Multi-terminal collaborative intelligent intervention and health management system and method for children obesity

Abstract

The invention discloses a multi-terminal cooperative intelligent intervention and health management system and method for children obesity, and relates to the technical field of medical information processing, and the system comprises an intelligent pre-inquiry module and an intelligent auxiliary diagnosis and treatment module; the intelligent pre-inquiry module is used for collecting medical history, living habits and family history information of obese child patients through interactive dialogue structuralization, and generating pre-diagnosis reports containing risk prompts based on the collected information; and the intelligent auxiliary diagnosis and treatment module is used for receiving and integrating the pre-diagnosis report and the subsequent examination data of the doctor, carrying out comprehensive evaluation based on a professional knowledge graph of children obesity and a machine learning model, and generating a personalized intervention scheme containing diet prescriptions, exercise prescriptions and living habit intervention suggestions. According to the method, multi-disciplinary professional knowledge can be deeply fused, so that the generation of the intervention scheme has standardization and individuation, and the diagnosis and treatment consistency and the long-term management effect can be improved.

Description

Technical Field

[0001] This invention relates to the field of medical information processing technology, and in particular to a multi-terminal collaborative intelligent intervention and health management system and method for childhood obesity. Background Technology

[0002] Childhood obesity has become a global public health problem, with its prevalence continuing to rise and closely related to the occurrence of various chronic diseases. In current clinical practice, the screening, diagnosis, and long-term intervention management of childhood obesity face a series of technical bottlenecks, specifically manifested in the non-standardization and inefficiency of the diagnosis and treatment process, the lack of professional management tools, and the difficulty in achieving personalized and continuous closed-loop health management.

[0003] To address these challenges, the field of medical information technology has developed several intelligent assisted diagnosis and treatment systems and health management platforms. These existing technologies attempt to collect patient information through digital means; some systems can provide basic health data recording functions, and some even apply simple rule engines to provide suggestions in the management of certain common chronic diseases.

[0004] However, these existing technologies still have significant drawbacks when applied to the field of childhood obesity. First, in the information collection phase, most systems lack a structured consultation logic tailored to the characteristics of children's growth and development and the etiology of obesity. Information collection is fragmented and relies on manual sorting, failing to efficiently generate high-quality pre-diagnosis reports. Second, in terms of decision support, existing systems often fail to deeply integrate multidisciplinary expertise related to childhood obesity, making it difficult to generate truly personalized comprehensive plans that include quantified dietary prescriptions, exercise prescriptions, and lifestyle intervention recommendations based on comprehensive pre-diagnosis information and examination data. Finally, at the long-term management level, existing tools typically cannot effectively support the dynamic adjustment of intervention plans and the closed-loop tracking of adherence.

[0005] Therefore, there is an urgent need for a technical solution that can systematically address the above-mentioned shortcomings, so as to achieve a seamless connection from intelligent pre-diagnosis to personalized intervention generation, and improve the efficiency, standardization and long-term effects of childhood obesity diagnosis and management. Summary of the Invention

[0006] The technical problem to be solved by this invention is to address the shortcomings of existing technologies, and the following technical solution is provided: 1) In a first aspect, the present invention provides a multi-terminal collaborative intelligent intervention and health management system for childhood obesity, the specific technical solution of which is as follows: Includes an intelligent pre-diagnosis module and an intelligent auxiliary diagnosis module; The intelligent pre-diagnosis module is used to: collect information on the medical history, lifestyle habits and family history of obese children through interactive dialogue in a structured manner, and generate a pre-diagnosis report containing risk warnings based on the collected information; The intelligent assisted diagnosis and treatment module is used to: receive and integrate pre-diagnosis reports and subsequent examination data from doctors, conduct comprehensive evaluation based on a professional knowledge map of childhood obesity and machine learning models, and generate personalized intervention plans that include dietary prescriptions, exercise prescriptions, and lifestyle intervention suggestions.

[0007] The beneficial effects of the multi-terminal collaborative intelligent intervention and health management system for childhood obesity provided by this invention are as follows: By collaborating with the intelligent pre-diagnosis module and the intelligent auxiliary diagnosis and treatment module, key shortcomings in existing childhood obesity diagnosis and management are effectively addressed. The intelligent pre-diagnosis module utilizes interactive dialogue to achieve structured collection of medical history, lifestyle habits, and family history information from obese children, overcoming the fragmented and inefficient nature of traditional information collection methods. The generated pre-diagnosis report, including risk warnings, provides doctors with a high-quality overview of the patient's condition, improving initial diagnosis efficiency. The intelligent auxiliary diagnosis and treatment module further receives and integrates the pre-diagnosis report and subsequent examination data from doctors, conducting a comprehensive evaluation based on a professional knowledge graph of childhood obesity and machine learning models. This generates personalized intervention plans that include specific dietary prescriptions, exercise prescriptions, and lifestyle intervention suggestions. This process deeply integrates multidisciplinary professional knowledge, ensuring that the generated intervention plans are both standardized and personalized, contributing to improved consistency in diagnosis and treatment and long-term management effectiveness.

[0008] Based on the above solution, the multi-terminal collaborative intelligent intervention and health management system for childhood obesity of the present invention can be further improved as follows.

[0009] Furthermore, it also includes a closed-loop management module, which is used to receive and analyze daily health data uploaded by obese child patients, and dynamically adjust personalized intervention plans based on the daily health data.

[0010] The beneficial effects of adopting the above-mentioned further approach are as follows: the closed-loop management module receives and analyzes the daily health data uploaded by obese children, enabling dynamic adjustments to personalized intervention plans. This process extends the in-hospital treatment plans generated based on pre-diagnosis reports and examination data to long-term implementation and follow-up in the home setting. Dynamically adjusting the plan based on actual implementation data can improve adherence to the plan by obese children and their parents, and enhance the adaptability and long-term effectiveness of weight management, thereby compensating for the shortcomings of traditional treatment models in continuous personalized management.

[0011] Furthermore, the closed-loop management module is specifically used to: receive and analyze daily health data uploaded by obese children, generate effect evaluation information and early warning information based on the analyzed weight change trend and compliance score, and dynamically adjust the personalized intervention plan according to the effect evaluation information and early warning information.

[0012] The beneficial effects of adopting the above-mentioned further approach are as follows: By analyzing daily health data to obtain weight change trends and adherence scores, and based on this, effect evaluation information and early warning information are generated. The closed-loop management module enables data-driven dynamic adjustment of personalized intervention plans. This method shifts the basis for plan adjustment from experience-based judgment to objective quantitative indicators, enabling more timely identification of implementation deviations and effectiveness bottlenecks. Adjustments based on effect evaluation information and early warning information are more targeted, thereby improving the accuracy of long-term health management and the effectiveness of interventions.

[0013] Furthermore, the intelligent pre-diagnosis module includes a dialogue management engine built on a professional knowledge graph of childhood obesity. The dialogue management engine is used to drive interactive dialogue and dynamically plan and output the next round of questions based on the obese child patient's answer to the current question, so as to collect information on the obese child patient's medical history, lifestyle habits and family history.

[0014] The beneficial effects of adopting the above-mentioned further approach are as follows: The dialogue management engine, built upon a professional knowledge graph of childhood obesity, drives interactive dialogue and dynamically plans and outputs the next round of questions based on the obese child patient's answers to the current question, thus achieving intelligent collection of medical history, lifestyle habits, and family history information. This dynamic planning-based consultation logic simulates the clinical thinking path of doctors, making the information collection process more targeted and efficient. Ultimately, this ensures the systematic nature and completeness of the collected information, laying a reliable data foundation for generating high-quality pre-diagnosis reports that include risk warnings.

[0015] 2) Secondly, the present invention also provides a multi-terminal collaborative intelligent intervention and health management method for childhood obesity, the specific technical solution of which is as follows: The system collects information on the medical history, lifestyle habits, and family history of obese children through interactive dialogue, and generates a pre-diagnosis report containing risk warnings based on the collected information. It receives and integrates pre-diagnosis reports and subsequent examination data from doctors, conducts a comprehensive assessment based on a professional knowledge map of childhood obesity and a machine learning model, and generates personalized intervention plans that include dietary prescriptions, exercise prescriptions, and lifestyle intervention suggestions.

[0016] Based on the above scheme, the multi-terminal collaborative intelligent intervention and health management method for childhood obesity of the present invention can be further improved as follows.

[0017] Furthermore, it also includes receiving and analyzing daily health data uploaded by obese child patients, and dynamically adjusting personalized intervention plans based on the daily health data.

[0018] Furthermore, the system receives and analyzes daily health data uploaded by obese children, and dynamically adjusts personalized intervention plans based on this data. This includes receiving and analyzing daily health data uploaded by obese children, generating effect evaluation information and early warning information based on the analyzed weight change trends and compliance scores, and dynamically adjusting personalized intervention plans based on the effect evaluation information and early warning information.

[0019] Furthermore, information on the medical history, lifestyle habits, and family history of obese children was collected, including: The dialogue management engine drives interactive dialogues and dynamically plans and outputs the next round of questions based on the obese child patient's answers to the current questions, in order to collect information on the obese child patient's medical history, lifestyle habits and family history.

[0020] 3) In a third aspect, the present invention also provides an electronic device, the electronic device including a processor coupled to a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor, so that the electronic device realizes any of the above-mentioned multi-terminal collaborative intelligent intervention and health management methods for childhood obesity.

[0021] 4) In a fourth aspect, the present invention also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements any of the above-mentioned multi-terminal collaborative intelligent intervention and health management methods for childhood obesity.

[0022] It should be noted that the beneficial effects of the technical solutions of the second to fourth aspects of the present invention and their corresponding possible implementations can be found in the above description of the technical effects of the first aspect and its corresponding possible implementations, and will not be repeated here. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below: Figure 1 This is a schematic diagram of a multi-terminal collaborative intelligent intervention and health management system for childhood obesity according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating a multi-terminal collaborative intelligent intervention and health management method for childhood obesity according to an embodiment of the present invention. Detailed Implementation

[0024] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0025] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0026] like Figure 1 As shown in the figure, an embodiment of the present invention provides a multi-terminal collaborative intelligent intervention and health management system for childhood obesity. The system includes an intelligent pre-diagnosis module and an intelligent auxiliary diagnosis and treatment module. The intelligent pre-diagnosis module is used to: collect information on the medical history, lifestyle habits and family history of obese children through interactive dialogue in a structured manner, and generate a pre-diagnosis report containing risk warnings based on the collected information; The intelligent pre-diagnosis module includes a dialogue management engine built on a professional knowledge graph of childhood obesity. This engine drives interactive dialogue and dynamically plans and outputs the next round of questions based on the obese child patient's answers to the current questions, in order to collect information on the obese child patient's medical history, lifestyle habits, and family history. Specifically: 1) When an obese child patient or their parents initiate the intelligent pre-diagnosis module via a mobile device, the dialogue management engine first loads a professional knowledge graph on childhood obesity. This graph stores entities, attributes, and relationships related to the diagnosis and treatment of childhood obesity in a structured form. For example, entities include "patient," "dietary habits," and "exercise status," attributes include "age" and "daily screen time," and relationships include "causing risk" and "belonging to assessment dimensions." The engine initializes a current dialogue state variable to record the information already collected, the information to be followed up, and the current question-and-answer node position. In the initial state, the engine determines the entry node from the knowledge graph, which is usually the collection of "basic information of the obese child patient," including age, gender, current height, and current weight.

[0027] 2) The dialogue management engine extracts the corresponding question template and candidate answer options from the knowledge graph based on the question-and-answer node pointed to by the current dialogue state. The question template is a natural language sentence with embedded variables. For example, for the "Breakfast Type" attribute under the "Eating Habits" entity, the question template might be "What is your usual breakfast?". Candidate answer options might include "porridge / noodles", "bread / milk", "high-oil fried food", etc., which are derived from attribute enumeration values ​​or classification criteria defined in the knowledge graph. The engine outputs the populated questions to the user through the mobile user interface.

[0028] 3) The dialogue management engine receives the user's answer to the current question. The answer can be free text input or a selection from candidate options. The engine uses natural language processing technology to extract and normalize key information from the free text, mapping it to predefined standard terms or numerical ranges in the knowledge graph. For example, if the user inputs "loves fried chicken and fries," the engine parses it and stores it in a structured format as "frequency of high-calorie snacks: high." The parsed information is updated in the current dialogue state variables and stored as factual data associated with the corresponding entity attributes in the knowledge graph.

[0029] 4) Based on the obese child patient's answer to the current question, combined with all accumulated structured information and the medical rules defined in the pediatric obesity knowledge graph, the engine dynamically calculates and selects the next optimal question-and-answer node. The decision-making logic is based on an objective function that aims to efficiently and completely collect all necessary information on the obese child patient's medical history, lifestyle habits, and family history with the fewest possible rounds. The decision-making process can formally utilize graph traversal and rule-triggered mechanisms. For example, a rule might be defined as: if "current BMIZ-Score ≥ 2.0" and "daily moderate-to-vigorous intensity exercise time < 60 minutes," then the next question node should prioritize jumping to a node related to "cause of movement disorder" for in-depth inquiry, rather than asking the next preset question in a fixed order. After the decision-making process is completed, the engine updates the current dialogue state variable to the new node position.

[0030] 5) The dialogue management engine repeats steps two through four above, forming an interactive dialogue loop. Each round of dialogue performs context-sensitive path planning based on the answers from the previous round. The loop termination condition is jointly determined by the preset collection target list and the dependencies in the knowledge graph. The loop terminates when the dialogue management engine determines that the current dialogue state has covered all key information items in the target list and there are no unfinished branches that must be followed up based on the collected information. For example, if the family history information does not report a history of diabetes, the in-depth follow-up node regarding "age of diabetes diagnosis" will not be activated, thus avoiding irrelevant questions.

[0031] 6) After the dialogue ends, the dialogue management engine organizes all collected, structured data according to the standard format of a medical report. The engine calls the risk assessment rules in the knowledge graph to automatically analyze key indicators. For example, combining "daily screen time > 2 hours" from "lifestyle information of obese children" with "positive history of diabetes in first-degree relatives" from "family history information," it generates a "High risk: sedentary behavior and family genetic risk" alert. Finally, a complete pre-diagnosis report containing a structured data list, key indicator warnings, and preliminary risk alerts is generated for use by the subsequent intelligent assisted diagnosis and treatment module.

[0032] The medical history information for obese children refers to past medical records and a sequence of health events related to the occurrence, development, and treatment of childhood obesity. This includes, but is not limited to, the child's birth weight and length, past medical history (such as whether they have been diagnosed with insulin resistance, non-alcoholic fatty liver disease, sleep apnea, etc.), past medication history, allergy history, and previous weight management interventions and their effects. This information forms the basis for assessing the causes of obesity, differentiating between simple and secondary obesity, and developing safe and effective intervention programs.

[0033] The lifestyle information for obese children refers to behavioral data reflecting their daily energy intake and expenditure balance patterns. It systematically covers three dimensions: eating behavior, physical activity behavior, and sedentary behavior. Eating behavior includes details of dietary structure, eating patterns and speed, and the frequency and types of snacks and sugary drinks consumed. Physical activity behavior includes the duration of moderate-to-vigorous intensity exercise daily, preferred exercise types, and daily transportation methods. Sedentary behavior mainly includes daily screen time outside of learning. This information is the most direct basis for assessing energy balance deficits, identifying obesity-related environmental factors, and developing personalized behavioral intervention prescriptions.

[0034] Family history information for obese children refers to genetic background information related to obesity and related metabolic diseases among the child's immediate family members and major extended family members. Core content includes a history of obesity, type 2 diabetes, hypertension, dyslipidemia, cardiovascular disease, and endocrine disorders in first- and second-degree relatives such as parents, siblings, and grandparents. Collecting family history information is of great value in assessing the child's genetic susceptibility, predicting the risk of long-term complications, and providing more targeted family health guidance in intervention management.

[0035] Among these, the pre-diagnosis report, which includes risk warnings, is generated based on the collected information, including: 1) After the dialogue management engine completes the interactive dialogue, all the collected raw answers are passed to the report generation submodule. This submodule maps and stores the raw answers as unified structured data objects based on the entity-attribute model defined in the pediatric obesity knowledge graph. For example, a free text answer about "breakfast type" is mapped to the "breakfast nutrition classification" attribute value under the "dietary habits" entity in the knowledge graph, such as "high carbohydrate, low protein." The family history information "father has diabetes" is structured as relational data, linking the "patient" entity and the "family member" entity, and labeled with "disease: type 2 diabetes" and "relationship: first-degree relative." This step transforms all collected medical history, lifestyle habits, and family history information of obese children into standardized data entries that are machine-queryable and computable, and stores them in a temporary patient information database.

[0036] 2) The report generation submodule traverses the structured data and identifies key indicators needed for risk assessment based on a predefined rule base. These indicators are divided into direct measurement indicators and derived indicators. Direct measurement indicators, such as age, height, and weight, are read directly. Derived indicators, on the other hand, need to be calculated using formulas. A core derived indicator is the Body Mass Index (BMI), whose calculation formula is: ,in, This indicates the weight of obese children, expressed in kilograms. This indicates the height of obese children, in meters.

[0037] Calculated The numerical values ​​will be compared with those stored in the knowledge graph based on the gender and age of Chinese children. Compare with percentile standard reference data and calculate The Z-score can determine the percentile of obesity level. Another derived indicator is the percentage of screen time, calculated as follows: ,in, This refers to the daily non-academic screen time of obese children, expressed in hours. This indicates the average daily awake time of obese children, expressed in hours. Used to quantify static behavior levels.

[0038] 3) The report generation submodule calls the built-in risk assessment rule engine. This engine matches the obtained key indicator values ​​with preset risk judgment thresholds in the knowledge graph and performs multi-condition logical judgments. The generation of each risk warning relies on a clear "condition-conclusion" rule. For example: ① Rule 1: If (obese children patients) If the Z-score is ≥2.0 and (the family history information of obese children includes a first-degree relative with a history of type 2 diabetes), then a risk warning will be generated: "High risk: Severe obesity combined with a family history of diabetes, significantly increased risk of insulin resistance and long-term metabolic syndrome".

[0039] Rule 2: If (in the lifestyle information of obese children) If (the daily moderate-to-high intensity exercise time in the lifestyle information of obese children is <1 hour), then a risk warning will be generated: "Medium risk: highly sedentary behavior combined with insufficient exercise, and severely low energy consumption".

[0040] ③ Rule 3: If (the medical history of obese children includes a diagnosis of "non-alcoholic fatty liver disease"), then a risk warning will be generated: "High risk: Non-alcoholic fatty liver disease has been developed, and liver-related biochemical tests and interventions should be prioritized."

[0041] The rules engine executes all relevant rules sequentially, collecting the triggered conclusions, i.e., risk warning texts, into a list. Each risk warning is associated with a corresponding source of evidence data; for example, the evidence triggering rule one is specific... The relationship between Z-score values ​​and specific family diseases.

[0042] 4) The report generation submodule integrates all components into a complete pre-diagnosis report. The report uses a fixed template, which includes the following sections: Patient Basic Information Form, Structured Summary, Detailed Data List, and Risk Warning Section. In the "Patient Basic Information Form," enter the obese child's age, gender, and calculated... The report also categorizes obesity levels. The "Structured Summary" concisely summarizes the key information regarding the obese child's medical history, lifestyle habits, and family history. The "Detailed Data List" presents all collected standardized data in tabular form. The "Risk Warning Zone" prominently displays the generated risk warning list, sorted by risk level (e.g., high risk, medium risk). Finally, the report generation submodule converts this formatted document into portable document format or Hypertext Markup Language format and outputs it to the doctor's workstation, completing the pre-diagnosis report generation process.

[0043] The pre-diagnosis report, which includes risk warnings, is a comprehensive medical document draft automatically generated by the intelligent pre-consultation module before the doctor's formal consultation. This report systematically integrates the medical history, lifestyle habits, and family history information of obese children collected through interactive dialogue, and provides preliminary medical analysis and interpretation. The core components of the report include an overview of the child's demographic and anthropometry data, a standardized structured list of health information, and risk assessment conclusions automatically generated based on predefined medical logic rules. The risk warning section clearly identifies potential or existing high-risk factors for obesity-related complications, metabolic abnormalities, behavioral problems, and genetic susceptibility, providing crucial data support and decision-making assistance for doctors to quickly grasp the key aspects of the condition, plan subsequent examinations, and develop intervention strategies.

[0044] The intelligent assisted diagnosis and treatment module is used to: receive and integrate pre-diagnosis reports and subsequent examination data from doctors; conduct a comprehensive assessment based on a professional knowledge map of childhood obesity and a machine learning model; and generate a personalized intervention plan that includes dietary prescriptions, exercise prescriptions, and lifestyle intervention suggestions. The specific implementation process is as follows: 1) The intelligent assisted diagnosis and treatment module receives structured pre-diagnosis reports from the pre-consultation stage via an application programming interface (API), as well as subsequent examination data manually entered by doctors at the workstation or imported through the system interface. Subsequent examination data includes detailed physical examination records and laboratory test results, such as waist circumference, blood pressure, fasting blood glucose, insulin, four lipid profiles, and liver function indicators. The module's data preprocessing unit aligns and integrates all this information, creating a temporary comprehensive profile containing panoramic information about the obese child patient based on the unified data model defined in the Childhood Obesity Professional Knowledge Graph. Each piece of information in the profile is transformed into a machine-readable standardized feature. For example, "daily vegetable intake in the pre-diagnosis report" is transformed into the feature x_diet_veg with a value of "insufficient"; "fasting insulin level in laboratory tests" is transformed into the feature x_lab_ins with a specific numerical value. The preprocessing unit combines all features into a multi-dimensional feature vector V to represent the current patient's complete condition. The dimension of the feature vector V is determined by all relevant assessment dimensions defined in the professional knowledge graph.

[0045] 2) The knowledge graph module and machine learning model module in the intelligent decision-making layer work in parallel to perform deep analysis on the feature vector V. The knowledge graph module performs rule-based reasoning evaluation. It traverses the features in the feature vector V and matches them with the diagnostic criteria and complication risk rules stored in the graph. For example, rule 1: if (BMIZ-score ≥ 2.0 and waist-to-height ratio ≥ 0.5), the evaluation conclusion is "central obesity". Rule 2: if (fasting blood glucose ≥ 5.6 mmol / L or fasting insulin elevation), the evaluation conclusion is "risk of insulin resistance". Simultaneously, the machine learning model module performs statistical pattern-based evaluation. A pre-trained ensemble learning model, such as a gradient boosting tree model, receives the feature vector V as input. This model learns from a large amount of historical patient data and can output more complex probabilistic judgments. The model output includes the differential diagnosis classification probability of obesity (e.g., the probability of simple obesity). The probability of secondary obesity ) and risk score for developing specific complications (such as fatty liver) in the future. The rule outputs from the knowledge graph module and the probability outputs from the machine learning model module are merged to form a comprehensive assessment report containing both deterministic conclusions and probabilistic judgments. The report includes obesity classification, complication risk levels, and major problem diagnoses.

[0046] 3) The treatment plan generation engine uses the generated comprehensive assessment report as direct input and the intervention rule base stored in the pediatric obesity professional knowledge graph and guidelines such as the "Expert Consensus on the Construction of Pediatric Obesity Clinics" as the blueprint for its generation logic. The engine performs calculations and content filling for each intervention dimension separately. For dietary prescriptions, the engine first calculates the estimated daily energy requirement based on the child's age, gender, height, current weight, and physical activity level. Specifically, a basal metabolic rate estimation formula (such as the Schofield formula) is used. ,in, This represents the basal metabolic rate, expressed in kilocalories. This represents the physical activity coefficient, a constant determined based on activity levels from lifestyle information. In total energy control (e.g., ... Within the framework of (kcal), the engine calculates the daily intake of carbohydrates, protein, and fat in grams based on the nutrient allocation ratios recommended in the guidelines, and maps this to specific food exchange recommendations. For exercise prescriptions, the engine selects a matching exercise type from the atlas's exercise prescription template library based on the child's degree of obesity, the presence of joint protection needs (determined from medical history), and exercise preferences (obtained from lifestyle information in the pre-diagnosis report). Exercise intensity, frequency, and duration are graded according to the child's age and initial fitness assessment; for example, the initial stage recommends "3-5 times per week, 30 minutes of moderate-intensity aerobic exercise (such as brisk walking or swimming)." For lifestyle intervention recommendations, the engine generates a structured list of adjustment suggestions for specific risky behaviors identified in the pre-diagnosis report. For example, if the feature vector shows screen time... If insufficient water intake is detected, the suggestion "limit daily non-learning screen time to 1 hour" is generated; if insufficient water intake is detected, the suggestion "ensure daily intake of 1200-1500 ml of plain water" is generated. All generated prescriptions and suggestion components, along with the key conclusions of the comprehensive assessment report, are automatically assembled into a standardized and clearly written personalized intervention plan document, which is then pushed to the doctor's workstation for review and distribution.

[0047] A dietary prescription is an individualized energy and nutrient intake plan for obese children. The prescription is based on the principles of controlling total energy intake, adjusting the proportion of macronutrients, and ensuring adequate micronutrient intake. The prescription clearly specifies the recommended total calorie intake per day or per meal, breaking it down into specific food types, quantities, and combinations, usually expressed using a "food exchange" system or specific recipe examples. For example, a dietary prescription might include specific guidance such as "Total daily calories 1600 kcal, Breakfast: 2 servings of whole-wheat bread, 1 serving of egg, 1 serving of milk; Lunch: 1.5 servings of brown rice, 1 serving of lean pork, 2 servings of vegetables," etc. The dietary prescription aims to create a safe negative energy balance while cultivating healthy dietary patterns.

[0048] An exercise prescription is an individualized physical activity plan for obese children. This prescription clearly specifies the recommended form, intensity, frequency, and duration of exercise to achieve the goals of weight control, improved metabolism, and enhanced physical fitness. The form of exercise is chosen based on the child's interests, fitness level, and the presence of any contraindications, such as swimming, brisk walking, or cycling. Exercise intensity is usually defined by heart rate range or subjective fatigue level. Exercise frequency and duration follow a gradual progression principle, for example, starting with "3 times a week, 20 minutes each time" and gradually increasing. The exercise prescription aims to safely and effectively increase energy expenditure and improve cardiopulmonary function.

[0049] The lifestyle intervention recommendations provide specific guidance on adjusting weight-related behaviors in obese children's daily lives, excluding diet and structured exercise. This section broadly covers various behavioral factors affecting energy balance, including reducing sedentary behavior (such as setting daily screen time limits), establishing regular sleep schedules (such as consistent sleep and wake-up times), adjusting eating habits (such as slowing down eating and avoiding eating while watching television), and improving hydration habits (such as replacing sugary drinks with plain water). These lifestyle intervention recommendations are typically presented as clear, actionable behavioral goals aimed at eliminating obesity-inducing environmental factors and consolidating the long-term effects of diet and exercise prescriptions.

[0050] The doctor's follow-up examination data refers to the results of a series of supplementary medical examinations performed or ordered by the attending physician during face-to-face consultations with obese children, based on the information provided in the pre-diagnosis report and the physician's clinical judgment, after the completion of the intelligent pre-diagnosis consultation and the generation of the pre-diagnosis report. This data provides important deepening and objective verification of the information in the pre-diagnosis report, and mainly includes two parts: first, detailed physical examination records, such as precise waist circumference measurements, blood pressure readings, and estimated body fat percentage; second, laboratory test results, usually derived from biochemical tests such as blood and urine, including specific numerical indicators such as fasting blood glucose, insulin, blood lipid profile, and liver function transaminase. The doctor's follow-up examination data provides objective evidence regarding the current metabolic status, presence of complications, and endocrine function of obese children, and is an indispensable key input for the intelligent assisted diagnosis and treatment module to conduct accurate comprehensive assessments and generate safe and effective personalized intervention plans.

[0051] In another feasible approach, the process of generating a personalized intervention plan includes: 1) Based on a professional knowledge graph of childhood obesity, pre-diagnosis reports and subsequent examination data are analyzed to extract key assessment features and map them to the intervention dimension framework defined by the graph, generating a structured intermediate representation that includes obesity classification, risk labels, and initial intervention guidelines. This step utilizes entity relationships and medical logic in the graph to transform unstructured text reports and discrete examination values ​​into a machine-computable, semantically related feature set, laying the data foundation for subsequent refined calculations. Specifically: The intelligent assisted diagnosis module receives structured pre-diagnosis reports from upstream sources and subsequent examination data entered by doctors on the workstation. The module's knowledge graph parsing engine is activated. This engine loads a professional knowledge graph on childhood obesity, which predefines a complete intervention dimension framework, such as "energy balance assessment," "metabolic risk," and "behavioral plasticity." The engine first performs deep semantic analysis on the text fields in the pre-diagnosis report, using natural language processing technology to identify and extract entities and attributes related to obesity assessment, such as "severe obesity," "family history of diabetes," and "high screen time." Simultaneously, the engine reads numerical results from the subsequent examination data, such as fasting insulin levels. and LDL cholesterol levels The core task of the parsing engine is to map and aggregate these discrete, diverse data points into specific feature slots within the intervention dimension framework, based on the semantic relationships between entities in the graph and medical logic rules. For example, associating "severe obesity" (diagnostic entity) with "BMIZ-score > 2.5" (examination data attribute) and mapping it to the "obesity classification" dimension generates the feature obesity_type: "severe". Similarly, associating "family history of diabetes" with elevated... The features are correlated and mapped to the "metabolic risk" dimension, generating the feature `metabolic_risk: "insulin_resistance_high"`. Finally, all extracted and mapped features are organized into a unified, machine-readable structured intermediate representation, typically a semantically labeled feature vector or a lightweight graph structure. This representation explicitly includes obesity subtyping, a series of risk labels, and preliminary intervention guidelines inferred from the graph, such as "emphasis on carbohydrate control".

[0052] 2) A multi-task machine learning model is invoked to perform parallel computation on the structured intermediate representation. The model includes a parameter prediction sub-model and a solution recommendation sub-model; the parameter prediction sub-model outputs the daily calorie requirement target value for the dietary prescription. The model outputs a set of candidate exercise prescription types and a priority sequence of lifestyle intervention suggestions, along with nutrient ratios and treatment plan recommendations. The calculation process integrates individual patient characteristics with group patterns. Specifically: The structured intermediate representation generated in the previous step is used as the input feature vector V and fed into a pre-trained multi-task machine learning model. This model architecture is designed to share a low-level feature encoding layer, with the upper layers branching into two parallel sub-model branches. The parameter prediction sub-model is a hybrid regression and classification model that focuses on accurate calculations related to energy and nutrition. This sub-model receives the feature vector V, passes it through its internal network layers, and outputs two key results: one is the daily calorie target value of the dietary prescription. The first is expressed in kilocalories; the second is the nutrient ratio expressed as a percentage, such as the percentage of carbohydrates. Protein percentage Fat percentage ,satisfy The recommended exercise sub-model is a multi-label classification and ranking model that focuses on the selection of behavioral intervention strategies. This sub-model receives the same feature vector V and outputs two results: one is a set of suitable exercise prescription candidate types recommended for the obese child patient. ,like The other is a priority sequence of intervention recommendations targeting common unhealthy lifestyle habits. The sequence is ordered according to the urgency of the recommended implementation and the expected effect, for example: The entire calculation process integrates the individual characteristics of obese children patients reflected in the feature vector V, as well as the group patterns learned by the model from a large amount of historical data.

[0053] 3) The output of the multi-task machine learning model is fused and conflict-resolved with the intervention rule base stored in the childhood obesity professional knowledge graph. The parameters recommended by the model are validated and adjusted based on the medical constraints and safety rules in the graph, for example, to ensure… The intake should not be lower than the safe lower limit corresponding to the patient's age, and unsuitable exercise types should be filtered based on the patient's medical history to form a consistent prescription draft. Specifically: This step introduces a rule system based on explicit medical knowledge to calibrate and optimize the prediction results of the machine learning model. The fusion engine synchronously accesses the output of the multi-task machine learning model and the intervention rule base in the childhood obesity knowledge graph. The rule base contains a series of "IF-THEN" format medical constraints and safety rules. The fusion engine executes conflict detection and resolution logic. For example, the output of the parameter prediction sub-model... It might be a value calculated based on statistical regularities. The rule base contains a safety rule: "For ages..." For children aged 3 years, daily calorie intake should not be lower than the safe minimum. The fusion engine will check inequalities. Is it true? If not, a conflict is triggered, and the engine will use the rules accordingly. Correction For example, the solution recommendation sub-model. The data may include "running". However, if the rule base contains a filtering rule based on medical history: "If the medical history information includes 'knee injury,' then high-impact sports are disabled," and this medical history is indeed present in the feature vector V, then the fusion engine will exclude "running" from the list. Removed from the set. After similar validation, adjustment, and logical filtering of all output parameters, the originally independent model output was transformed into a draft prescription that had been reviewed by medical knowledge and internally coordinated. This draft prescription retained the essence of the model's personalized recommendation while ensuring safety and medical rationality.

[0054] 4) Based on the integrated draft prescription, a final personalized intervention plan document is automatically generated, containing specific quantitative indicators, implementation instructions, and adjustment rationale. The dietary prescription specifies the total daily calories, meal distribution, and food exchange examples; the exercise prescription specifies the type, frequency, intensity, and duration; and the lifestyle intervention recommendations list actionable behavioral goals. The plan is output in a structured format to the doctor's workstation. Specifically: The document generator receives a fused and deconstructed draft prescription as input. The generator has built-in natural language generation templates for doctors and patients. For the diet prescription section, the generator will quantify... , , , This translates into specific meal distribution suggestions and food exchange examples, such as "Total daily calories 1500 kcal, breakfast: 2 servings of whole-wheat bread, 1 serving of milk, 1 egg." For the exercise prescription section, the generator starts from... Select 1-2 items from the set as primary recommendations, specifying the frequency (e.g., 5 times per week), intensity (e.g., maintaining a heart rate of 120-140 beats per minute), and duration (e.g., 40 minutes per session). For lifestyle intervention recommendations, the generator will... The priority sequence generates a list of actionable behavioral goals. Furthermore, the generator extracts key adjustment criteria from the fusion process, briefly explaining them in annotation form, such as "Based on the safe calorie lower limit rule, set the daily calorie intake to 1500 kcal." Finally, all content is assembled into a well-structured, standardized language document containing specific quantitative indicators, implementation instructions, and adjustment criteria. This document is output to the doctor's workstation in a structured data format (such as JSON or XML) and an easy-to-read document format (such as PDF) for final confirmation and distribution by the doctor.

[0055] Optionally, the above technical solution also includes a closed-loop management module. This module is used to: receive and analyze daily health data uploaded by obese children, and dynamically adjust the personalized intervention plan based on the daily health data. Specifically, it receives and analyzes the daily health data uploaded by obese children, generates effect evaluation information and early warning information based on the analyzed weight change trend and compliance score, and dynamically adjusts the personalized intervention plan based on the effect evaluation information and early warning information. The specific implementation process is as follows: 1) Obese children or their parents regularly enter daily health data through a mobile application or automatically upload it via wearable devices. The closed-loop management module's data access service continuously monitors and receives these data streams. The received raw data is immediately cleaned and standardized, for example, by standardizing weight to kilograms and exercise duration to minutes, and by using an image recognition model to parse food images into estimated food types and portions. The processed data is structured and stored in the daily record table of the child's information database along with the obese child's unique identifier and data timestamp. The stored data items strictly correspond to the indicators required to be tracked in the personalized intervention plan, forming a structured dataset arranged in time series.

[0056] 2) The trend analysis submodule of the closed-loop management module periodically (e.g., weekly) extracts the latest weight data sequence of the obese child patient from the database. ,in This indicates the most recently recorded weight. Indicates the number of records. The module first calculates the average weekly weight change rate over a recent window period (e.g., the last four weeks). This rate of change can be estimated using the slope of a linear regression or a simple difference mean. One calculation method is: in, This indicates the number of weeks between the two most recent calculations of weekly variation. The module will also display the current weight. The phased target weight set in the personalized intervention program Compare and calculate the percentage of progress towards the target. : in, This represents the initial weight at the start of the intervention. Analysis results include current weight, recent weight change trend direction (decreasing, stable, increasing), and rate of change. and target progress .

[0057] 3) The compliance assessment submodule of the closed-loop management module calculates a comprehensive compliance score based on the diverse daily health data uploaded within the same time period. This score quantifies the extent to which obese children adhere to the guidelines of a personalized intervention program. The scoring model typically uses a weighted summation method, decomposing the program into multiple observable dimensions and assigning weights to each dimension. For example: in, The dietary prescription adherence score is calculated based on the degree of conformity between the daily uploaded dietary records and the types and quantities of food recommended in the prescription. The exercise prescription adherence score is calculated based on the degree to which the weekly exercise frequency and duration meet the prescription requirements. The compliance score for lifestyle intervention recommendations is calculated based on the achievement of target behaviors such as screen time and sleep duration. , , These represent the preset weights for each dimension, and Sub-ratings for each dimension It is itself a standardized value (e.g., 0 to 100 points) calculated based on the deviation between specific behavioral records and target values.

[0058] 4) The effect evaluation information generator receives the results of weight change trend analysis and compliance scores. It compares actual data with the pre-set expected thresholds in the personalized intervention plan. The effectiveness evaluation information is a structured summary, including: "Average weekly weight loss over the past four weeks..." "kg, below the expected target range", "Current target achievement progress is..." "%", "Current overall compliance score is" The score is considered "medium level". Warning messages are alert signals triggered by predefined rules. For example, rule A: If (for two consecutive weeks...) If (), then a "weight rebound warning" is triggered. Rule B: If ( The absolute value is less than the expected lower limit and If (), then a "poor solution effectiveness and low compliance warning" will be triggered. Rule C: If ( If this occurs, a "high compliance risk warning" will be triggered. The warning information includes the warning level, specific content, and related data evidence.

[0059] 5) The program adjustment engine is the execution terminal of the closed-loop management module. It is activated upon receiving effect evaluation and warning information. The adjustment logic is based on the program adjustment rule base defined in the Childhood Obesity Professional Knowledge Graph. Adjustments can be automatically proposed by the system or implemented after confirmation by a doctor. For example, in response to a warning of "too slow weight loss rate and low dietary adherence score," the adjustment rules might suggest: further clarifying or simplifying the food exchange examples in the dietary prescription, or adjusting the daily total calorie intake. Make a slight adjustment to a fixed value based on the original value. Generate new heat targets For patients with low exercise adherence and a history of joint discomfort, the adjustment rules might suggest changing the exercise type from running to swimming in the exercise prescription and appropriately reducing the initial intensity. Based on the matching rules, the adjustment engine generates a draft of a personalized intervention plan, clearly outlining specific modifications to the original diet prescription, exercise prescription, or lifestyle intervention recommendations. This draft is then sent to the doctor's workstation for review, where the doctor can adopt, modify, or reject these adjustments. Confirmed adjustments are merged to form a new version of the personalized intervention plan and immediately updated to the obese child patient's mobile device, thus completing one dynamic adjustment cycle.

[0060] Routine health data refers to a series of quantitative and qualitative records uploaded regularly or continuously by obese children or their parents via mobile devices during the implementation of personalized intervention programs. These records reflect short-term health status and behavioral adherence. This data serves as the raw material for closed-loop management and primarily includes regular self-monitoring records of anthropometric measurements such as weight and waist circumference; daily dietary intake types, quantities, or photographic records; daily exercise types, durations, and intensity records; and logs of key lifestyle indicators such as sleep duration and screen time. Routine health data is characterized by its temporal sequence, multi-dimensionality, and high frequency, providing objective evidence for assessing the immediate effectiveness of interventions and patient compliance.

[0061] The effectiveness evaluation information is a structured summary report on the effectiveness of personalized intervention programs, generated through periodic calculations, analysis, and interpretation of uploaded daily health data. This information systematically compares quantitative results such as actual weight change trends, progress towards goals, and adherence scores with the initial expected goals set in the intervention program. The effectiveness evaluation information not only reports "what happened," such as the specific rate of weight loss, but also attempts to correlate "why it happened," for example, linking unsatisfactory weight change with lower dietary adherence scores. It provides doctors and patients with clear, quantitative feedback on the effectiveness of the current intervention pathway.

[0062] The early warning information is an automatic alert signal generated by the closed-loop management module when it detects potential risks or adverse trends by monitoring daily health data in real time and applying preset medical logic rules and threshold conditions. Early warning information indicates that the actual implementation has deviated from the expected safe or effective trajectory, such as weight rebound, a much slower-than-expected rate of weight loss, compliance levels falling below dangerous thresholds, or the appearance of potential data anomalies. Early warning information typically includes the warning level, the specific abnormal indicator that triggered the warning, and possible causes. Its purpose is to proactively alert doctors and patients to take intervention measures before problems accumulate or worsen.

[0063] In another possible approach, the process of dynamically adjusting personalized intervention plans is as follows: 1) Construct a multidimensional time series dataset based on daily health data, and use feature engineering methods to extract slope features that reflect the trend of weight change. Stability characteristics reflecting compliance fluctuations Slope characteristics Stability characteristics were obtained by linear fitting of recent weight sequences. By calculating the standard deviation of the compliance rating sequence, a pair of quantitative indicators for effectiveness evaluation is formed. Specifically: The data processing unit of the closed-loop management module periodically extracts all daily health data of designated obese children within a continuous time window from the patient information database. This data is sorted by timestamp and constructed into a multidimensional time-series dataset. Each line contains a time point. and the corresponding weight value at that time point. Dietary adherence score Exercise compliance score and lifestyle habit compliance score Fields such as weight. The feature engineering submodule first targets the weight sequence. Perform the operation. This submodule selects the N most recent valid weight data points and uses the least squares method to perform linear fitting. The fitted line equation is: ,in This refers to the slope characteristic of the extracted weight change trend, expressed in kilograms per day. This is the intercept. The sign and magnitude of the value directly quantify the average direction and rate of recent weight change. Simultaneously, the feature engineering submodule calculates the daily comprehensive compliance score within this time window. The sequence, where Depend on , , The weighted calculation yielded the results. The volatility of the series was determined by calculating the standard deviation. To measure, the formula is: ,in It is the mean of the sequence. This reflects the stability characteristic of compliance fluctuations; a larger value indicates a lower regularity in performance behavior. Slope characteristic. With stability characteristics Together, they form a quantitative indicator pair for subsequent in-depth effect evaluation. .

[0064] 2) The quantitative indicators are input into a pre-trained reinforcement learning model. This model uses the current personalized intervention plan as its state and plan adjustments as its action space. The model evaluates the effectiveness of the current state based on the indicator pairs and outputs a set of preliminary strategy suggestions including the direction and magnitude of adjustment, such as suggesting increasing exercise intensity or fine-tuning dietary calories. Specifically: Quantitative indicators Compared with the currently implemented personalized intervention program Together, they serve as input to the pre-trained reinforcement learning model. Within the model's framework, This is encoded as the "state" in which the model exists. The model's "action space" is defined as a series of adjustments that can be made to the plan, such as "increasing the total daily calories." "kcal", "increase the number of times you exercise per week" "Second time," "Reinforce item A in the lifestyle intervention recommendations." The reinforcement learning model internally includes a value evaluation network that evaluates the input state (i.e., the current plan). and performance indicators The model assesses the long-term value of the current strategy. Based on its policy network, it outputs a set of preliminary policy recommendations that are most likely to improve future long-term value. Every suggestion They are all structured adjustment instructions, for example Possible adjustment: "Adjust direction: Increase energy expenditure; Adjustment range: It is recommended to increase the duration of moderate-intensity exercise by 10 minutes per day"; Possible suggestions include "Adjust direction: Optimize dietary structure; Adjustment magnitude: It is recommended to reduce the proportion of carbohydrates in dinner by 5%." These suggestions are based on the optimal response strategy learned by the model in the simulated environment.

[0065] 3) The initial strategy suggestions output by the reinforcement learning model are matched and optimized with the dynamic adjustment rules in the knowledge graph of childhood obesity. Based on the principles of medical safety and progressiveness, the knowledge graph rules constrain and calibrate the scope of the strategy suggestions, generating a safe and feasible set of specific adjustment instructions. Specifically: This step introduces a rule system based on explicit medical knowledge to constrain the safety and feasibility of the strategies explored by the reinforcement learning model. The rule matching engine simultaneously accesses the initial set of strategy suggestions. and the rule base specifically designed for dynamic adjustment in the childhood obesity knowledge graph. Rule base It includes constraints based on principles of medical safety and gradual improvement. The matching and optimization process handles each constraint individually. The recommendations in the text. For example, one recommendation is to "reduce your total daily calorie intake by 300 kcal". There might be a corresponding rule in the rule base: "The absolute value of a single calorie adjustment must not exceed the safety threshold." The engine will check the conditions. If the condition is not met, the adjustment range will be corrected. And it retains the direction. For example, for the suggestion to "increase running frequency," the rule base might contain a filtering rule based on medical history: "If there is a history of severe knee joint disease, then increasing the frequency of high-impact exercise is prohibited." The engine will query the current patient's medical history information; if the condition is triggered, this suggestion will be removed from the set or replaced with low-impact exercise. After similar matching, validation, amplitude calibration, and filtering of all preliminary strategy suggestions, a safe and feasible set of specific adjustment instructions is generated. . The instructions in the document both preserve the optimization direction of reinforcement learning model exploration and strictly comply with the restrictions of medical rules.

[0066] 4) Based on the specific set of adjustment instructions, automatically generate the revised portion of the new personalized intervention plan and integrate it into a complete updated plan document. The updated document clearly lists the adjustment items, the basis for the adjustment, and the expected goals. After confirmation by the doctor, it is synchronized to the child's or parent's mobile device and the doctor's workstation to complete the dynamic adjustment. Specifically: The closed-loop management module's scheme update generator receives a set of specific adjustment instructions. As input, the generator's task is to create a clear and executable new version of the plan. The generator first parses the currently implemented personalized intervention plan. The generator identifies the document structure and locates the parts that will be modified, such as calorie values ​​in a diet prescription, frequency in an exercise prescription, or entries in a list of lifestyle intervention recommendations. Then, the generator... Each instruction in the text, for The corresponding parts need to be precisely modified. For example, if the instruction is "to reduce the total daily calories from..." Adjusted to If the value is displayed correctly, the generator will update it in the diet prescription section. After all changes are complete, the generator will integrate the changed parts with the unchanged parts to form a complete new version of the personalized intervention plan document. At the same time, the generator will create a separate solution update document, which clearly lists all adjustments and the quantitative indicators on which each adjustment is based. Or rules, and the adjusted expected goals. Update documents and The data was pushed to the doctor's workstation for review and confirmation. Once confirmed, the new personalized intervention plan was implemented. The system immediately synchronizes the data to the patient's or parent's mobile device and the doctor's workstation, overriding the old plan, thus completing a full, data-driven, dynamic adjustment loop.

[0067] The technical solution of the present invention will be further described through the following embodiments. The purpose of this invention is to address the problems of low efficiency, insufficient standardization, and lack of integrated and personalized long-term management tools in the diagnosis and treatment of childhood obesity in existing technologies. Specifically, this invention aims to achieve the following objectives: ① Improve consultation efficiency and information quality: Through an intelligent pre-consultation module, an interactive dialogue management engine based on a professional knowledge graph of childhood obesity is used to drive interactive dialogue before the consultation. This systematically and structurally collects the child's medical history, lifestyle habits, and family history information, generating a pre-consultation report containing risk warnings to assist doctors in quickly understanding the condition. ② Achieve intelligent and standardized diagnosis and treatment: An intelligent auxiliary diagnosis and treatment module based on a professional knowledge graph of childhood obesity and machine learning models comprehensively evaluates the pre-consultation report and the doctor's subsequent examination data, providing intelligent auxiliary diagnostic suggestions, risk assessments, and generating personalized intervention plans including dietary prescriptions, exercise prescriptions, and lifestyle intervention suggestions, ensuring that the diagnosis and treatment process follows the latest expert consensus and guidelines. ③ Build a multi-terminal collaborative closed-loop management system: Through mobile terminals for children or parents and workstation terminals for doctors, the system's various modules achieve seamless integration of data collection, plan execution, effect tracking, and dynamic adjustment. Specifically, the closed-loop management module receives and analyzes daily health data uploaded by obese children. Based on the analyzed weight change trends and compliance scores, it generates effect evaluation and early warning information. The personalized intervention plan is then dynamically adjusted based on the effect evaluation and early warning information, thereby supporting multidisciplinary collaboration and long-term follow-up.

[0068] The system architecture includes a multi-terminal user interaction layer, an application service layer, an intelligent decision-making layer, and a data layer, specifically: 1) The multi-terminal user interaction layer includes mobile terminals for children or parents and doctor workstations. This layer enables user data input, information display, and system interaction. The mobile terminal is used for interactive dialogue collection and daily health data recording in the intelligent pre-diagnosis module; the workstation is used by doctors to review pre-diagnosis reports, enter examination data, review personalized intervention plans, and manage them. The application service layer includes pre-diagnosis services, treatment services, and management services. This layer provides core business logic, including consultation process control, pre-diagnosis report generation, personalized intervention plan distribution, and visualization of daily health data. The intelligent pre-diagnosis module and closed-loop management module are deployed in this layer. The intelligent decision-making layer includes a knowledge graph module and a machine learning model module. Based on a professional knowledge graph of childhood obesity and machine learning models, this layer provides intelligent consultation logic, auxiliary diagnostic suggestions, risk prediction, and personalized intervention plan recommendations. The intelligent auxiliary treatment module is deployed in this layer. The data layer includes a patient information database, a knowledge base, and a model library. This layer stores patient medical records, daily health data, a professional knowledge graph of childhood obesity, treatment guidelines, and trained machine learning model parameters.

[0069] The intelligent pre-diagnosis module, located in the application service layer, employs a dialogue management engine built upon a professional knowledge graph of childhood obesity. Simulating a doctor's consultation process, it systematically collects basic information about the child, including growth and development history, dietary habits, exercise status, family history, and past medical history, through multiple rounds of interactive dialogue. This constitutes the medical history, lifestyle habits, and family history information of obese children. The dialogue management engine dynamically plans and outputs the next round of questions based on the obese child's answers to the current questions. The collected medical history, lifestyle habits, and family history information of obese children is standardized and structured to generate a pre-diagnosis report containing key indicator warnings and risk alerts for doctors to quickly review.

[0070] The intelligent auxiliary diagnosis and treatment module, located at the intelligent decision-making layer, includes a comprehensive assessment submodule and a personalized intervention plan recommendation submodule. The comprehensive assessment submodule is used for obesity classification and risk assessment. Specifically, based on the latest diagnostic criteria for childhood obesity, it automatically grades the degree of obesity, determines metabolic type, and predicts the risk of complications. The comprehensive assessment submodule also uses machine learning models to assist diagnosis. Specifically, it employs an ensemble learning model, combining medical history, physical examination data from pre-diagnosis reports, and subsequent laboratory test data to provide auxiliary diagnostic suggestions, along with the diagnostic basis and confidence level. The personalized intervention plan recommendation submodule is used to generate prescriptions. Specifically, based on the comprehensive assessment results and guidelines such as the "Expert Consensus on the Construction of Childhood Obesity Clinics," it automatically generates personalized intervention plans that include dietary prescriptions, exercise prescriptions, and lifestyle intervention recommendations. The personalized intervention plan recommendation submodule also supports multidisciplinary collaboration, specifically providing interfaces for online collaboration and plan adjustment among multidisciplinary experts such as nutritionists, sports rehabilitation therapists, and endocrinologists.

[0071] The closed-loop management module resides in the application service layer. This module tracks health data, specifically receiving and storing daily health data uploaded by children or parents via mobile devices in real time, such as weight changes, exercise levels, and dietary records, and performing visual trend analysis. The closed-loop management module also dynamically adjusts treatment plans. Specifically, it analyzes weight change trends and adherence scores using a pre-set effectiveness evaluation model, generating effectiveness evaluation information and early warning information, and prompting or assisting doctors in dynamically adjusting personalized intervention plans accordingly.

[0072] A specific embodiment of the present invention is as follows: Scenario: The parents of Xiaoming (10 years old, BMI overweight) are using this system for the first time. Specifically, the parents access the system via mobile device, and the system initiates an interactive dialogue in the intelligent pre-diagnosis module. The system first collects basic information (age, gender, height, weight), automatically calculates BMI, and indicates "obese (severe)". The system further asks structured questions such as: "What is Xiaoming's usual breakfast?", "How much time does he spend watching TV or playing on electronic devices each day?", and "Is there a family history of diabetes?". The parents input information as prompted, and the dialogue management engine dynamically plans subsequent questions based on the answers. The system structures the information in real time and generates a pre-diagnosis report containing "High Risk: Long Sedentary Time, Family History of Diabetes".

[0073] The doctor opens Xiaoming's medical record on the workstation and directly sees the structured pre-diagnosis report and risk warning generated by the system. Based on the report, the doctor conducts targeted physical examinations and laboratory tests (such as blood lipids and blood sugar). The doctor enters the test results into the system, and the intelligent auxiliary diagnosis module immediately performs a comprehensive evaluation of the pre-diagnosis report and test data, providing a "diagnostic suggestion: simple obesity, with a mild risk of insulin resistance," and recommending a personalized intervention plan that includes a diet prescription, exercise prescription, and lifestyle intervention suggestions. The doctor adopts the suggestions and fine-tunes the exercise prescription (for example, changing running to swimming to protect the knee joint), and sends the final plan to the parents' mobile device with one click.

[0074] Parents receive dietary prescriptions, exercise prescriptions, and lifestyle intervention suggestions on their mobile devices. Parents record Xiaoming's weight, food photos, and exercise logs daily via their mobile devices, uploading daily health data. The system monitors the data in real time. The closed-loop management module analyzes the data, and in week 3, based on the analyzed weight change trend and adherence score, generates effectiveness evaluation information and a "poor program effectiveness warning." The closed-loop management module automatically sends the warning information to the doctor's workstation. Upon receiving the warning information, the doctor sends a message to the parents through the system: "Please pay attention to the exercise log; we recommend adding one parent-child outdoor activity this week." Based on the effectiveness evaluation information, the doctor fine-tunes the dietary prescription, completing the dynamic adjustment of the personalized intervention plan and achieving dynamic, personalized long-term management.

[0075] Compared with existing technologies, this invention has the following beneficial effects: ① Improved diagnostic efficiency: The intelligent pre-consultation module advances and structures a large amount of information collection, allowing doctors to focus more on diagnosis and treatment plan development within limited outpatient time, which is expected to save more than 30% of outpatient time. ② Guaranteed standardization and professionalism in diagnosis and treatment: The system's built-in professional knowledge graph and machine learning model for childhood obesity ensure that the generation of diagnoses and personalized intervention plans strictly follows authoritative guidelines, reducing the differences in diagnosis and treatment among primary healthcare institutions. ③ Personalized closed-loop management: The dynamic adjustment mechanism of the multi-terminal collaboration and closed-loop management module extends diagnosis and treatment from hospitals to homes, improving compliance among children and parents and contributing to long-term, effective weight management. ④ Promoted clinical research and technological iteration: The structured and standardized data accumulation provides a high-quality data foundation for clinical research on childhood obesity and the continuous optimization of machine learning models.

[0076] like Figure 2 As shown in the figure, an embodiment of the present invention provides a multi-terminal collaborative intelligent intervention and health management method for childhood obesity, comprising: S1. Collect information on the medical history, lifestyle habits and family history of obese children through interactive dialogue in a structured manner, and generate a pre-diagnosis report containing risk warnings based on the collected information; S2. Receive and integrate pre-diagnosis reports and subsequent examination data from doctors, conduct a comprehensive evaluation based on a professional knowledge map of childhood obesity and a machine learning model, and generate a personalized intervention plan that includes dietary prescriptions, exercise prescriptions, and lifestyle intervention suggestions.

[0077] Optionally, the above technical solution also includes: receiving and analyzing daily health data uploaded by obese child patients, and dynamically adjusting the personalized intervention plan based on the daily health data.

[0078] Optionally, in the above technical solution, receiving and analyzing daily health data uploaded by obese children, and dynamically adjusting the personalized intervention plan based on the daily health data, includes: receiving and analyzing daily health data uploaded by obese children, generating effect evaluation information and early warning information based on the analyzed weight change trend and compliance score, and dynamically adjusting the personalized intervention plan based on the effect evaluation information and early warning information.

[0079] Optionally, in the above technical solution, information on the medical history, lifestyle habits, and family history of obese children is collected, including: The dialogue management engine drives interactive dialogues and dynamically plans and outputs the next round of questions based on the obese child patient's answers to the current questions, in order to collect information on the obese child patient's medical history, lifestyle habits and family history.

[0080] It should be noted that the beneficial effects of the multi-terminal collaborative intelligent intervention and health management method for childhood obesity provided in the above embodiments are the same as the beneficial effects of the multi-terminal collaborative intelligent intervention and health management system for childhood obesity provided in the above embodiments, and will not be repeated here. Furthermore, the methods and system embodiments provided in the above embodiments belong to the same concept, and their specific implementation processes are detailed in the system embodiments, and will not be repeated here.

[0081] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned multi-terminal collaborative intelligent intervention and health management methods for childhood obesity.

[0082] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the above-mentioned multi-terminal collaborative intelligent intervention and health management methods for childhood obesity.

[0083] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0084] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A multi-terminal collaborative intelligent intervention and health management system for childhood obesity, characterized in that, Includes an intelligent pre-diagnosis module and an intelligent auxiliary diagnosis module; The intelligent pre-diagnosis module is used to: collect information on the medical history, lifestyle habits and family history of obese children through interactive dialogue in a structured manner, and generate a pre-diagnosis report containing risk warnings based on the collected information; The intelligent auxiliary diagnosis and treatment module is used to: receive and integrate the pre-diagnosis report and the doctor's subsequent examination data, conduct a comprehensive evaluation based on the professional knowledge map of childhood obesity and machine learning model, and generate a personalized intervention plan that includes dietary prescriptions, exercise prescriptions and lifestyle intervention suggestions.

2. The multi-terminal collaborative intelligent intervention and health management system for childhood obesity according to claim 1, characterized in that, It also includes a closed-loop management module, which is used to: receive and analyze the daily health data uploaded by the obese child patients, and dynamically adjust the personalized intervention plan based on the daily health data.

3. The multi-terminal collaborative intelligent intervention and health management system for childhood obesity according to claim 2, characterized in that, The closed-loop management module is specifically used to: receive and analyze the daily health data uploaded by the obese child patients, generate effect evaluation information and early warning information based on the analyzed weight change trend and compliance score, and dynamically adjust the personalized intervention plan according to the effect evaluation information and early warning information.

4. A multi-terminal collaborative intelligent intervention and health management system for childhood obesity according to any one of claims 1 to 3, characterized in that, The intelligent pre-diagnosis module includes a dialogue management engine built on a professional knowledge graph of childhood obesity. The dialogue management engine is used to: drive the interactive dialogue, and dynamically plan and output the next round of questions based on the obese child patient's answer to the current question, so as to collect the obese child patient's medical history, lifestyle habits and family history information.

5. A multi-terminal collaborative intelligent intervention and health management method for childhood obesity, characterized in that, include: The system collects information on the medical history, lifestyle habits, and family history of obese children through interactive dialogue, and generates a pre-diagnosis report containing risk warnings based on the collected information. The system receives and integrates the pre-diagnosis reports and subsequent examination data from doctors, conducts a comprehensive evaluation based on a professional knowledge map of childhood obesity and a machine learning model, and generates a personalized intervention plan that includes dietary prescriptions, exercise prescriptions, and lifestyle intervention suggestions.

6. A multi-terminal collaborative intelligent intervention and health management method for childhood obesity according to claim 5, characterized in that, Also includes: The system receives and analyzes the daily health data uploaded by the obese child patients, and dynamically adjusts the personalized intervention plan based on the daily health data.

7. A multi-terminal collaborative intelligent intervention and health management method for childhood obesity according to claim 6, characterized in that, Receiving and analyzing daily health data uploaded by the obese child patients, and dynamically adjusting the personalized intervention plan based on the daily health data, includes: receiving and analyzing daily health data uploaded by the obese child patients, generating effect evaluation information and early warning information based on the analyzed weight change trend and compliance score, and dynamically adjusting the personalized intervention plan based on the effect evaluation information and early warning information.

8. A multi-terminal collaborative intelligent intervention and health management method for childhood obesity according to any one of claims 5 to 7, characterized in that, Collect medical history, lifestyle habits, and family history information of obese children, including: The interactive dialogue is driven by a dialogue management engine, and the next round of questions is dynamically planned and output based on the obese child patient's answer to the current question, so as to collect information on the obese child patient's medical history, lifestyle habits and family history.

9. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the multi-terminal collaborative intelligent intervention and health management method for childhood obesity as described in any one of claims 5 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the multi-terminal collaborative intelligent intervention and health management method for childhood obesity as described in any one of claims 5 to 8.

Images