned by the server in step 12.

[0259] Data processing: The terminal displays health level, reason, diet and medication usage plan, behavior guidance, economic discount information and addiction reminders by region; at the same time, it saves discount parameters and validity period, as well as the current addiction intake limit in local storage, for later reference in the payment interface and reminder interface.

[0260] Output: A user-friendly visual interface and a set of control parameters cached locally on the terminal, supporting automatic discount applications and intake limit reminders during subsequent order generation and electronic payments.

[0261] Alternatively, an emotion engine for inferring user emotions can be combined. That is, the specific processing unit 290 can also use the emotion-specific model 59 to infer user emotions and perform specific processing using user emotions.

[0262] Example 2 The flow of a specific process in Example 2 will be described. Each part of the system described below is implemented by the data processing device 12 and the smart device 14. The data processing device 12 will be referred to as the "server," and the smart device 14 as the "terminal."

[0263] In existing technologies, computer-based solutions for individual health management typically only perform simple rule matching or static evaluation of the user's health-related data on a server, and then output fixed template-based dietary or exercise recommendations. These solutions suffer from the following technical problems: (1) The data processing flow on the server side is mostly to perform simple queries and condition judgments on the raw structured data. There is a lack of a mechanism to use data processing programs and machine learning programs to perform unified modeling and feature extraction on multi-source health data. As a result, the server cannot fully explore the implicit features related to disease risk and lifestyle risk at the computational level, and the overall health assessment accuracy and personalization of the system are limited.

[0264] (2) Even if existing systems introduce generative artificial intelligence models, they often directly combine the user's original information and send it to the model. They lack the ability to construct high-quality prompts based on machine learning risk assessment results and calculated body indicators. The prompts lack structured alignment with the model's internal representation, resulting in unstable generation results and insufficient correlation with medical prevention goals, which limits the effective use of generative artificial intelligence models in the field of health management.

[0265] (3) Many existing solutions do not deeply integrate the health assessment results on the server side with external services (such as food service) at the system level. Even if it is possible to jump to external services, there is a lack of a technical process in which the server automatically extracts recommended food categories and dietary menu candidates based on the generated results and generates data that can be directly used for ordering external services. This makes it difficult for the calculation results on the server side to be transformed into executable behavior paths in a timely manner, resulting in a poor user experience.

[0266] (4) In terms of emotional state and addiction intake control, traditional systems usually only provide simple reminders or static restrictions at the front end. They lack the ability of servers to dynamically calculate addiction intake restriction conditions based on emotion-related information and risk assessment information, and to generate restriction adjustment suggestions through generative artificial intelligence models. They cannot form an adaptive restriction control closed loop within the computer system, and the system lacks flexibility and intelligence in responding to changes in user state.

[0267] (5) Regarding stress management, most systems only provide general text suggestions. They lack a way to uniformly encode health-related information, emotion-related information and risk assessment information on the server and use it as input to a generative artificial intelligence model to generate stress management suggestions closely related to individual health risks. This makes it difficult to improve the relevance and feasibility of the suggestions at the algorithm level.

[0268] Therefore, there is an urgent need for a technical solution that improves data preprocessing, feature extraction, prompt statement construction, generative artificial intelligence model invocation, and linkage mechanism with external services on the server side, thereby improving the intelligence, personalization, and end-to-end execution efficiency of the health management system at the computer technology level.

[0269] The specific processing performed by the specific processing unit 290 of the data processing apparatus 12 in Embodiment 2 is achieved by the following means.

[0270] In this invention, the server includes a data acquisition unit for receiving individual health-related information and emotion-related information from a communication terminal and generating structured information; a data preprocessing unit for running an information processing program on general computing resources to convert the structured information into tabular data and calculating body indicators and extracting feature quantities through a numerical calculation program; a risk assessment unit for calling a machine learning program to calculate disease risk and lifestyle risk based on the feature quantities; a prompt statement generation unit for generating health status description information based on the structured information and the risk assessment information and constructing prompt statements for a generative artificial intelligence model; a generation result acquisition unit for inputting the prompt statements into the generative artificial intelligence model and obtaining suggestion information including nutritional intake suggestions, exercise plan suggestions, and drug use suggestions based on preventive medicine; an external service linkage unit for converting the suggestion information into visual information and sending it to the communication terminal, and for linking information with external dietary services based on diet-related suggestions to support product selection and ordering operations; and a restriction control unit for calculating addiction intake restrictions based on emotion-related information and risk assessment information and generating restriction adjustment suggestions through a generative artificial intelligence model to update the addiction intake restrictions. This allows for the formation of a complete computational chain within the server, from structured modeling, feature extraction, and risk assessment of multi-source health and emotion data, to the construction of high-quality prompt statements and generative artificial intelligence model inference, and finally to the execution in conjunction with external services. This significantly improves the data processing capabilities, model inference effectiveness, and end-to-end service execution efficiency of the individual health management system at the computer technology level, and enables adaptive intelligent support for nutrition management, exercise planning, medication use, addiction control, and stress management.

[0271] "Individual health-related information" refers to data related to an individual's physical condition, including but not limited to basic attribute information, body measurement information, past medical history, physiological indicators, and lifestyle information.

[0272] "Emotion-related information" refers to data used to represent an individual's emotional or psychological state at a specific time, including but not limited to emotion labels such as tension, anger, anxiety, depression, and relaxation, as well as quantitative information related to the intensity and duration of emotions.

[0273] "Communication terminal" refers to an information processing device used to interact with a server, including but not limited to smartphones, tablet computing devices, portable computing devices, and desktop computing devices.

[0274] "Structured information" refers to health-related and emotion-related information that has been formatted and has clearly defined fields and data types, including data stored and transmitted in the form of key-value pairs, tabular data, and record sets.

[0275] "General-purpose computing resources" refer to information processing hardware resources that are not specifically optimized for any particular application and are capable of running a variety of software programs, including but not limited to central processing units, storage devices, and network interfaces.

[0276] "Information processing program" refers to a program that runs on general computing resources and is used to transform, clean, format and integrate structured information, including but not limited to data parsing programs, data conversion programs and data verification programs.

[0277] "Tabular data" refers to a binary data format that uses rows and columns as its basic structure to represent multiple records and their fields, including but not limited to data tables, data frames, and spreadsheets.

[0278] "Numerical computation processing" refers to the processing of numerical data through mathematical operations, including but not limited to addition, subtraction, multiplication, division, exponentiation, normalization, statistical calculation, and indicator derivation.

[0279] "Body metrics" refer to quantitative indicators that reflect an individual's physical condition, obtained from body measurement information and numerical processing. These include, but are not limited to, body mass index, body fat estimation parameters, blood pressure level classification, and metabolic risk scores.

[0280] "Features" refer to numerical or categorical variables extracted from structured information and physical indicators to describe an individual's health status and for use by machine learning programs, including but not limited to age codes, disease presence markers, lifestyle habit levels, and normalized health parameters.

[0281] "Machine learning program" refers to a software program that uses a pre-trained model to perform inference calculations on input features to output prediction or evaluation results, including but not limited to classification models, regression models, and risk scoring models.

[0282] "Risk assessment information" refers to the quantitative or graded results obtained by machine learning programs based on feature quantities, used to represent the risk of an individual's disease occurrence and the risk of lifestyle habits, including but not limited to risk probability values, risk level labels, and multidimensional risk score vectors.

[0283] "Health status description information" refers to textual information that comprehensively describes an individual's current health status in natural language based on structured information and risk assessment information, including but not limited to descriptions of major health problems, potential risks, and lifestyle characteristics.

[0284] "Generative artificial intelligence models" refer to models that are based on statistical learning and deep learning techniques and can automatically generate natural language text content based on input text or other forms of data, including but not limited to large-scale language models and sequence generation models.

[0285] "Prompt statements" refer to natural language or semi-structured instruction text provided to generative artificial intelligence models to describe the input context, limit the generation task, and specify the output format, including but not limited to problem descriptions, role settings, and output requirements.

[0286] "Nutritional intake recommendations" refer to natural language suggestions based on preventive medicine principles and individual health status, regarding dietary structure, food types, and intake amounts, including but not limited to staple food pairings, vegetable and fruit intake, and types of foods that need to be restricted.

[0287] "Exercise plan recommendations" refer to exercise arrangement suggestions based on individual health conditions and risk assessment results, including but not limited to the type of exercise, duration of each exercise session, weekly exercise frequency, and exercise intensity level.

[0288] "Medication use advice" refers to general natural language advice that individuals should pay attention to regarding medication use, based on medical professional principles. This includes, but is not limited to, taking medication as prescribed, avoiding self-adjustment of dosage, and adhering to follow-up appointment schedules.

[0289] "Suggestion information" refers to a collection of natural language texts generated by generative artificial intelligence models under the constraints of prompt statements, which includes suggestions on nutrition intake, exercise plans, medication use, and other health management.

[0290] "Visual information" refers to the formatted display data of suggested information for presentation on communication terminals, including but not limited to text lists, icons, charts, and highlighted marks divided into modules.

[0291] "External food service" refers to services provided by third-party service systems for selecting and delivering food products, including but not limited to food ordering platforms, food delivery service systems, and online nutrition meal planning service systems.

[0292] "Product selection and ordering" refers to the interactive process of browsing, filtering, confirming, and submitting order requests based on candidate product information when providing external food and beverage services.

[0293] "Constraints on consumption of addictive substances" refers to control rules, such as maximum intake, frequency of intake, or prohibition conditions, set for addictive consumer products like alcoholic beverages and sugary drinks, based on individual health status and risk assessment results.

[0294] "Restriction adjustment suggestions" refer to natural language suggestions generated by generative artificial intelligence models based on the latest emotion-related information and risk assessment information, used to increase, decrease, or refine restrictions on the intake of addictive substances.

[0295] "Stress management suggestions" refer to natural language suggestions aimed at reducing mental burden and emotional stress, focusing on aspects such as sleep habits, relaxation training, time management, and social support.

[0296] The embodiments of this invention will be described in detail, combining hardware structure, software modules, data structure, and the internal algorithm flow of the generative artificial intelligence model, to illustrate the system's composition and technical effects. The following embodiments are illustrative of the invention and not intended to limit it; various modifications and variations are possible without departing from the spirit of the invention.

[0297] In the following description, the server, terminal, and user are described as subjects respectively.

[0298] I. Overall System Composition The server includes a general-purpose processor, storage devices, a network interface, and an optional graphics processing unit (GPU). The general-purpose processor can be a multi-core general-purpose central processing unit, and the GPU can be a parallel processing unit supporting general-purpose computing. Storage devices may include main memory and persistent memory. The network interface is used for communication with multiple terminals via wired or wireless networks.

[0299] A terminal includes a processor, memory, display, and input unit. The terminal can be a smartphone, tablet, portable computing device, or desktop computing device. The terminal communicates bidirectionally with the server through the network protocol stack provided by the operating system.

[0300] The server stores multiple software modules in its storage device for executing the method of the present invention, including: a data receiving module, a data preprocessing module, a feature extraction module, a risk assessment module, a prompt statement generation module, a generative artificial intelligence model invocation module, a result post-processing module, an external service linkage module, and a restriction control module. These modules can be implemented using general-purpose programming languages ​​and rely on data processing libraries (such as data analysis libraries and numerical computation libraries) and machine learning libraries (such as deep learning frameworks) to perform specific data processing and operations.

[0301] II. Data Acquisition and Structured Modeling Users input health-related and emotion-related information through a graphical user interface on a terminal. Health-related information may include age, gender, height, weight, past medical conditions, daily exercise frequency, dietary habits, and sleep patterns. Emotion-related information may include the current emotion category, intensity, and duration. The terminal organizes this input data into a key-value structure and represents it using a hierarchical field structure, for example, assigning a separate field to each type of information. After performing basic validation of the input, the terminal sends it to the server as structured data via a network interface.

[0302] After receiving structured data requests from multiple terminals, the server converts them into a unified tabular internal data structure. During this process, the server uses a data analysis library to map the key-value structures uploaded from different terminals to a data table with a unified column definition, where each row represents an individual data record. The server uses a numerical computation library to perform type conversion, unit conversion, and missing value handling on the numerical fields. For example, the server uniformly converts height to meters, maps gender in text form to standardized category codes, and applies estimates based on historical data to missing weight values.

[0303] During the structured modeling process described above, the server also standardizes and encodes the existing disease descriptions in free text using internal rules and mapping tables, converting them into a predefined set of disease categories. Through this structured modeling, the server unifies the originally heterogeneous input data into a data matrix that can be directly used in machine learning and numerical computation, improving the stability and efficiency of subsequent calculations.

[0304] This unified data modeling approach enables servers to perform parallel processing on data from different individuals in batch or streaming mode, thereby reducing format conversion overhead and exception handling branches at the computer technology level and improving the overall data pipeline's processing throughput.

[0305] III. Body Indicator Calculation and Feature Extraction The server uses a tabular data structure and a numerical calculation library to calculate body indicators from raw measurements. It combines height and weight fields to calculate body mass index (BMI) and other indicators reflecting body shape, mapping these indicators to category labels based on preset thresholds. The server can also statistically analyze the average and standard deviation of indicators such as blood pressure and blood sugar from historical records, and deduce risk predisposition based on the statistical results.

[0306] In the feature extraction module, the server combines basic attributes, disease codes, lifestyle habits, emotional labels, and the calculated physical indicators into a multi-dimensional feature vector. These feature vectors can include continuous numerical features, discrete categorical features, and multi-label features. The server then uses a series of feature transformation rules, including normalization, bucketing, one-hot encoding, and embedding index mapping, to uniformly convert these features into numerical tensors that are adapted to the input format of machine learning models.

[0307] Through this feature extraction process, the server compresses raw health and emotion information into feature representations with higher information density, enabling subsequent models to perform efficient reasoning in a lower-dimensional space, reducing memory usage and computational load. This feature extraction creates a structured, reusable feature space within the computer, which significantly improves the accuracy of risk assessment and suggestion generation compared to traditional methods that directly use raw numerical values ​​and rule matching.

[0308] IV. Structure and Calculation of Risk Assessment Model The server runs a pre-trained machine learning model in the risk assessment module to output quantitative results of disease risk and lifestyle risk. This machine learning model can employ a feedforward neural network structure, containing multiple fully connected layers and non-linear activation functions. The server inputs the aforementioned feature vectors into the model, performing matrix operations and non-linear transformations layer by layer from the input layer, first-level hidden layer, second-level hidden layer, to the output layer.

[0309] During model training, the server uses a large number of labeled historical health records as training data, with each record including known disease occurrences and lifestyle habit assessments. The server employs supervised learning methods, using a loss function to measure the difference between the model output and the true labels, and continuously updates the network weights using gradient descent-like optimization algorithms. The server can use regularization and parameter constraints to prevent overfitting, and selects the optimal model structure and hyperparameters through cross-validation.

[0310] During the inference phase, the server performs a forward propagation on the new input data based on the final network weights, obtaining multiple risk scores. These scores can include cardiovascular risk scores, metabolic risk scores, and health risk scores based on lifestyle habits. The server further converts these scores into low, medium, and high risk level labels.

[0311] Compared to traditional scoring methods based on fixed rules, this risk assessment method based on multi-layer neural networks can automatically learn complex feature combinations and nonlinear correlations from large-scale data, improve the ability to identify marginal cases and complex interaction factors, and achieve improved prediction accuracy and reduced error at the computer technology level.

[0312] V. Construction of Prompt Statements and Invocation of Generative Artificial Intelligence Models In the prompt statement generation module, the server converts structured information and risk assessment information into natural language descriptions. Following a standardized text template, the server combines age, gender, major diseases, body mass index category, risk level, and lifestyle characteristics into a health status description text. The server then appends task instructions to this description text to form a complete prompt statement.

[0313] For example, the server can generate the following prompts for input into a generative artificial intelligence model: "User age: 45 years old; Gender: Male; Medical history: Hypertension; Body mass index: 27.5 (overweight); Cardiovascular risk assessment: moderately elevated; Metabolic risk assessment: mildly elevated. As an expert in preventive medicine and nutrition, please generate a personalized nutritional intake recommendation and exercise plan based on the above information, and provide general advice on medication use when medical attention is needed. Please answer in Simplified Chinese and list the daily dietary recommendations and weekly exercise schedule in item form." In psychological stress management scenarios, the server can generate additional prompts, such as: "User age: 38 years old; Gender: Male; Current mood: tense and anxious; Recent work pressure is high, sleep time is less than 6 hours; Cardiovascular risk assessment: moderate; Metabolic risk assessment: mild. Based on the above health and mood information, please provide a stress management suggestion aimed at reducing mental stress and improving sleep, including daily routine adjustments, relaxation training and exercise suggestions." By automatically constructing prompts based on structured data and risk assessment results, the server aligns the input of the generative AI model more closely with its internal feature space, thereby improving the relevance and stability of the generated results, reducing irrelevant content and logical jumps, and ultimately enhancing the effectiveness of model reasoning and output quality at the algorithm level.

[0314] The server loads a pre-tuned sequence generation model in the generative AI model invocation module. This model can employ a multi-layer self-attention network structure, including encoding and decoding sub-modules. The server converts the prompt statement into a sequence of lexical units, maps it to a vector space via a word embedding layer, and then processes it sequentially through a multi-head self-attention layer, a feedforward network layer, and a normalization layer. During the generation phase, the server iteratively generates the next lexical unit based on the given prompt statement's encoding state until a termination condition is met.

[0315] During model training and fine-tuning, the server uses a large-scale corpus of medical texts and health advice texts as training targets, employing an autoregressive language modeling loss function and updating model weights through gradient backpropagation. The server can also employ data augmentation strategies, such as synonym substitution, structural restructuring, and multi-perspective narration, to improve the model's robustness to different expression methods. Through this continuous learning and fine-tuning, the server enables the generative AI model to achieve high language fluency and medical relevance in the health advice generation task.

[0316] VI. Post-processing and visualization of generated results In the post-processing module, the server parses the natural language text output by the generative artificial intelligence model. Using text segmentation rules and keyword recognition, the server divides the original text into different categories such as nutrition recommendations, exercise plan recommendations, medication use recommendations, and stress management recommendations, and further breaks down each category into a list of items. The server can simplify or remove redundancy from the text as needed to adapt it for terminal display.

[0317] The server then combines the structured recommendations with visual configurations to generate a data structure suitable for terminal display. Upon receiving this data, the terminal displays various recommendations in a modular layout on its screen. High-risk recommendations can be highlighted with colors or icons, and timelines or progress bars can be added to help users understand the execution schedule.

[0318] This post-processing workflow, from generated text to structured display, enables the generated results to be presented in a clearer and more executable form, reducing the burden on users to understand, and optimizing the user interaction experience through collaborative front-end interface design and back-end text parsing.

[0319] VII. External Food Service Coordination and Actual Implementation In the external service linkage module, the server automatically extracts food categories, nutritional requirements, and limitations from nutritional intake recommendations. The server can define a food category mapping table to map food names in text to standard food categories, and then match these categories with product data provided by external dietary services. Based on the matching results, the server generates a candidate product list and returns it to the terminal in a structured format.

[0320] After receiving candidate product information, the terminal associates product names, nutrition labels, prices, and other information with corresponding suggested items, allowing users to browse available products while reading the suggestions. Once the user selects and confirms a product on the terminal, the terminal generates an order request and sends it to the server. The server then converts the request into a format acceptable to external service interfaces and submits it via a network interface. In this way, the server not only generates suggestions but also transforms them into directly executable real-world operations through a computational process.

[0321] This linkage method creates a closed loop between the server's internal calculation results and real-world dietary behavior. Technically, it is reflected in the combination of backend data matching algorithms and external interface calls, reducing the number of manual search and configuration steps for users across different systems and achieving end-to-end automated execution.

[0322] 8. Restricting and controlling the intake of emotionally driven addictions In the restriction control module, the server calculates the intake limits for each addiction based on emotion-related information and risk assessment information. The server can set a basic safety threshold for each addiction and dynamically adjust this threshold based on risk scores and emotional states. For example, when the cardiovascular risk score is high and the emotional state is tense or anxious, the server lowers the acceptable upper limit of addiction intake or raises the prohibited level.

[0323] When the server needs to interpret or customize the constraints, it will construct a specific prompt statement to request a generative artificial intelligence model to generate constraint adjustment suggestions. For example, the server can generate the following prompt statement: "User's current emotional state: tense with anxiety; cardiovascular risk assessment: moderately high; past history of high-salt diet and frequent alcohol consumption. Based on the above information, please provide specific recommendations on limiting alcohol intake and controlling high-salt food intake, including the recommended maximum number of drinks per week, the approximate amount consumed each time, and how to gradually reduce the frequency of drinking." After receiving suggested adjustments, the server combines them with numerical limits to update the addiction intake rules maintained internally. The server can then synchronize these rules to the user's device, providing alerts or warnings to the user, thus creating a control mechanism that adaptively adjusts based on the calculation results.

[0324] This emotion-driven dynamic constraint control not only relies on the language generation capabilities of generative artificial intelligence models, but also achieves automatic conversion from text suggestions to parameter updates through the rule engine and numerical condition calculations inside the server. This makes the whole process a programmable and adjustable feedback loop, thereby enhancing the system's flexibility and control precision in response to changes in user state.

[0325] IX. Technical Effects and Improvements in Computer Technology Through the collaborative design of the aforementioned multi-layered data preprocessing, feature extraction, risk assessment, prompt statement construction, generative artificial intelligence model reasoning, and result structured processing, the server achieves specific improvements to computer technology in the following aspects: By using a unified data structure and preprocessing process, the server reduces the computational burden caused by format compatibility and exception handling, enabling large-scale user data to be processed in batches at higher throughput, thereby improving the overall processing speed.

[0326] The server significantly reduces noise and bias in the generated results through a risk assessment model based on multi-layer neural networks and a high-quality prompt generation process, thereby improving the accuracy of disease risk prediction and the relevance of the suggestions, and reducing misjudgments caused by simple rules.

[0327] By explicitly encoding risk assessment results and physical indicators in the prompt statements, the server enables the generative AI model to focus more on medical-related information in its internal attention mechanism, thereby increasing the model's attention to key health factors and effectively constraining the direction of text generation.

[0328] The server separates feature extraction and text generation through a modular data pipeline and uses well-defined data structures in the middle layer, which facilitates replacement or upgrades between different models and algorithms, improving the maintainability and scalability of the system.

[0329] Through external service linkage modules and restriction control modules, the server enables the calculation results to directly influence real-world dietary choices and addiction behaviors. Technically, this is manifested as a tight coupling between internal computer data flow and external execution actions, going beyond the scope of simple information display and achieving a deeper level of device control and behavior guidance.

[0330] 10. Other Implementation Methods In different implementations, the server can employ different types of machine learning models, such as using gradient boosting decision trees as a risk assessment model, or using sequence generation models of varying sizes as generative artificial intelligence models. The server can also choose to run a smaller-scale model using only the central processing unit, or run a large-scale model to achieve higher accuracy when a graphics processing unit is available, depending on resource availability.

[0331] In another implementation, the server can also introduce an incremental learning mechanism to retrain the risk assessment model and the generative artificial intelligence model by periodically collecting user feedback data and execution status, thereby continuously optimizing system performance over time.

[0332] Through the above implementation, the server, terminal, and user form a close collaboration in the stages of data acquisition, calculation, generation, and execution. This invention not only provides application solutions in the field of health management, but also brings substantial technical improvements to the design of data structures, algorithm processes, and model calling methods within the computer.

[0333] use Figure 13The processing flow is explained.

[0334] Step 1: Users input health-related and emotion-related information using the terminal.

[0335] Users enter health-related information such as age, gender, height, weight, past medical conditions, daily exercise frequency, dietary habits, and sleep patterns on the terminal's graphical interface, and select or fill in emotion-related information such as the current emotion type (e.g., tension, anxiety, relaxation), emotion intensity, and duration.

[0336] Input: Raw text and numerical inputs entered by the user on the terminal interface.

[0337] Output: A set of key-value pair structured data formed internally by the terminal, used to represent the user's health-related information and emotion-related information.

[0338] Step 2: The terminal performs basic validation on the input data and packages it into a structured request.

[0339] The terminal checks whether required fields are complete, whether numeric fields are valid, and whether option fields are within the preset option range. If any missing or invalid values ​​are found, the terminal prompts the user to correct them on the interface. After successful verification, the terminal organizes all fields into a structured data object with uniform field names and adds a timestamp and device identifier to it.

[0340] Input: Key-value pair data entered by the user in step 1 and temporarily stored in the terminal memory.

[0341] Output: A structured data object with validated and normalized field names, ready to be sent to the server over the network.

[0342] Step 3: The terminal sends structured data to the server.

[0343] The terminal constructs a request message containing structured data using the network protocol stack provided by the operating system. The request header identifies the data type and encoding method, and the request is sent to the server's predetermined network address via an encrypted communication channel. Before sending, the terminal serializes the request content, converting the internal structured objects into a network transmission format.

[0344] Input: Structured health-related and emotion-related objects.

[0345] Output: The request message carried on the communication line, a structured data stream that can be parsed by the server.

[0346] Step 4: The server receives and parses structured data from the terminal.

[0347] After receiving a request message at the network interface, the server passes the message content to the application. The application parses the transmission format and restores it into a structured data object that it can process internally. The server verifies the data integrity and basic format, such as checking required fields and data types, and writes the data to a temporary buffer in memory after verification.

[0348] Input: A structured data stream that arrives at the server via network transmission.

[0349] Output: Structured data objects stored in the server's memory for subsequent data preprocessing.

[0350] Step 5: The server converts structured information into tabular data.

[0351] In the data preprocessing module, the server maps structured data objects to data tables represented by rows and columns. Each column corresponds to a standardized field, such as age, gender, and height, and each row corresponds to a user record. The server merges or splits multi-valued fields (such as multiple past illnesses) into multiple binary columns and maps text category fields to standard category labels, thus obtaining a unified tabular data structure.

[0352] Input: Structured health-related and emotion-related information objects in the server's memory.

[0353] Output: Tabular data containing one or more rows of records, used for numerical calculations and feature extraction.

[0354] Step 6: The server performs numerical calculations to determine body metrics.

[0355] The server uses a numerical calculation program to read numerical fields such as height and weight from the table data, calculates body indicators such as body mass index through arithmetic operations, and constructs other combined indicators based on preset formulas and parameters, such as calculating the blood pressure level index based on historical average blood pressure. During the calculation process, the server performs unit conversions and outlier pruning to ensure the stability of the calculation results.

[0356] Input: Raw numerical fields in tabular data (such as height, weight, historical measurements).

[0357] Output: Tabular data containing newly added body metric columns, such as the added Body Mass Index (BMI) column and its corresponding numerical results.

[0358] Step 7: The server extracts features based on structured information and body indicators.

[0359] The server selects fields meaningful for risk assessment from the tabular data, normalizes continuous numerical fields, encodes discrete categorical fields, and converts sentiment-related information into quantitative or hierarchical features according to rules. The server combines these processed fields into feature vectors, generating a set of multidimensional numerical features for each user, which are stored in memory in a unified matrix format for use by the risk assessment model.

[0360] Input: Data in tabular form containing raw fields and calculated body metrics.

[0361] Output: A set of feature vectors representing the health and emotional states of each user, i.e., the numerical feature matrix used as input to the model.

[0362] Step 8: The server uses machine learning programs to perform risk assessment.

[0363] The server inputs the feature matrix into a pre-trained risk assessment model, and calculates disease risk and lifestyle risk scores through multi-layer parameter operations. Following the model structure, the server sequentially performs operations such as weighted summation and non-linear activation from the input layer to the output layer to obtain the probability or score of various risks, and then maps the continuous scores to risk level labels.

[0364] Input: A feature matrix containing multidimensional features.

[0365] Output: Risk assessment information for each user, including risk score vector and risk level label.

[0366] Step 9: The server generates health status information.

[0367] The server combines structured information and risk assessment information into a natural language description, selects key fields (such as age, gender, major pre-existing conditions, body mass index, and risk level), and generates a summary description text according to a preset text template. During the generation process, the server performs conditional checks, adding the corresponding description only when a specific disease or high-risk label is present, making the description content individualized and focused.

[0368] Input: Structured health-related information, emotion-related information, and corresponding risk assessment information.

[0369] Output: Health status description text written in natural language, used for constructing subsequent prompt statements.

[0370] Step 10: The server provides prompts for constructing generative artificial intelligence models.

[0371] In the prompt generation module, the server uses health status information as background and appends task descriptions, output requirements, and constraints to form a complete prompt. Depending on the usage scenario (e.g., nutrition and exercise recommendations, stress management recommendations, and addiction restriction / adjustment recommendations), the server selects different templates and instructions to guide the generative AI model in generating corresponding types of output.

[0372] Input: Health status description and current task type (e.g., nutrition advice, stress management).

[0373] Output: Complete prompt text for generative artificial intelligence models, used to drive the natural language generation process.

[0374] Step 11: The server inputs prompts into the generative artificial intelligence model and obtains suggested information.

[0375] In the model generation module, the server converts the prompts into internal vector representations, which are then passed to the generative AI model to perform multi-step inference based on model parameters to generate response text. During the generation process, the server balances the length and diversity of the model output by setting a maximum generation length and control parameters. After generation, the server recovers the natural language text from the model output sequence, obtaining suggested information including nutritional intake recommendations, exercise plan suggestions, and medication usage recommendations.

[0376] Input: The prompt statement generated in step 10.

[0377] Output: Natural language suggestion text output by the generative artificial intelligence model.

[0378] Step 12: The server categorizes and structures the suggested information.

[0379] The server performs segmentation and classification on the generated text, identifying different types of suggestion blocks through keywords, delimiters, or serial numbers. It extracts content related to nutrition intake, exercise plans, and medication use, and breaks each category down into a list of entries. The server associates these entries with the original fields, giving them clear category labels in the data structure for easy display on the terminal and subsequent processing.

[0380] Input: The original suggestion information text output by the generative artificial intelligence model.

[0381] Output: A structured suggested dataset based on category partitioning, containing multiple category fields and a set of entries.

[0382] Step 13: The server extracts food categories and dietary menu candidates from nutritional intake recommendations.

[0383] In the external service linkage module, the server scans nutrition advice text, identifies words related to food and dietary structure, compares them with a predefined food category mapping table, and determines the recommended food categories and dietary patterns. Based on the extraction results, the server generates an internally represented list of food categories and dietary menu candidates for querying product information from external dietary services.

[0384] Input: Text entries for categorized nutrition intake recommendations.

[0385] Output: A list of food categories and a set of candidate menu items for matching against an external service goods library.

[0386] Step 14: The server works in conjunction with external food service providers to obtain information on candidate products.

[0387] Based on the food category list and dietary menu candidates, the server sends a query request to an external dietary service interface, requesting product data that matches the recommended food categories. The server then filters the returned results for products that meet nutritional requirements and limitations, organizing fields such as product name, basic nutritional information, and price into structured candidate product information for synchronized display on the terminal along with the suggested content.

[0388] Input: A list of recommended food categories and a set of candidate menus.

[0389] Output: A set of external candidate product information corresponding to the nutritional intake recommendations.

[0390] Step 15: The server calculates the restrictions on addiction intake based on emotion-related information and risk assessment information.

[0391] The server reads emotional state, emotional intensity, and related risk scores from the restriction control module, maps these factors into adjustment coefficients according to preset rules, and weights the upper limit of basic craving intake. The server calculates the maximum acceptable intake of cravings per day or week and the prohibited situations, and stores the currently effective restrictions as parameters for subsequent reminders and control.

[0392] Input: Emotion-related information and corresponding disease risk and lifestyle risk assessment information.

[0393] Output: A set of parameters for limiting the intake of addictive substances, calculated based on the current state.

[0394] Step 16: The server generates additional prompts and obtains adjustment suggestions for the restrictions.

[0395] When the server needs to explain or adjust restrictions on addictions, it constructs a specific prompt statement based on the latest emotional state and restriction parameters. This prompt describes the current restriction status and user behavior patterns in natural language and instructs the generative AI model to provide suggestions on how to adjust or gradually strengthen the restrictions. The server inputs this prompt statement into the generative AI model to obtain a suggested restriction adjustment text containing the reasons for the restriction adjustment and the execution steps.

[0396] Input: Current addiction intake restrictions and emotional state information.

[0397] Output: Text suggesting adjustments to addiction restrictions from the generative artificial intelligence model.

[0398] Step 17: The server updates the restrictions on addiction intake and generates a final set of recommendations.

[0399] The server parses the restriction adjustment suggestion text, extracts specific items regarding increasing or decreasing intake limits, modifying frequency, or introducing phased plans, and converts these items into new parameter values ​​to update the internal restrictions. The server combines the updated restriction parameters with health advice, stress management advice, and candidate product information to form a unified final suggestion set for the user.

[0400] Input: Suggested text for adjusting restrictions and existing limits on the intake of addictive substances.

[0401] Output: A comprehensive recommended dataset including the updated constraints.

[0402] Step 18: The server converts the suggested data into visual information and sends it to the terminal.

[0403] Based on the terminal's display capabilities and interface layout requirements, the server generates an adapted display structure from the comprehensive suggestion data, including module division, item sorting, and highlighting strategies. The server assigns display blocks to different categories of suggestions and adds risk labels and execution priority prompts when necessary. The server sends the visualization information to the terminal via a network interface, ensuring that the terminal can directly use this structure for rendering.

[0404] Input: Comprehensive suggestion dataset (nutrition, exercise, medication, stress management, constraints, candidate product information).

[0405] Output: A terminal-oriented visual data structure containing multiple display modules and entry configurations.

[0406] Step 19: The terminal receives visual information and displays suggestions and candidate products on the interface.

[0407] After receiving the visual data structure from the server, the terminal draws the corresponding interface elements on the display, presenting nutrition advice, exercise plans, medication precautions, stress management suggestions, and addiction restrictions in a zoned format. The terminal displays a list of candidate products in the area corresponding to the nutrition advice, allowing users to browse the suggestions while simultaneously viewing available products.

[0408] Input: A visual data structure sent by the server.

[0409] Output: A graphical interface displayed on the terminal screen, including text suggestions, icons, and a list of candidate products.

[0410] Step 20: Users view the suggestions and perform actions on the terminal.

[0411] Users can read various suggestions on the terminal interface, choose whether to adopt them based on their personal circumstances, and click on candidate products to view and add them to their order. Users can also adjust their behavior based on the addiction restriction prompts and record the implementation status or feedback information when necessary.

[0412] Input: The suggested interface and candidate product information presented by the terminal.

[0413] Output: User-made behavioral decisions and operational instructions (such as product selection, order confirmation, marking as executed, etc.).

[0414] Step 21: The terminal sends the user's operation results back to the server.

[0415] After detecting a user's selection of candidate products, order confirmation, or marking of execution status, the terminal converts these actions into structured feedback data and sends it to the server. The feedback data includes the user's actual selected products, execution frequency, and preference information, which is used by the server for subsequent analysis and model optimization.

[0416] Input: The user's specific operation record on the terminal.

[0417] Output: Feedback data and actual behavior records sent to the server.

[0418] Step 22: The server records feedback data and uses it for continuous model optimization.

[0419] After receiving feedback data from the terminal, the server stores it in a historical record database in a persistent storage device to track user acceptance and implementation of suggestions. During offline training or periodic update phases, the server uses this feedback data as new training samples or evaluation benchmarks to retrain or adjust the parameters of the risk assessment model and the generative artificial intelligence model, thereby continuously improving the model's prediction accuracy and suggestion quality in real-world usage environments.

[0420] Input: User behavior feedback data and order execution records.

[0421] Output: The updated state of the historical data storage and the new training dataset that can be used in subsequent training.

[0422] Application Example 2 The process flow corresponding to the specific processing in Use Case 2 will be described below. The various parts of the system described below are implemented by the data processing device 12 and the intelligent device 14. In addition, the data processing device 12 is referred to as the "server" and the intelligent device 14 is referred to as the "terminal".

[0423] While existing computer systems generate dietary or medication recommendations based on health information, several technical problems remain: First, servers typically only perform simple rule-based judgments on static health data, lacking the ability to uniformly model and extract features from physiological measurements and lifestyle activity information that change over time. This results in recommendations being insensitive to individual state changes, low system processing capacity for time-series data, and inefficient utilization of computational resources. Second, servers often fail to integrate multimodal psychological information such as speech, images, and text, relying solely on a single emotion source. This leads to limited emotion recognition accuracy and an inability to quantify emotions such as "stress" at the computational level, hindering precise control over the input to generative AI models. Third, existing systems often use fixed or manually written prompts when calling generative AI models, failing to dynamically construct prompts based on individual physiological characteristics, emotional characteristics, and preventive medicine constraints. This results in generated results that are inconsistent with individual risk characteristics and... The lack of strong constraints between resource constraints leads to poor controllability of model output, excessive repetitive calls and manual corrections, increasing the computational load and response latency on the server side. Fourth, the interaction between the server and external food delivery services is usually limited to simple link jumps, failing to structure the output of the generative AI model into a data structure that can be directly used to build order data. This results in multiple parsing and transformations at the application layer, increasing the complexity and potential failure points of the data processing chain. Fifth, existing technologies often separate user emotions from drinking advice, lacking a unified computational process for automatically generating alcohol intake restriction prompts based on emotional characteristics and adjusting them through the generative AI model, making it difficult to achieve adaptive optimization of alcohol intake restriction strategies. Sixth, existing systems lack a closed-loop update mechanism driven by user feedback. Servers generally do not dynamically optimize prompt templates and constraints based on user evaluation information, thus failing to continuously improve generative AI model calling strategies and resource scheduling strategies at the computer level.

[0424] Therefore, a new computer implementation is needed to enable servers to: automatically preprocess individual physiological and psychological information and extract numerical and emotional features; automatically generate dynamic prompts for generative artificial intelligence models based on preventive medicine knowledge; incorporate alcohol restriction strategies into the generation and adjustment process within the same computational flow; and iteratively update prompts and constraints based on user feedback, thereby improving data processing efficiency, model call controllability, and the ability to interact with external services from a computer technology perspective.

[0425] The specific processing performed by the specific processing unit 290 of the data processing apparatus 12 in Application Example 2 is achieved by the following means.

[0426] In this invention, the server includes a device for acquiring individual physiological information and individual psychological information; a device for preprocessing the individual physiological information and individual psychological information to generate numerical features and emotional features; a device for setting nutritional intake constraints, drug use constraints, and target conditions based on preventive medicine according to the numerical features and emotional features; a device for generating prompt statements for a generative artificial intelligence model based on the constraints and target conditions and in combination with the individual physiological information and individual psychological information, and inputting the prompt statements into the generative artificial intelligence model to generate suggestion information containing nutritional intake suggestions and drug use suggestions; a device for generating food delivery service information based on the nutritional intake suggestions contained in the suggestion information and communicating with external food service providers to generate order information; a device for determining alcohol-related restrictions based on the individual psychological information and emotional features, generating prompt statements corresponding to the restrictions, inputting them into the generative artificial intelligence model, and adjusting the alcohol-related restrictions based on the output results; and a device for updating the prompt statements and constraints based on user evaluation information from user terminals. This allows for the formation of a data processing pipeline on the server side, centered on feature extraction, constraint setting, automatic generation of prompts, generative AI model invocation, result structuring, external service linkage, and feedback-driven updates. This enables the server to efficiently process time-series health data and multimodal sentiment data, automatically construct prompts consistent with individual risk and resource constraints, reduce the number of trial-and-error steps in model invocation and human intervention, optimize computing resource utilization, shorten response time, and directly output structured data usable for order generation when collaborating with external food delivery services. Ultimately, this improves the overall performance and scalability of the health management advice generation system at the computer technology level.

[0427] "Individual physiological information" refers to various types of data related to the physical condition of a specific individual, including but not limited to physiological measurement data such as weight, height, blood pressure, heart rate, body temperature, sleep time, and steps, as well as other physical condition data obtained by wearable devices, medical devices, or user input.

[0428] "Individual psychological information" refers to various types of data that reflect the psychological or emotional state of a specific individual, including but not limited to voice information, image information, text information, and subjective emotional descriptions and behavioral pattern data collected by questionnaires, interactive interfaces, or sensors.

[0429] "Preprocessing" refers to the process of performing at least one of the following on the acquired raw data: format conversion, missing value imputation, outlier handling, normalization, standardization, feature extraction, and dimensionality reduction, in order to generate structured data that can be used for subsequent analysis or model input.

[0430] "Numerical features" refer to the numerical values ​​of features obtained by preprocessing and calculating individual physiological information, including but not limited to mean, variance, maximum value, minimum value, trend coefficient, exponential index, and embedding vectors obtained after model transformation.

[0431] "Emotional characteristics" refer to the feature quantities used to characterize emotional states obtained through the analysis and evaluation of an individual's psychological information, including but not limited to emotion category labels, emotion intensity scores, pleasure, arousal, stress levels, and their corresponding vectorized representations.

[0432] "Nutritional intake constraints" refer to the rules or parameter ranges set to limit or regulate the intake of food energy, salt, sugar, fat, protein and other nutrients, based on the principles of preventive medicine and individual physiological information and emotional characteristics.

[0433] "Drug use constraints" refer to the rules or parameter ranges set to regulate the types, dosages, frequencies, and interactions of drugs, based on the principles of preventive medicine and individual health status, past medical history, and current medication use.

[0434] "Target conditions" refer to the health-related goals that the system expects to achieve when generating suggestions, including but not limited to weight control, blood pressure control, blood sugar control, sleep improvement, stress relief, and corresponding quantitative or qualitative indicators.

[0435] "Generative AI models" refer to AI models that can automatically generate text, structured data, or other forms of output based on input prompts, including but not limited to deep learning-based language models, sequence-to-sequence models, and other generative models.

[0436] "Prompt statements" refer to the instructional text input into the generative artificial intelligence model. The instructional text describes individual physiological information, individual psychological information, constraints, and target conditions, and guides the generative artificial intelligence model to generate output content that meets the requirements.

[0437] "Suggested information" refers to individual-oriented recommendations output by generative artificial intelligence models based on prompts, including but not limited to nutritional intake recommendations, medication use recommendations, lifestyle recommendations, and health-related text descriptions.

[0438] "Information for food delivery services" refers to data constructed for interacting with external food service providers, including but not limited to dish names, ingredient categories, portion sizes, nutrition labels, merchant logos, and other parameters required to generate an order.

[0439] "External food service" refers to service systems that receive food orders and arrange for their preparation and delivery through online platforms, including but not limited to online food ordering platforms, supermarket delivery systems, and other food delivery systems.

[0440] "Order information" refers to structured data generated for food delivery in the service of external food providers, including but not limited to user identification, delivery address, menu list, quantity, price, delivery time, and payment-related information.

[0441] "Drinking-related restrictions" refer to a range of rules or parameters set based on individual physiological information and emotional characteristics to limit the types of alcoholic beverages, the frequency of drinking, the amount of alcohol consumed, or to recommend avoiding drinking.

[0442] "Display data" refers to the interface data generated for displaying suggested information, drinking restrictions, order information, etc. on user terminals, including but not limited to text, icons, list structures, and layout information for graphical user interfaces.

[0443] "User terminal" refers to an electronic device operated by a user to interact with a server, including but not limited to smartphones, tablet computers, personal computers, wearable devices, and other terminal devices with display and communication functions.

[0444] "User feedback information" refers to data provided by users through their terminals after viewing or using suggested information and related services, used to evaluate the quality of the system's output. This includes, but is not limited to, ratings, likes or dislikes, subjective comments, and behavioral data such as whether suggestions were adopted or orders were placed.

[0445] In a preferred embodiment of the present invention, the server, terminal, and user together constitute the basic hardware environment for implementing the system of the present invention. The server can be a computing device deployed in a data center or cloud computing platform, including a central processing unit, storage device, and optional graphics processing unit. The terminal can be a smartphone, tablet computer, or personal computer, equipped with a display unit, input unit, microphone, camera, and wireless communication module. The user interacts with the server through the terminal.

[0446] At the software level, a server includes an operating system, a network communication module, a database management system, and application services. The server can use a general-purpose server operating system and either a relational or non-relational database to store user data. The server's application services are preferably implemented using a programming language, combined with data processing libraries, numerical computation libraries, and machine learning frameworks. The server can use data processing libraries to preprocess individual physiological and psychological information; use numerical computation libraries to perform statistical operations and feature transformations; and use machine learning frameworks to build and run generative artificial intelligence models and sentiment analysis models.

[0447] At the software level, a terminal includes its operating system, communication module, and user interface program. The user interface program can be a native application or a web application, responsible for presenting interface elements, collecting user input, utilizing local sensors (such as microphones and cameras), and exchanging data with the server through a secure communication channel.

[0448] Users can fill in or update health information and subjective emotional descriptions on the interface through the terminal. Users can input physiological data such as age, gender, height, weight, past medical history, allergy information, daily exercise frequency, and sleep status, and can also input text descriptions such as "I've been under a lot of work pressure lately and often stay up late" or "I feel a bit tired and depressed today." With authorization, users can also allow the terminal to connect to wearable devices via Bluetooth to automatically acquire physiological measurement information such as heart rate, steps, and sleep duration, and allow the terminal to use a camera and microphone to collect facial images and voice signals.

[0449] In this embodiment, the terminal is responsible for the initial formatting of user input data and sensor data. The terminal can package various types of data into structured data objects, using predetermined field formats to describe user identifiers, timestamps, physiological measurements, subjective emotion text, audio data identifiers, image data identifiers, etc. The terminal establishes a secure connection with the server through an encrypted transmission protocol and sends the structured data to the server.

[0450] After receiving data from the terminal, the server persistently stores the raw data in the storage device. The server can write physiologically relevant data to the "Physiological Measurement Record Table," subjective emotional text, audio references, and image references to the "Psychological Information Record Table," and basic user information to the "User Profile Table." The server effectively manages data at different points in time through a database indexing mechanism, making subsequent query and aggregation operations highly efficient, thereby reducing database access latency in large-scale user scenarios.

[0451] When preprocessing individual physiological information, the server uses a data processing library to build data tables, aggregating various physiological indicators from wearable devices, manual input from terminals, and historical records in chronological order. The server can calculate numerical features including weight, height, body mass index, mean blood pressure, extreme blood pressure, heart rate statistics, moving average of steps, and rolling average of sleep duration. The server can also normalize or standardize these features to generate numerical feature vectors suitable for subsequent model input. This preprocessing helps eliminate dimensional differences, making the training and inference processes of deep learning models more stable, reducing the risk of gradient explosion or gradient vanishing, thereby technically improving the model's convergence speed and prediction accuracy.

[0452] When preprocessing individual psychological information, the server uses natural language processing methods to segment, tokenize, and vectorize text information, mapping the text to a fixed-dimensional semantic vector space. For speech information, the server first generates text using a speech recognition module and then inputs the text into the sentiment analysis model. For image information, the server extracts facial expression features using a convolutional neural network structure and maps them into vectors representing emotion categories and intensities. The server then weights and fuses the text sentiment vectors, speech sentiment vectors, and image sentiment vectors in the feature space to generate unified sentiment features. Through this multimodal fusion approach, the server reduces the impact of single-modal noise on the sentiment recognition results at the computational level, reducing sentiment judgment errors and thus making the sentiment conditions for generating subsequent prompts and suggestions more accurate.

[0453] After obtaining numerical and emotional characteristics, the server constructs nutritional intake constraints, medication use constraints, and target conditions based on pre-defined preventive medicine rules and individual health records. The server can use a rule engine or specific logic modules to set different upper and lower limits for risks such as hypertension, abnormal glucose metabolism, obesity, and sleep deprivation. For example, it can limit daily salt intake, control simple carbohydrate intake, and lower the upper limit for saturated fat intake. The server can also set restrictions on drinking behavior based on indicators such as stress levels and pleasure levels in emotional characteristics, such as completely prohibiting drinking, limiting drinking frequency, or suggesting only small amounts of alcohol in specific situations. By uniformly generating constraints and target conditions on the server side, the system tightly binds health risk assessment with suggestion generation. This technically constrains the potential output space before the generative AI model is invoked. This pre-filtering mechanism reduces the generation of invalid or non-compliant content, lowers subsequent filtering overhead, and improves overall computational efficiency.

[0454] When generating prompts for generative AI models, the server uses template and parameter filling mechanisms to combine user profiles, numerical features, emotional features, constraints, and target conditions into natural language text. The server maintains different prompt templates for different task types (e.g., "one-meal dietary suggestions," "weekly diet plan," "alcohol restriction instructions") and dynamically selects the appropriate template and fills in the current user's feature values. The prompts generated by the server may include the following examples: "Please act as a preventive medicine and nutrition expert and generate a detailed dinner recommendation based on the following user information. User: 50-year-old male with a history of hypertension and a high BMI. Average blood pressure over the past week: 148 / 92; average steps: 4000; average sleep duration: 5.8 hours. Emotional state: Based on voice and facial expression analysis, the user is judged to be under high stress, subjectively describing it as 'recently experiencing significant work stress and frequently staying up late.' Constraints: 1. Dinner must be low in salt and saturated fat, avoiding fried foods and sugary drinks; 2. Due to the user's current high stress level, please do not recommend any alcoholic beverages; sugar-free drinks or herbal teas are acceptable; 3. Provide 2-3 common food types that can be found on food delivery platforms, along with brief descriptions of their health benefits; 4. Use simplified Chinese language, geared towards general users, and a friendly tone." "Based on the following user's health and emotional information, generate a health recommendation regarding alcohol intake restriction. User: 40-year-old male, family history of hypertension, occasional alcohol consumption. Recent indicators: high blood pressure, insufficient sleep. Emotional state: Emotional engine indicates high stress. Requirements: 1. Explain the dangers of drinking alcohol under high stress and high blood pressure risk; 2. Recommend avoiding alcohol as much as possible this week, and provide at least three non-alcoholic relaxation methods; 3. Use a simple, non-threatening tone." When the server invokes the generative AI model, it employs a pre-trained language generation model. This generative AI model can be built upon a multi-layer self-attention network structure, including word embedding layers, multi-layer encoder-decoder modules, feedforward network layers, and an output layer. The server trains this model using a large corpus of health management and nutrition texts, employing supervised learning for the next word prediction task. The server evaluates the difference between the predicted and true sequences using a cross-entropy loss function and updates the model parameters using gradient descent and adaptive learning rate optimization algorithms. During training, the server can use data augmentation methods, such as rewriting semantically equivalent sentences, to improve the model's robustness to different prompt sentence structures. In the inference phase, the server inputs the aforementioned prompt sentences into the generative AI model. The model uses a multi-head attention mechanism to integrate health parameters and constraint information from the prompt sentences, generating an output sequence word-by-word containing nutritional intake suggestions, medication usage recommendations, or alcohol restriction instructions.

[0455] After receiving the model output, the server performs rule checks and structuring on the output text. The server can use keyword recognition and pattern matching methods to extract elements such as dish name, ingredient category, cooking method, and nutritional attributes from the generated text. The server maps these elements to an internally unified dish data structure and compares them with pre-maintained dish templates in the database. If the output contains ingredients that do not match the user's allergy information or constraints, the server can replace them with rules or re-invoke the generative AI model to obtain new suggestions that meet the conditions. This reduces the impact of erroneous model output on the user and minimizes manual screening on the user side, achieving automated control of the model output quality from a technical perspective.

[0456] When generating information for food delivery services, the server maps extracted dish names and attributes to the menu database of external food service providers. The server can maintain a mapping table that associates internal dish identifiers with dish identifiers on different external platforms. Based on user location information, platform availability, and time constraints, the server selects a suitable service provider and generates a set of order parameters including a list of dishes, quantities, delivery address, and expected delivery time. The server communicates with external food service providers via a standardized application programming interface (API), sending the order parameters as data packets. Because the server internally handles the conversion from natural language suggestions to structured order data, external services can directly use this data to create orders, avoiding repetitive text parsing across different terminals or services. This reduces the data conversion burden between different systems and lowers communication and parsing overhead.

[0457] When processing alcohol-related restrictions, the server first sets basic restriction rules based on indicators such as stress level and emotional polarity in emotional characteristics. For example, under high stress, the upper limit of alcohol intake is reduced to zero or near zero. The server then generates a prompt statement corresponding to the restriction and obtains specific health instructions and alternative behavior suggestions through a generative artificial intelligence model. Based on the descriptions of drinking frequency, amount of alcohol consumed, and alternative behaviors in the output text, the server further refines and adjusts the internally stored restriction parameters to ensure that they meet both medical risk control requirements and have good user acceptability. Through this two-way mechanism, the server technically achieves adaptive restriction strategy optimization based on emotional characteristics and generative model output, thereby reducing the problems of being too rigid or unsuitable for individual situations that may arise from simply relying on fixed rules.

[0458] When displaying server-generated suggestions and alcohol restriction information, the terminal renders the data sent by the server into a user interface. The terminal can present dinner suggestions in a graphic and textual format, listing recommended dishes, main ingredients, and corresponding health benefits. The terminal can embed an "Order Now" button in the interface; when the user triggers this button, the terminal completes the order process by calling an external food service provider's interface or launching the corresponding application. In addition, the terminal can also display textual explanations related to alcohol restrictions and recommended alternative relaxation methods, such as "a 10-minute evening walk" or "deep breathing exercises," to help users understand the health recommendations provided by the system.

[0459] After reading the suggestions and restrictions, users can rate and provide feedback on the suggestions through the terminal interface. Users can provide a numerical rating (e.g., 1 to 5 points) and enter text comments, such as "The suggested dishes are a bit bland but acceptable" or "The explanation regarding alcohol restrictions is clear." Users can also choose whether to adopt the suggestions and place an order. The terminal then uploads this feedback information to the server.

[0460] After receiving user feedback, the server associates and stores the feedback data with corresponding prompts, constraints, and model outputs. The server performs statistical analysis on a large sample of feedback records to evaluate the impact of different prompt templates and constraint configurations on user satisfaction and adoption rates. The server can calculate the relationship between a specific prompt template version and its average rating and order rate, and further construct a prompt generation strategy model using machine learning methods to predict which prompt structure is more likely to achieve higher user satisfaction and a higher actual execution rate given health and emotional characteristics. Based on this, the server gradually adjusts the prompt template library and constraint settings parameters, making the input for subsequent calls to the generative artificial intelligence model more refined and personalized, technically implementing a feedback-driven adaptive optimization mechanism for prompts.

[0461] Through the collaboration of the various modules mentioned above, the system of this invention is not merely a simple automated replacement of the work of a human health advisor. Rather, within the computer architecture, through improvements in specific data structure design, feature extraction methods, constraint modeling methods, and generative artificial intelligence model prompting strategies, it achieves efficient processing of time-series health data and multimodal emotion data, and technically enhances the controllability and compliance of model output. Therefore, this invention represents a substantial technical improvement over traditional systems in multiple dimensions, including processing speed, recommendation accuracy, error control, data management, and efficiency in integrating with external services.

[0462] use Figure 14 The processing flow is explained.

[0463] Step 1: Users input health and mood-related information on the terminal. Users fill in or update their personal information and daily status through the application interface on the terminal.

[0464] Input: User inputs age, gender, height, weight, medical history, allergy information, lifestyle habits (smoking, drinking, exercise frequency, etc.), subjective feelings on the day (e.g., "I've been under a lot of work pressure lately and often stay up late"), and data such as heart rate, steps, and sleep duration transmitted by wearable devices with user authorization.

[0465] The terminal formats the input, organizing each field into a predefined data structure and generating a timestamp and user identifier. The terminal does not perform complex calculations on the content; it only checks data types (e.g., whether it is a number or empty) and performs simple validity checks.

[0466] Output: A structured data object containing user identifiers, timestamps, physiological data, psychological text data, and sensor data references, ready to be sent to the server.

[0467] Step 2: Terminal sends structured data to server The terminal packages the structured data generated in step 1 into a request message and sends it to the server through a secure communication channel.

[0468] Input: A structured data object containing user health and mood-related data.

[0469] The terminal uses a pre-defined network communication module to serialize the data into a format conforming to the interface specification, establishes a connection with the server via an encrypted transmission protocol, and sends the data to the server's designated interface address. The terminal records the sending time and request identifier for subsequent error retries or log analysis.

[0470] Output: The request data message sent to the server, and the status indicating the result of the sending (e.g., a local flag indicating success or failure).

[0471] Step 3: The server receives and verifies user data. The server receives request messages from the terminal at the backend interface and parses and verifies the data.

[0472] Input: The request data message transmitted by the terminal over the network.

[0473] The server first parses the message, extracting fields such as user identifier, physiological data, psychological text data, and sensor data references. Then, it checks whether the fields are complete and of correct type according to a predefined data pattern. For missing or obviously abnormal fields, the server can mark them as missing values ​​or generate an error log. The server then writes the valid data into the corresponding table in the database and generates a unique record identifier for each record.

[0474] Output: The original user records stored in the database, and a confirmation message (e.g., "Data received successfully") returned to the terminal.

[0475] Step 4: The server preprocesses individual physiological information and generates numerical features. The server reads the latest physiological records from the database within a certain time range and performs statistical analysis and feature calculations.

[0476] Input: User's historical and current physiological data records (including blood pressure, heart rate, weight, steps, sleep duration, etc.).

[0477] The server sorts this data by time and constructs a time series. It calculates statistics (such as mean, maximum, minimum, and standard deviation) for each indicator and also calculates derived indicators such as body mass index, blood pressure deviation, and step moving average. The server uses numerical calculation methods to normalize or standardize each feature, mapping indicators with different dimensions to a unified numerical range and reducing scale differences in subsequent model inputs.

[0478] Output: A set of numerical feature vectors for subsequent calculations, which includes multiple normalized physiological indicators and derived health risk indicators.

[0479] Step 5: The server preprocesses individual psychological information and generates emotional characteristics. The server performs emotion analysis and feature extraction on the user's subjective emotional text, speech-to-text, and facial image data.

[0480] Input: Emotion-related data stored in the database, including emotional text input by the user, voice data references, and image data references.

[0481] The server first converts speech data into text using a speech recognition module, and obtains emotion category probability vectors from image data using an image analysis module. Then, it segments and vectorizes all text (user-input text and speech-to-text) to obtain text emotion vectors. The server then performs weighted fusion of the speech emotion vectors, image emotion vectors, and text emotion vectors, calculating a unified emotion feature vector, including emotion category (e.g., tension, calm) and intensity rating, through weighted summation or a multilayer perceptron structure.

[0482] Output: Emotional characteristic data representing the current user's emotional state, including emotion category labels, emotion intensity, and stress level.

[0483] Step 6: The server generates health constraints and target conditions based on features. The server sets nutritional intake constraints, drug use constraints, and overall health goals based on numerical and emotional characteristics combined with preventive medicine rules.

[0484] Input: The numerical feature vector obtained in step 4 and the emotional feature data obtained in step 5, as well as the user's past medical history and allergy information.

[0485] The server uses rule-based logic to determine if a user has a risk of high blood pressure, obesity, or sleep deprivation, and then generates corresponding restrictions (such as limiting salt intake, controlling energy intake, and avoiding specific foods). The server also determines the strength of alcohol restrictions based on stress levels in the user's emotional profile, for example, setting the permitted alcohol intake to zero or reducing it to a very low level. The server also sets target conditions, such as "controlling blood pressure," "improving sleep," and "reducing stress," to provide directional constraints for subsequent suggestion generation.

[0486] Output: Data describing the constraints and target configurations for nutritional intake constraints, drug use constraints, alcohol consumption restrictions, and health goals.

[0487] Step 7: The server generates prompts for generative artificial intelligence models. The server uses a template mechanism to combine physiological characteristics, emotional characteristics, constraints, and target conditions into natural language prompts.

[0488] Input: User profile information, numerical features, emotional features, constraints, and target conditions.

[0489] The server selects the appropriate prompt template based on the task type (e.g., "Dinner Suggestion Template" or "Alcohol Restriction Instructions Template"), and fills in placeholders in the template with user data such as age, gender, average blood pressure over the past week, exercise level, sleep duration, and stress level. The prompt statement generated by the server is a complete natural language text, clearly indicating the type of suggestion to be generated, the constraints that must be followed, and the output language style.

[0490] Output: A prompt text adapted to the current user and task, used as input for the generative artificial intelligence model.

[0491] Step 8: The server invokes a generative artificial intelligence model and generates suggestion information. The server inputs the prompt into the generative artificial intelligence model to obtain personalized suggestion text for that user.

[0492] Input: The prompt text output in step 7.

[0493] The server inputs prompts into a pre-trained language generation model via a model interface. The model encodes the input sequence based on internal parameters and then generates output words sequentially through a decoding layer. The server receives the complete text generated by the model, which includes dietary recommendations, medication advice, or instructions on alcohol restrictions based on the user's current health and emotional state. The server records the prompts and model version used in this call for subsequent feedback analysis.

[0494] Output: Generated text containing recommendations on nutritional intake, medication use, and alcohol consumption restrictions, i.e., recommendation information.

[0495] Step 9: The server parses and structures the suggestion information. The server will generate text, parse it into structured data, and extract key elements that can be used for order placement and display.

[0496] Input: The suggestion information text generated in step 8.

[0497] The server uses text parsing algorithms to detect information such as dish names, ingredient types, cooking methods, nutritional characteristics, and precautions, mapping this content to internally defined dish and ingredient data structures. The server also checks for content that conflicts with nutritional constraints or allergy information. If any non-compliant ingredients are found, the server can replace or delete the relevant sections, ensuring that the structured data technically complies with pre-defined safety and health rules.

[0498] Output: A structured list of dishes, health instructions, and alcohol restriction parameters for subsequent ordering and display.

[0499] Step 10: The server generates information for food delivery services and interacts with external services. The server generates order-related information based on structured food data, which can be directly used for external food delivery services.

[0500] Input: The structured menu generated in step 9, along with user address, time preference, and other information.

[0501] The server uses the mapping relationships maintained in the system to map internal dish identifiers to external platform dish numbers, and selects an available service provider based on the user's location information. The server assembles order parameters including dish number, quantity, delivery address, contact information, and expected delivery time, encapsulates them into a data format that meets the requirements of the external service interface, and sends them to the external food delivery service system via the network interface.

[0502] Output: Information on food delivery services used for placing orders with external services, and order confirmation data returned by external services.

[0503] Step 11: The server generates display data and sends it to the terminal. The server organizes the suggestion information and order information into a data structure suitable for display on the terminal interface.

[0504] Input: a structured menu, health information text, alcohol restriction information text, and order confirmation data.

[0505] The server generates display data based on front-end display specifications, including titles, descriptive text, menu images and text, order button configuration, and alcohol restriction reminders, ensuring clear and easy-to-read information on the terminal screen. The server then sends this display data to the corresponding user's terminal via a network interface, recording the sending time and content version for future tracking.

[0506] Output: A display data object for terminal rendering.

[0507] Step 12: The terminal displays suggested information and provides an interactive interface. The terminal receives display data from the server and generates a visual interface for user operation.

[0508] Input: The data to be displayed in step 11.

[0509] The terminal parses the display data and, based on the layout information, draws a suggestion list, menu details, alcohol restriction information, and an order button on the screen. The terminal also provides rating input controls and text input boxes, allowing users to easily rate the suggestions. The terminal listens for user clicks; when a user clicks the order button, the terminal invokes the corresponding application or opens the order link based on the accompanying external service information.

[0510] Output: The interactive interface displayed on the terminal screen and the user's interaction event data.

[0511] Step 13: Users provide feedback on suggestions and submit evaluation information. Users can subjectively evaluate the system's recommendations based on the content displayed on their devices and choose to implement those recommendations.

[0512] Input: Suggested information and interactive controls displayed on the terminal.

[0513] Users can choose whether to order from the recommended dishes, or they can click on the rating component to give a score and enter feedback such as "The dish combination is reasonable," "The taste is mild but healthy," or "The alcohol restriction statement is convincing" in the text box. After completing the input, users can click the submit button.

[0514] Output: Feedback data, including rating values, text evaluations, and whether to implement suggestions, is prepared by the terminal and sent to the server.

[0515] Step 14: Terminal uploads user review information The terminal packages the feedback data entered by the user in step 13 and sends it to the server.

[0516] Input: User ratings, text reviews, and action records generated on the terminal.

[0517] The terminal performs simple verification on the feedback data (e.g., whether the scoring range is valid, whether the text length is too long), and attaches a user identifier and timestamp; then it sends the data packet to the server's feedback receiving interface through the same secure communication channel as before.

[0518] Output: User review data packets transmitted to the server.

[0519] Step 15: The server updates the prompts and constraint policies based on the evaluation information. The server analyzes the accumulated user review information to optimize the generation of subsequent prompts and the setting of constraints.

[0520] Input: User review data received in step 14, as well as historically stored prompts, model outputs, and adoption records.

[0521] The server correlates the prompts used each time a suggestion is generated, along with the corresponding suggestion information and user feedback, to construct a sample set that includes prompt type, health characteristics, emotional characteristics, constraint configuration, and user satisfaction. The server uses statistical analysis and machine learning methods to evaluate the impact of different prompt templates and constraint combinations on user satisfaction and adoption rate, and then adjusts the weight and content of the prompt templates accordingly. For example, the server can add wording that has proven to be more acceptable, reduce expressions that cause user rejection, and appropriately adjust the severity of certain nutritional or alcohol restriction parameters under different emotional states.

[0522] Output: An updated prompt statement template library and constraint policy configuration. These updates will be used in the processing of subsequent user requests, thus forming a generative artificial intelligence model invocation and health suggestion generation mechanism based on feedback adaptive optimization.

[0523] The specific processing unit 290 sends the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires sound representing user input regarding the result of the specific processing. The control unit 46A sends the sound data representing user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the sound data.

[0524] Data generation model 58 is a so-called generative AI (Artificial Intelligence). Examples of data generation models 58 include ChatGPT (registered trademark) (accessible via the internet (URL: https: / / openai.com / blog / chatgpt)). Data generation model 58 is obtained through deep learning on a neural network. Input to data generation model 58 are prompt words containing instructions, and inference data such as sound data representing sound, text data representing text, and image data representing images (e.g., still image data or animation data). Data generation model 58 infers from the input inference data based on the instructions represented by the prompt words and outputs the inference result in one or more data forms, such as sound data, text data, and image data. Data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or induction. The specific processing unit 290 performs the aforementioned specific processing while using data generation model 58. The data generation model 58 can also be a model finely tuned to output inference results from prompts that do not contain instructions. In this case, the data generation model 58 can output inference results based on prompts that do not contain instructions. The data processing apparatus 12, etc., includes various data generation models 58, including AI other than the generation AI. AI other than the generation AI can be, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and can perform various processes, but is not limited to this example. Furthermore, the AI ​​can also be an AI agent. Furthermore, when the processing of the above-mentioned parts is performed by AI, the processing can be performed partially or entirely by AI, but is not limited to this example. Furthermore, the processing performed by the AI ​​including the generation AI can be replaced by processing in the rule base, and the processing in the rule base can also be replaced by processing performed by the AI ​​including the generation AI.

[0525] Furthermore, the processing of the aforementioned data processing system 10 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart device 14, but it can also be performed by both the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart device 14. Additionally, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart device 14 or external devices, and the smart device 14 acquires or collects information required for processing from the data processing device 12 or external devices.

[0526] For example, the collection unit is implemented by the control unit 46A of the smart device 14 or the specific processing unit 290 of the data processing device 12. For example, the acquisition unit uses the camera 42 or communication I / F 44 of the smart device 14 to acquire step data, which is then processed by the specific processing unit 290 of the data processing device 12. For example, the analysis unit is implemented by the specific processing unit 290 of the data processing device 12, which analyzes the data from the collection unit and the acquisition unit. For example, the generation unit is implemented by the specific processing unit 290 of the data processing device 12, which uses a generation AI to generate a menu. For example, the serving unit is implemented by the output device 40 of the smart device 14 or the specific processing unit 290 of the data processing device 12, which provides the generated menu to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and various changes can be made.

[0527] In the above embodiments, examples of specific processing by the data processing device 12 are given, but the technology disclosed herein is not limited to this, and specific processing may also be performed by the smart device 14.

[0528] Second Implementation Method Figure 3 An example of the configuration of the data processing system 210 according to the second embodiment is shown.

[0529] like Figure 3 As shown, the data processing system 210 includes a data processing device 12 and smart glasses 214. A server can be cited as an example of the data processing device 12.

[0530] The data processing apparatus 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" as understood in this disclosure. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, the database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0531] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, camera 42, and communication I / F 44 are also connected to the bus 52.

[0532] Microphone 238 receives instructions from user 20 by receiving sounds emitted by user 20. Microphone 238 captures sounds emitted by user 20 and converts the captured sounds into sound data, which is then output to processor 46. Speaker 240 outputs sound according to instructions from processor 46.

[0533] Camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and imaging elements such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, which captures images of the user 20's surroundings (e.g., the field of view defined by an angle equivalent to the field of vision of an average healthy person).

[0534] Communication I / F44 is connected to network 54. Communication I / F44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F44 and 26 is performed in a secure state.

[0535] Figure 4 This illustrates an example of the main functions of the data processing device 12 and the smart glasses 214. For example... Figure 4 As shown, in the data processing device 12, specific processing is performed by the processor 28. The specific processing program 56 is stored in the memory 32.

[0536] The specific processing program 56 is an example of a "program" involved in the technology of this disclosure. The processor 28 reads the specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. Specific processing is implemented by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0537] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. In the emotion inference function (emotion-specific function) using the emotion-specific model 59, various inferences and predictions related to the user's emotions are performed, including inferences and predictions of the user's emotions, but this is not limited to this example. Furthermore, emotion inference and prediction may also include, for example, emotion analysis (parsing).

[0538] In the smart glasses 214, the processor 46 performs reception and output processing. The memory 50 stores the reception and output program 60. The processor 46 reads the reception and output program 60 from the memory 50 and executes the read reception and output program 60 on the RAM 48. The reception and output processing is implemented by the processor 46 operating as a control unit 46A according to the reception and output program 60 executed on the RAM 48. Furthermore, the smart glasses 214 has the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and these models can also be used to perform the same processing as the specific processing unit 290.

[0539] Next, the specific processing of the specific processing unit 290 of the data processing device 12 will be described. Each part of the system described below is implemented by the data processing device 12 and the smart glasses 214. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0540] Example 1 The process is the same as that of the specific process described in Embodiment 1 in the first embodiment above, so the description is omitted.

[0541] Application Example 1 The process is the same as that in the specific processing described in Application Example 1 of the first embodiment above, so the description is omitted.

[0542] Example 2 The process is the same as that of the specific process in Embodiment 2 described in the first embodiment above, so the description is omitted.

[0543] Application Example 2 The process is the same as that in the specific processing described in Application Example 2 of the first embodiment above, so the description is omitted.

[0544] The specific processing unit 290 sends the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A outputs the result of the specific processing to the speaker 240. The microphone 238 acquires sound input representing the user's input regarding the result of the specific processing. The control unit 46A sends the sound data representing the user's input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the sound data.

[0545] Data generation model 58 is a so-called generative AI (Artificial Intelligence). Examples of data generation models 58 include ChatGPT (registered trademark) (accessible via the internet (URL: https: / / openai.com / blog / chatgpt)). Data generation model 58 is obtained through deep learning on a neural network. Input to data generation model 58 are prompt words containing instructions, and inference data such as sound data representing sound, text data representing text, and image data representing images (e.g., still image data or animation data). Data generation model 58 infers from the input inference data based on the instructions represented by the prompt words and outputs the inference result in one or more data forms, such as sound data, text data, and image data. Data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or induction. The specific processing unit 290 performs the aforementioned specific processing while using data generation model 58. The data generation model 58 can also be a model finely tuned to output inference results from prompts that do not contain instructions. In this case, the data generation model 58 can output inference results based on prompts that do not contain instructions. The data processing apparatus 12, etc., includes various data generation models 58, including AI other than the generation AI. AI other than the generation AI can be, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and can perform various processes, but is not limited to this example. Furthermore, the AI ​​can also be an AI agent. Furthermore, when the processing of the above-mentioned parts is performed by AI, the processing can be performed partially or entirely by AI, but is not limited to this example. Furthermore, the processing performed by the AI ​​including the generation AI can be replaced by processing in the rule base, and the processing in the rule base can also be replaced by processing performed by the AI ​​including the generation AI.

[0546] Furthermore, the processing of the aforementioned data processing system 10 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart glasses 214, but it can also be performed by both the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart glasses 214. Additionally, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart glasses 214 or external devices, and the smart glasses 214 acquires or collects information required for processing from the data processing device 12 or external devices.

[0547] For example, the collection unit is implemented by the control unit 46A of the smart glasses 214 or the specific processing unit 290 of the data processing device 12. For example, the acquisition unit uses the camera 42 or communication I / F 44 of the smart glasses 214 to acquire step data, which is then processed by the specific processing unit 290 of the data processing device 12. For example, the analysis unit is implemented by the specific processing unit 290 of the data processing device 12, which analyzes the data from the collection unit and the acquisition unit. For example, the generation unit is implemented by the specific processing unit 290 of the data processing device 12, which uses a generation AI to generate a menu. For example, the serving unit is implemented by the speaker 240 of the smart glasses 214 or the specific processing unit 290 of the data processing device 12, which provides the generated menu to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and various changes can be made.

[0548] In the above embodiments, examples of specific processing by the data processing device 12 are given, but the technology disclosed herein is not limited to this, and specific processing may also be performed by the smart glasses 214.

[0549] Third Implementation Method Figure 5 An example of the configuration of the data processing system 310 according to the third embodiment is shown.

[0550] like Figure 5 As shown, the data processing system 310 includes a data processing device 12 and a head-mounted terminal 314. A server can be cited as an example of the data processing device 12.

[0551] The data processing apparatus 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" as understood in this disclosure. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, the database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0552] The head-mounted terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, camera 42, display 343, and communication I / F 44 are also connected to the bus 52.

[0553] Microphone 238 receives instructions from user 20 by receiving sounds emitted by user 20. Microphone 238 captures sounds emitted by user 20 and converts the captured sounds into sound data, which is then output to processor 46. Speaker 240 outputs sound according to instructions from processor 46.

[0554] Camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and imaging elements such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, which captures images of the user 20's surroundings (e.g., the field of view defined by an angle equivalent to the field of vision of an average healthy person).

[0555] Communication I / F44 is connected to network 54. Communication I / F44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F44 and 26 is performed in a secure state.

[0556] Figure 6 This illustrates an example of the main functions of the data processing device 12 and the head-mounted terminal 314. For example... Figure 6 As shown, in the data processing device 12, specific processing is performed by the processor 28. The specific processing program 56 is stored in the memory 32.

[0557] The specific processing program 56 is an example of a "program" involved in the technology of this disclosure. The processor 28 reads the specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. Specific processing is implemented by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0558] The memory 32 stores the data generation model 58 and the emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290.

[0559] In the head-mounted terminal 314, the processor 46 performs the acceptance / output processing. The memory 50 stores the acceptance / output program 60. The processor 46 reads the acceptance / output program 60 from the memory 50 and executes the read acceptance / output program 60 on the RAM 48. The acceptance / output processing is implemented by the processor 46 operating as a control unit 46A according to the acceptance / output program 60 executed on the RAM 48.

[0560] Next, the specific processing of the specific processing unit 290 of the data processing device 12 will be described. Each part of the system described below is implemented by the data processing device 12 and the head-mounted terminal 314. In the following description, the data processing device 12 will be referred to as the "server" and the head-mounted terminal 314 will be referred to as the "terminal".

[0561] Example 1 The process is the same as that of the specific process described in Embodiment 1 in the first embodiment above, so the description is omitted.

[0562] Application Example 1 The process is the same as that in the specific processing described in Application Example 1 of the first embodiment above, so the description is omitted.

[0563] Example 2 The process is the same as that of the specific process in Embodiment 2 described in the first embodiment above, so the description is omitted.

[0564] Application Example 2 The process is the same as that in the specific processing described in Application Example 2 of the first embodiment above, so the description is omitted.

[0565] The specific processing unit 290 sends the result of the specific processing to the head-mounted terminal 314. In the head-mounted terminal 314, the control unit 46A outputs the result of the specific processing to the speaker 240 and the display 343. The microphone 238 acquires sound input representing the user's input regarding the result of the specific processing. The control unit 46A sends the sound data representing the user's input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the sound data.

[0566] Data generation model 58 is a so-called generative AI (Artificial Intelligence). Examples of data generation models 58 include ChatGPT (registered trademark) (accessible via the internet (URL: https: / / openai.com / blog / chatgpt)). Data generation model 58 is obtained through deep learning on a neural network. Input to data generation model 58 includes prompt words containing instructions, and inference data such as sound data representing sound, text data representing text, and image data representing images (e.g., still image data or animation data). Data generation model 58 infers the input inference data based on the instructions represented by the prompt words and outputs the inference result in one or more data forms such as sound data, text data, and image data. Data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or induction. The specific processing unit 290 performs the aforementioned specific processing while using data generation model 58. The data generation model 58 can also be a model finely tuned to output inference results from prompts that do not contain instructions. In this case, the data generation model 58 can output inference results based on prompts that do not contain instructions. The data processing apparatus 12, etc., includes various data generation models 58, including AI other than the generation AI. AI other than the generation AI can be, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and can perform various processes, but is not limited to this example. Furthermore, the AI ​​can also be an AI agent. Furthermore, when the processing of the above-mentioned parts is performed by AI, the processing can be performed partially or entirely by AI, but is not limited to this example. Furthermore, the processing performed by the AI ​​including the generation AI can be replaced by processing in the rule base, and the processing in the rule base can also be replaced by processing performed by the AI ​​including the generation AI.

[0567] Furthermore, the processing of the aforementioned data processing system 10 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the head-mounted terminal 314, but it can also be performed by both the specific processing unit 290 of the data processing device 12 and the control unit 46A of the head-mounted terminal 314. Additionally, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the head-mounted terminal 314 or external devices, and the head-mounted terminal 314 acquires or collects information required for processing from the data processing device 12 or external devices.

[0568] For example, the collection unit is implemented by the control unit 46A of the head-mounted terminal 314 or the specific processing unit 290 of the data processing device 12. For example, the acquisition unit uses the camera 42 or communication I / F 44 of the head-mounted terminal 314 to acquire step data, which is then processed by the specific processing unit 290 of the data processing device 12. For example, the analysis unit is implemented by the specific processing unit 290 of the data processing device 12 to analyze the data from the collection unit and the acquisition unit. For example, the generation unit is implemented by the specific processing unit 290 of the data processing device 12 to generate a menu using a generation AI. For example, the serving unit is implemented by the speaker 240 and display 343 of the head-mounted terminal 314 or the specific processing unit 290 of the data processing device 12 to provide the generated menu to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and various changes can be made.

[0569] In the above embodiments, examples of specific processing by the data processing device 12 are given, but the technology disclosed herein is not limited to this, and specific processing may also be performed by the head-mounted terminal 314.

[0570] Fourth Implementation Method Figure 7 An example of the configuration of the data processing system 410 according to the fourth embodiment is shown.

[0571] like Figure 7 As shown, the data processing system 410 includes a data processing device 12 and a robot 414. A server can be cited as an example of the data processing device 12.

[0572] The data processing apparatus 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" as understood in this disclosure. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, the database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0573] Robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, camera 42, controlled object 443, and communication I / F 44 are also connected to the bus 52.

[0574] Microphone 238 receives instructions from user 20 by receiving sounds emitted by user 20. Microphone 238 captures sounds emitted by user 20 and converts the captured sounds into sound data, which is then output to processor 46. Speaker 240 outputs sound according to instructions from processor 46.

[0575] Camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and imaging elements such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, to photograph the area around robot 414 (e.g., the field of view defined by a perspective equivalent to the field of vision of an average healthy person).

[0576] Communication I / F44 is connected to network 54. Communication I / F44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F44 and 26 is performed in a secure state.

[0577] The controlled object 443 includes a display device, LEDs (light-emitting diodes) for the eyes, and motors for driving the arms, hands, and feet. The posture or movement of the robot 414 is controlled by controlling the motors in the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. In addition, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[0578] Figure 8 This illustrates an example of the main functions of the data processing device 12 and the robot 414. For example... Figure 8 As shown, in the data processing device 12, specific processing is performed by the processor 28. The specific processing program 56 is stored in the memory 32.

[0579] The specific processing program 56 is an example of a "program" involved in the technology of this disclosure. The processor 28 reads the specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. Specific processing is implemented by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0580] The memory 32 stores the data generation model 58 and the emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290.

[0581] In robot 414, the processor 46 performs the acceptance and output processing. The memory 50 stores the acceptance and output program 60. The processor 46 reads the acceptance and output program 60 from the memory 50 and executes the read acceptance and output program 60 on RAM 48. The acceptance and output processing is implemented by the processor 46 acting as the control unit 46A according to the acceptance and output program 60 executed on RAM 48.

[0582] Next, the specific processing of the specific processing unit 290 of the data processing device 12 will be described. Each part of the system described below is implemented by the data processing device 12 and the robot 414. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 will be referred to as the "terminal".

[0583] Example 1 The process is the same as that of the specific process described in Embodiment 1 in the first embodiment above, so the description is omitted.

[0584] Application Example 1 The process is the same as that in the specific processing described in Application Example 1 of the first embodiment above, so the description is omitted.

[0585] Example 2 The process is the same as that of the specific process in Embodiment 2 described in the first embodiment above, so the description is omitted.

[0586] Application Example 2 The process is the same as that in the specific processing described in Application Example 2 of the first embodiment above, so the description is omitted.

[0587] The specific processing unit 290 sends the result of the specific processing to the robot 414. In the robot 414, the control unit 46A outputs the result of the specific processing to the speaker 240 and the controlled object 443. The microphone 238 acquires sound input representing the result of the specific processing. The control unit 46A sends the sound data representing the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the sound data.

[0588] Data generation model 58 is a so-called generative AI (Artificial Intelligence). Examples of data generation models 58 include ChatGPT (registered trademark) (accessible via the internet (URL: https: / / openai.com / blog / chatgpt)). Data generation model 58 is obtained through deep learning on a neural network. Input to data generation model 58 are prompt words containing instructions, and inference data such as sound data representing sound, text data representing text, and image data representing images (e.g., still image data or animation data). Data generation model 58 infers from the input inference data based on the instructions represented by the prompt words and outputs the inference result in one or more data forms, such as sound data, text data, and image data. Data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or induction. The specific processing unit 290 performs the aforementioned specific processing while using data generation model 58. The data generation model 58 can also be a model finely tuned to output inference results from prompts that do not contain instructions. In this case, the data generation model 58 can output inference results based on prompts that do not contain instructions. The data processing apparatus 12, etc., includes various data generation models 58, including AI other than the generation AI. AI other than the generation AI can be, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and can perform various processes, but is not limited to this example. Furthermore, the AI ​​can also be an AI agent. Furthermore, when the processing of the above-mentioned parts is performed by AI, the processing can be performed partially or entirely by AI, but is not limited to this example. Furthermore, the processing performed by the AI ​​including the generation AI can be replaced by processing in the rule base, and the processing in the rule base can also be replaced by processing performed by the AI ​​including the generation AI.

[0589] Furthermore, the processing of the aforementioned data processing system 10 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the robot 414, but it can also be performed by both the specific processing unit 290 of the data processing device 12 and the control unit 46A of the robot 414. Additionally, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the robot 414 or external devices, and the robot 414 acquires or collects information required for processing from the data processing device 12 or external devices.

[0590] For example, the collection unit is implemented by the control unit 46A of the robot 414 or the specific processing unit 290 of the data processing device 12. For example, the acquisition unit uses the camera 42 or communication I / F 44 of the robot 414 to acquire step data, which is then processed by the specific processing unit 290 of the data processing device 12. For example, the analysis unit is implemented by the specific processing unit 290 of the data processing device 12, which analyzes the data from the collection unit and the acquisition unit. For example, the generation unit is implemented by the specific processing unit 290 of the data processing device 12, which uses a generation AI to generate a menu. For example, the serving unit is implemented by the speaker 240 of the robot 414 and the control object 443 or the specific processing unit 290 of the data processing device 12, which provides the generated menu to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and various changes can be made.

[0591] In the above embodiments, examples of specific processing by the data processing device 12 are given, but the technology disclosed herein is not limited to this, and specific processing may also be performed by the robot 414.

[0592] Furthermore, the emotion-specific model 59, acting as an emotion engine, can determine a user's emotion based on a specific mapping. Specifically, the emotion-specific model 59 can determine a user's emotion based on an emotion graph that serves as a specific mapping (see...). Figure 9 The emotion-specific model 59 can also determine the robot's emotion, and the specific processing unit 290 performs specific processing based on the robot's emotions.

[0593] Figure 9 This is a diagram representing an emotion map 400 that maps multiple emotions. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotion is. On the outer side of the concentric circles, emotions representing states or behaviors arising from mood are arranged. Emotions are concepts that include feelings and mental states. Emotions generated by reactions occurring in the brain are arranged roughly to the left of the concentric circles. Emotions derived from situational judgments are arranged roughly to the right of the concentric circles. Emotions generated by reactions occurring in the brain and derived from situational judgments are arranged roughly above and below the concentric circles. Furthermore, "pleasant" emotions are arranged above the concentric circles, and "unpleasant" emotions are arranged below them. Thus, in the emotion map 400, multiple emotions are mapped based on the structure that generates emotions, and emotions that are likely to occur simultaneously are mapped close to each other.

[0594] These emotions are distributed at the three o'clock position of the emotion map 400, typically fluctuating between peace and anxiety. In the right half of the emotion map 400, situational awareness dominates over internal sensation, thus resulting in an impression of calm.

[0595] The inner side of the emotion map 400 represents the inner state, while the outer side represents behavior. Therefore, the further outward you are from the emotion map 400, the more visible the emotion becomes (manifested in behavior).

[0596] Here, human emotions are based on various balances such as posture and blood sugar levels. When these balances deviate from an ideal state, it indicates an unpleasant state; when they approach the ideal state, it indicates a pleasant state. Emotions in robots, cars, motorcycles, etc., can also be created in the following way: based on various balances such as posture and remaining battery power, when these balances deviate from an ideal state, it indicates an unpleasant state; when they approach the ideal state, it indicates a pleasant state. Emotion maps can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a Brain Physiological Signal Analysis System for Voice Emotion Recognition and Emotion, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). In the left half of the emotion map, emotions belonging to the sensory-dominated region, called "response," are arranged. Furthermore, in the right half of the emotion map, emotions belonging to the situational cognition-dominated region, called "situation," are arranged.

[0597] In the emotion map, two types of emotions that promote learning are defined. One is a negative emotion on the situational side, in the middle or peripheral region of "repentance" or "reflection." This occurs when the robot experiences negative emotions such as "I don't want to experience this feeling again" or "I don't want to be blamed again." The other is a positive emotion on the response side, near the "desire" region. This occurs when there are positive feelings such as "wanting more" or "wanting to know more."

[0598] The emotion-specific model 59 inputs user input into a pre-trained neural network to obtain emotion values ​​representing each emotion shown in the emotion map 400, thereby determining the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network... Figure 10 As shown in the sentiment graph 900, it was trained in a way that sentiments that are configured close to each other have similar values. Figure 10 The text shows examples of emotions such as "peace of mind", "stability", and "reassurance" that have similar emotion values.

[0599] The above description focuses on the functions of the data processing device 12, but the system of this disclosure is not necessarily installed on a server. The system of this disclosure can also be installed as a general information processing system. This disclosure can also be installed, for example, as a software program running on a personal computer, an application running on a smartphone, etc. The method of this disclosure can also be provided to users in the form of SaaS (Software as a Service).

[0600] In the above embodiments, an example of a specific process being performed by a single computer 22 is given. However, the technology disclosed herein is not limited to this, and the specific process can also be distributed among multiple computers, including computer 22. For example, the data generation model 58 can be located on an external device of the data processing apparatus 12, where data is generated based on the input data.

[0601] In the above embodiments, examples of storing a specific processing program 56 in the memory 32 have been described, but the technology disclosed herein is not limited thereto. For example, the specific processing program 56 may also be stored in a portable computer-readable non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed into the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[0602] Alternatively, a specific processing program 56 may be pre-stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 according to the requirements of the data processing device 12.

[0603] In addition, it is not necessary to store all the specific processing program 56 in the storage device such as the server connected to the data processing device 12 via the network 54 or in the memory 32; a portion of the specific processing program 56 may be stored in advance.

[0604] As hardware resources for performing specific processes, various processors, as shown below, can be used. For example, a CPU can be listed as a processor, which functions as a general-purpose processor that performs specific processes by executing software, i.e., a program. Furthermore, processors can be listed as special-purpose circuits such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application-Specific Integrated Circuits), which are processors with circuitry specifically designed to perform specific processes. Each processor has built-in or connected memory, and each processor executes specific processes using that memory.

[0605] The hardware resources for performing a specific process can consist of one of these various processors, or a combination of two or more processors of the same or different types (e.g., a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resources for performing a specific process can be a single processor.

[0606] As an example of a single processor, there are two approaches: First, a processor is composed of a combination of one or more CPUs and software, which functions as a hardware resource to perform a specific process; second, as represented by a SoC (System-on-a-chip), a processor is used to implement the functionality of the entire system, which includes multiple hardware resources for performing a specific process, using a single IC (Integrated Circuit) chip. In this way, the specific process is implemented by using one or more of the aforementioned processors as hardware resources.

[0607] Furthermore, the hardware architecture of these various processors, more specifically, can utilize circuits that combine semiconductor elements and other circuit components. Moreover, the specific process described above is just one example. Therefore, without departing from the main point, unnecessary steps can certainly be deleted, new steps added, or the processing order changed.

[0608] The descriptions and illustrations above are detailed explanations of a portion of the technology disclosed herein, and are merely one example of the technology disclosed herein. For example, the above descriptions of the structure, function, effect, and results are just one example of the structure, function, effect, and results of a portion of the technology disclosed herein. Therefore, without departing from the spirit of the technology disclosed herein, unnecessary parts may be deleted, new elements added, or replacements may be made to the descriptions and illustrations above. Furthermore, to avoid confusion and facilitate understanding of a portion of the technology disclosed herein, explanations of common technical knowledge that do not require special explanation under the premise of being able to implement the technology disclosed herein have been omitted from the descriptions and illustrations above.

[0609] All documents, patent applications and technical specifications set forth in this specification are incorporated herein by reference to the same extent that each document, patent application and technical specification is specifically and individually described therein and referenced by reference.

[0610] In addition, the following notes are provided in response to the above explanation.

[0611] Example 1 (Note 1) An information processing system, characterized in that it comprises: A device for receiving an individual's attribute information, medical history information, lifestyle information, and emotional state information from a terminal having a display device and an input device via a communication network; An apparatus for standardizing and storing attribute information, medical history information, lifestyle information, and emotional state information in a structured form according to an information management structure stored in an information storage device. An apparatus for calculating body mass index and other health indicators based on the stored information using statistical processing programs and data processing components, and generating an individual's health status analysis results through risk assessment that combines past disease information, family medical history information and lifestyle information. An apparatus for constructing prompt statements for input to a generative artificial intelligence model based on the health status analysis results and the emotional state information, using character sequence generation logic or template processing logic. A means for sending the prompt statement to an external information processing device that provides a generative artificial intelligence model, so that the external information processing device generates a response text based on preventive medical knowledge, including dietary advice, medication-related advice, stress management advice, alcohol intake restriction guidelines, and exercise guidelines, and for obtaining the response text from the external information processing device; A device for generating linked information indicating food selection candidates and order information based on dietary suggestions contained in the response text, and sending the linked information to an external information processing device that provides the item service, thereby enabling the individual to place an order through the item service; An apparatus for outputting the response text and the health status analysis results to the terminal in a visual form, generating additional prompts based on the emotional state information to adjust the intensity or frequency of alcohol intake restrictions, and dynamically modifying alcohol-related suggestions in the response text based on the additional prompts.

[0612] (Note 2) According to the information processing system described in Appendix 1, the device for constructing prompt statements for input to a generative artificial intelligence model is configured to embed, when constructing the prompt statements, risk assessment results, body mass index classifications, and quantitative information related to lifestyle habits extracted from the health status analysis results as independent items into the prompt statements, and to include in the prompt statements constraints and output format specification information for constraining the output of the generative artificial intelligence model, so as to prompt the generative artificial intelligence model to generate structured health advice response text in item units.

[0613] (Note 3) According to the information processing system described in Appendix 1, the system further includes: a device for storing historical prompt statements corresponding to the health status analysis results and the emotional state information, as well as evaluation information provided by the individual for the prompt statements, in the form of historical records in an information storage device; and a device for updating the content or constituent elements of the prompt statements based on the statistical processing results of the historical records, thereby gradually optimizing the suggested content generated by the generative artificial intelligence model.

[0614] Application Example 1 (Note 1) An information processing system, characterized in that it comprises: A control unit used to acquire individual medical information and emotional state from an input device connected to an information processing device; A control unit is used to preprocess the acquired individual medical information and emotional state to perform numerical transformation, feature calculation and statistical processing, and generate physical state indicators related to the individual. A control unit that generates structured data or prompts for inputting into a generative artificial intelligence model based on individual medical information and physical condition indicators, and inputs the generated structured data or prompts into the generative artificial intelligence model to obtain a health status assessment result, wherein the health status assessment result divides the health status into multiple levels. A control unit that generates dietary plans, medication plans, behavioral guidelines, and economic incentive information related to preventive medicine based on the obtained health status assessment results and emotional state, and determines the economic incentive information as a fee discount or service provision content for use in electronic payment processing. Control unit for enabling order processing corresponding to the diet plan and medication use plan to be executed in conjunction with external service providing devices, and for enabling economic incentive information generated based on the health status assessment results to be applied to the external service providing devices or electronic payment processing; A control unit is used to generate prompts for dynamically adjusting the intake limit or frequency of addictive substances based on emotional state and health status assessment results obtained from a generative artificial intelligence model, and to update the intake limit or frequency of addictive substances according to the prompts. The control unit is used to present the health status assessment results, diet plan, medication plan, and economic benefits information to the individual, and to operate and control the external service providing device and electronic payment processing according to the individual's selection.

[0615] (Note 2) The information processing system according to Appendix 1 is characterized in that, The control unit for inputting structured data or prompts into the generative artificial intelligence model is configured to include an individual's lifestyle and behavioral history information in the prompts, so that the generative artificial intelligence model outputs a health status assessment result with accompanying explanatory information and improvement suggestions, and generates guidance information for stress management and behavior change based on the explanatory information and improvement suggestions.

[0616] (Note 3) The information processing system according to Appendix 1 is characterized in that, The control unit for generating economic incentive information is configured to determine the incentive content and discount rate from multiple economic incentive candidates based on health status assessment results and explanatory information, store the determined incentive content and discount rate in association with individual identification information, and automatically apply the incentive content and discount rate corresponding to the individual identification information to calculate the payment amount when an electronic payment request is received.

[0617] Example 2 (Note 1) An information processing system, characterized in that it comprises: Means for receiving individual health-related information and emotion-related information from a communication terminal, and for obtaining structured information including individual basic attribute information, body measurement information, past disease information, lifestyle habit information, and the emotion-related information; A means for converting the structured information into tabular data using an information processing program running on general computing resources, calculating body indicators through numerical operations, and extracting feature quantities based on the body indicators and the structured information. A means for inputting the feature values ​​into a machine learning program to calculate risk assessment information representing the individual's disease risk and lifestyle risk; Means for generating health status description information in natural language based on the structured information and the risk assessment information, and generating prompt statements for generating a generative artificial intelligence model based on preventive medicine nutritional intake recommendations, exercise plan recommendations and drug use recommendations in response to the health status description information. A means for inputting the prompt statement into a generative artificial intelligence model and obtaining natural language text from the generative artificial intelligence model containing the nutritional intake recommendations, exercise plan recommendations, and drug use recommendations based on preventive medicine; This is a means of converting the suggested information into visual information and sending it to the communication terminal, prompting the individual on the communication terminal, and linking information with external dietary services based on the dietary-related suggestions contained in the suggested information, so as to make it possible to select and order products for the dietary services. A means for calculating addiction intake restrictions based on the emotion-related information and the risk assessment information, generating a prompt statement for a generative artificial intelligence model that describes the adjustment content of the addiction intake restriction conditions in natural language, obtaining restriction adjustment suggestions containing the adjustment content of the addiction intake restriction conditions from the generative artificial intelligence model using the prompt statement, and updating the addiction intake restriction conditions based on the restriction adjustment suggestions.

[0618] (Note 2) The information processing system according to Appendix 1 is characterized in that, The system is configured to: in addition to generating the nutritional intake recommendations, exercise plan recommendations, and medication use recommendations based on preventive medicine, also obtain stress management recommendations aimed at reducing mental burden by providing additional prompts to the generative artificial intelligence model based on the individual's health-related information, the emotion-related information, and the risk assessment information, and include the stress management recommendations in the visualization information and display them on the communication terminal.

[0619] (Note 3) The information processing system according to Appendix 1 is characterized in that, The system is configured to: extract recommended food categories and dietary menu candidates for the individual from the preventive medicine-based nutritional intake recommendations included in the recommendation information; obtain candidate product information from the external dietary services based on the extraction results; display the candidate product information in association with the recommendation information on the communication terminal; and generate order information for the external dietary services based on the selection of the candidate product information.

[0620] Application Example 2 (Note 1) An information processing system, characterized in that it comprises: A device for acquiring individual physiological and psychological information; A device for preprocessing the individual's physiological and psychological information to generate numerical and emotional features; Apparatus for setting nutritional intake constraints, drug use constraints, and target conditions based on the numerical characteristics and the emotional characteristics, in accordance with preventive medicine. An apparatus for generating prompt statements for a generative artificial intelligence model based on the constraints and the target conditions, combined with the individual's physiological information and psychological information, and inputting the prompt statements into the generative artificial intelligence model to generate suggestion information containing nutritional intake suggestions and drug use suggestions; An apparatus for generating information for food delivery services based on nutritional intake recommendations contained in the recommended information, and for communicating with external food service providers to generate order information for the external food service providers; An apparatus for determining alcohol-related restrictions based on the individual's psychological information and emotional characteristics, generating prompt statements corresponding to the restrictions and inputting them into the generative artificial intelligence model, and adjusting the alcohol-related restrictions based on the output obtained from the generative artificial intelligence model; A device for generating display data based on the suggested information and the adjusted drinking restrictions, and sending the display data to a user terminal; An apparatus for updating the prompt statement and the constraints based on user evaluation information obtained from the user terminal.

[0621] (Note 2) According to the information processing system described in Appendix 1, the device for acquiring individual physiological information is configured to acquire physiological measurement information and daily activity information that change over time, and the device for preprocessing is configured to calculate health risk indicators through time-series analysis and reflect the health risk indicators in the constraints and the prompt statements.

[0622] (Note 3) According to the information processing system described in Appendix 1, the device for acquiring individual psychological information is configured to acquire voice information, image information, and text information, and the device for preprocessing is configured to perform a comprehensive emotion assessment on the voice information, the image information, and the text information using voice analysis, image analysis, and natural language processing, and adjust at least one of the nutritional intake recommendations and the ingredient categories and cooking methods in the food delivery service information based on the emotion assessment.

Claims

1. An information processing system, characterized in that, include: processor; The processor is configured as follows: Receive personal health information and emotional state input; The individual's health information and emotional state are analyzed to generate prompts that indicate dietary and medication recommendations from a preventive medicine perspective. The generated recommendations are provided to the individual and linked with external food delivery services to enable ordering; Based on the individual's emotional state, a prompt message is generated to instruct adjustments to alcohol intake limits, and the alcohol intake limits are adjusted using the prompt message.

2. The information processing system according to claim 1, characterized in that, The processor is also configured to generate stress management recommendations based on the individual's health information and emotional state.

3. The information processing system according to claim 1, characterized in that, The processor is also configured to generate suitable food information based on the individual's health information and emotional state, and to link with the external food delivery service to enable ordering based on the information of the food.

Images