Auxiliary decision-making method and system for cardiovascular chronic disease management

Abstract

The invention provides an auxiliary decision-making method and system for cardiovascular chronic disease management, electronic equipment, a computer readable storage medium and a computer program product, and relates to the technical field of cardiovascular disease monitoring and prevention, and the method comprises the steps: obtaining the current personal information of a target user; the current personal information comprises current feature information and current state information; predicting the current physique information of the target user according to the similarity between the current feature information and a historical physique contrast set; performing risk analysis on the current state information, generating risk factor information, and performing risk quantification; and obtaining an auxiliary prediction result based on the mapping relationship between the current physique information, the risk factor information and the acupuncture points, and using the auxiliary prediction result to assist professionals in determining a target acupuncture point so as to apply the auxiliary prediction result to cardiovascular chronic disease management.

Description

Technical Field

[0001] The present invention relates to the technical field of monitoring and prevention of cardiovascular diseases, and in particular to a decision-making support method, system, electronic device, computer-readable storage medium and computer program product for the management of chronic cardiovascular diseases. Background Art

[0002] Cardiovascular disease is a common chronic disease that may cause symptoms such as palpitations, shortness of breath, chest pain, dizziness, insomnia, etc., seriously affecting patients' daily life and work ability. In severe cases, it may cause myocardial infarction, stroke, heart failure, etc., and even lead to death or disability.

[0003] With the acceleration of the aging process of the population and the transition to a modern lifestyle, the high incidence, high prevalence, high recurrence rate and high mortality rate are more prominent in the manifestations of cardiovascular diseases.

[0004] In order to cope with the severe situation of cardiovascular disease, effective management and intervention of chronic cardiovascular diseases are needed. However, the problems with existing intervention methods mainly include: (1) relying on annual physical examinations or occasional medical visits, which cannot capture dynamic physiological fluctuations and lead to delayed early warning capabilities; (2) current guidelines (such as hypertension management guidelines) are based on group statistics and do not consider individual differences.

[0005] A decision-making support method, system, electronic device, computer-readable storage medium and computer program product for the management of chronic cardiovascular diseases are hereby proposed. Summary of the Invention

[0006] This specification provides an auxiliary decision-making method, system, electronic device, computer-readable storage medium and computer program product for the management of chronic cardiovascular diseases. By combining data from traditional Chinese medicine and Western medicine, the current physical condition information and risk factor information of the target user are predicted to achieve multi-dimensional evaluation and prediction; based on the mapping relationship between the current physical condition information, the risk factor information and acupoints, auxiliary prediction results are obtained to assist professionals in determining target acupoints, so as to facilitate their application in the management of chronic cardiovascular diseases, thereby improving the management efficiency and accuracy of cardiovascular diseases.

[0007] The present application provides a decision-making support method for cardiovascular chronic disease management using the following technical solutions, including:

[0008] Obtaining the current personal information of the target user; the current personal information includes: current feature information and current status information;

[0009] Predicting the current physical condition information of the target user based on the similarity between the current feature information and the historical physical condition reference set;

[0010] Performing risk analysis on the current state information to generate risk factor information;

[0011] Based on the mapping relationship between the current physical condition information, the risk factor information and the acupoints, an auxiliary prediction result is obtained.

[0012] Optionally, predicting the current physical condition information of the target user based on the similarity between the current feature information and a historical physical condition reference set includes:

[0013] Constructing the historical constitution control set;

[0014] performing a similarity analysis on the current feature information based on the historical physical condition comparison set to obtain a similarity analysis result;

[0015] Screening out target sample pairs according to the similarity analysis results;

[0016] The physical fitness information in the target sample pair is used as the current physical fitness information.

[0017] Optionally, performing similarity analysis on the current feature information based on the historical physical condition reference set to obtain similarity analysis results includes:

[0018] Traversing the historical physical condition comparison set, calculating the similarity between the current feature information and each piece of historical feature information, and obtaining a first similarity;

[0019] searching, according to the first similarity and the first similarity threshold, for historical physical sample pairs that meet the first similarity condition, as original sample pairs;

[0020] The current feature information and the original sample pair are compared to determine the same feature attributes and different feature attributes, which are used as similarity analysis results.

[0021] Optionally, based on the similarity analysis result, screening out target sample pairs includes:

[0022] Determine the similarity weight based on the weight corresponding to each identical feature attribute; determine the difference weight based on the weight corresponding to each different feature attribute;

[0023] Determine the second similarity of each original sample pair by combining the feature attributes of the difference, the difference weight, and the similarity weight;

[0024] The target sample pair is selected from the original sample pair based on the second similarity.

[0025] Optionally, performing risk analysis on the current state information to generate risk factor information includes:

[0026] Performing state analysis on the current state information to obtain comprehensive state parameters;

[0027] Performing risk analysis on the current state information to obtain at least one risk parameter;

[0028] The comprehensive state parameter and all the risk parameters are integrated to generate risk factor information.

[0029] Optionally, the risk parameter includes: at least one of a current risk parameter and a future risk parameter;

[0030] Optionally, performing risk analysis on the current state information to obtain at least one risk parameter includes:

[0031] Perform current risk analysis on current status information to obtain current risk parameters;

[0032] Conduct future risk analysis on current status information and generate future risk parameters.

[0033] Optionally, also include:

[0034] receiving decision feedback related to the auxiliary prediction result and configuring the acupoint stimulation device;

[0035] After the configuration is completed, acupoint stimulation reminders are performed based on the execution order of the operation knobs.

[0036] The present application provides a decision-making support method for cardiovascular chronic disease management using the following technical solutions, including:

[0037] An information acquisition module is used to acquire the current personal information of the target user; the current personal information includes: current feature information and current status information;

[0038] A physical fitness prediction module, configured to predict the current physical fitness information of the target user based on the similarity between the current feature information and a historical physical fitness reference set;

[0039] A risk prediction module, configured to perform risk analysis on the current state information and generate risk factor information;

[0040] The auxiliary decision-making module is used to obtain auxiliary prediction results based on the mapping relationship between the current physical condition information, the risk factor information and the acupoints.

[0041] Optionally, the physical fitness prediction module includes:

[0042] A set construction submodule, used to construct the historical constitution control set;

[0043] A similarity analysis submodule, configured to perform a similarity analysis on the current feature information based on the historical constitution reference set to obtain a similarity analysis result;

[0044] A screening submodule, configured to screen out target sample pairs based on the similarity analysis results;

[0045] The extraction submodule is configured to use the physical fitness information in the target sample pair as the current physical fitness information.

[0046] Optionally, the similarity analysis submodule includes:

[0047] a first calculation unit, configured to traverse the historical physical condition comparison set, calculate the similarity between the current feature information and each piece of historical feature information, and obtain a first similarity;

[0048] a similarity search unit, configured to search for a pair of historical physical sample meeting a first similarity condition as an original sample pair based on the first similarity and a first similarity threshold;

[0049] The attribute comparison unit is used to compare the current feature information with the original sample pair, determine the same feature attributes and different feature attributes, and use them as the similarity analysis result.

[0050] Optionally, the screening submodule includes:

[0051] A weight determination unit, configured to determine a similarity weight based on the weight corresponding to each identical feature attribute; and to determine a difference weight based on the weight corresponding to each different feature attribute;

[0052] a second calculation unit, configured to determine a second similarity of each original sample pair by combining the characteristic attributes of the difference, the difference weight, and the similarity weight;

[0053] A screening unit is configured to screen out the target sample pair from the original sample pair based on the second similarity.

[0054] Optionally, the risk prediction module includes:

[0055] A state prediction submodule is used to perform state analysis on the current state information to obtain comprehensive state parameters;

[0056] a risk prediction submodule, configured to perform risk analysis on the current state information to obtain at least one risk parameter;

[0057] The risk integration submodule is used to integrate the comprehensive status parameter with all the risk parameters to generate risk factor information.

[0058] Optionally, the risk parameter includes: at least one of a current risk parameter and a future risk parameter;

[0059] Optionally, the risk prediction submodule includes:

[0060] A first analysis unit is used to perform current risk analysis on current state information to obtain current risk parameters;

[0061] The second analysis unit is used to perform future risk analysis on the current state information and generate future risk parameters.

[0062] Optionally, also include:

[0063] a configuration module, configured to receive decision feedback related to the auxiliary prediction result and configure the acupoint stimulation device;

[0064] The reminder module is used to provide acupoint stimulation reminders based on the execution sequence of the operation knobs after the configuration is completed.

[0065] This specification also provides an electronic device, wherein the electronic device includes:

[0066] processor; and,

[0067] A memory storing computer executable instructions, which, when executed, cause the processor to perform any of the above methods.

[0068] This specification also provides a computer-readable storage medium, wherein the computer-readable storage medium stores one or more programs, and when the one or more programs are executed by a processor, any of the above methods is implemented.

[0069] This specification also provides a computer program product, wherein the computer program product includes: a computer program / instructions, and when the computer program / instructions are executed by a processor, any of the above methods is implemented.

[0070] In this application, the current personal information of the target user is obtained; the current personal information includes: current feature information and current status information; the current physical information of the target user is predicted based on the similarity between the current feature information and the historical physical control set; the current status information is subjected to risk analysis, risk factor information is generated, and risk quantification is performed; based on the mapping relationship between the current physical information, the risk factor information and the acupoints, an auxiliary prediction result is obtained to assist professionals in determining the target acupoints, so as to facilitate its application in the management of chronic cardiovascular diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 A schematic diagram of the principle of a decision-making support method for cardiovascular chronic disease management provided in an embodiment of this specification;

[0072] Figure 2 A flowchart of a decision-making support method for cardiovascular chronic disease management provided in an embodiment of this specification;

[0073] Figure 3 A schematic diagram of the structure of a decision support system for cardiovascular chronic disease management provided in an embodiment of this specification;

[0074] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of this specification;

[0075] Figure 5 A schematic diagram of a computer-readable medium provided in accordance with an embodiment of this specification. DETAILED DESCRIPTION

[0076] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are for illustrative purposes only, and those skilled in the art will readily appreciate other obvious variations. The basic principles of the present invention defined in the following description may be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the present invention.

[0077] The exemplary embodiments of the present invention will now be described more fully with reference to the accompanying drawings. Under the premise of being consistent with the technical concept of the present invention, the features, structures, characteristics or other details described in a particular embodiment do not exclude that they can be combined in one or more other embodiments in a suitable manner.

[0078] The term "and / or" or "and / or" includes all combinations of any one or more of the associated listed items.

[0079] If the technical solution of this application involves personal information, the product that applies the technical solution of this application has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing personal information. If the technical solution of this application involves sensitive personal information, the product that applies the technical solution of this application has obtained the individual's separate consent before processing sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, a clear and prominent sign is set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that they agree to the collection of their personal information; or on the personal information processing device, when the personal information processing rules are notified by obvious signs / information, the individual's authorization is obtained through pop-up information or by asking the individual to upload their personal information; among which, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

[0080] Figure 1A schematic diagram of the principle of a decision-making support method for cardiovascular chronic disease management provided in an embodiment of this specification, the method comprising:

[0081] S1 obtains the current personal information of the target user; the current personal information includes: current feature information and current status information;

[0082] S2 predicting the current physical condition information of the target user based on the similarity between the current feature information and the historical physical condition reference set;

[0083] S3 performs risk analysis on the current state information to generate risk factor information;

[0084] S4 obtains an auxiliary prediction result based on the mapping relationship between the current physical condition information, the risk factor information and the acupoints.

[0085] The existing cardiovascular disease prevention and control system relies heavily on annual physical examinations and sporadic clinical data. Routine physical examinations are typically conducted over six months apart, and fluctuations in cardiovascular risk factors (such as blood pressure, blood lipids, and blood sugar) have circadian and seasonal characteristics. Therefore, their temporal resolution is insufficient. Furthermore, a single measurement cannot reflect the physiological changes in patients' real-life scenarios. For example, transient hemodynamic changes caused by factors such as exercise, diet, and mood swings are often obscured by static data, thus missing the spatial dimension of cardiovascular chronic disease management.

[0086] The formulation of current guidelines (such as the JNC8 hypertension guidelines and the ESC heart failure guidelines) relies on statistical conclusions from large-scale cohort studies and cannot address individual differences.

[0087] Considering that there is a close correlation between cardiovascular chronic disease management and TCM constitution, based on this, in order to improve the management efficiency and accuracy of cardiovascular chronic disease management, the present invention proposes an auxiliary decision-making method for cardiovascular chronic disease management, which combines the data of TCM and Western medicine to obtain auxiliary prediction results. Specifically, Figure 2 As shown, the method includes:

[0088] S1 obtains the current personal information of the target user;

[0089] Create a user profile for each new user; assign a unique identifier to each user; obtain the user's personal information and store it in the user profile. When the user's latest personal information is subsequently obtained, overwrite and / or update the records in the user profile.

[0090] Current personal information includes: current feature information and current status information;

[0091] S11 collects the user's current feature information;

[0092] The current feature information includes a plurality of feature data, each of which includes at least one feature attribute, and is used to characterize the user's TCM features. Specifically, the current feature information includes one or more of facial feature data, odor feature data, pulse feature data, and question-answer feature data.

[0093] S111 performs a first feature analysis on the user's face to obtain facial feature data;

[0094] Face includes: the user's face and tongue coating;

[0095] The facial feature data includes: face feature data and tongue feature data.

[0096] In one embodiment of the present specification, a professional manually observes and judges the user's face and tongue coating, obtains the manual judgment result, and uses it as facial feature data.

[0097] In another embodiment of the present specification, facial feature data includes: facial feature attributes; specifically, a facial image of the user is collected, and the facial image includes: an initial facial image and an initial tongue coating image; an HSV color space analysis is performed on the initial facial image to obtain a color analysis result, and the color analysis result includes: pigment concentration of the H channel and glossiness of the V channel; a DLI index calculation is performed based on the color analysis result to generate a DLI value representing the blood oxygen status as a facial feature attribute for grading the facial color.

[0098] In another embodiment of the present specification, the tongue feature data includes: tongue quality feature attributes and tongue coating texture feature attributes. Specifically, the initial tongue coating image is segmented into ROIs, interference areas such as teeth and lips are excluded, the tongue contour is determined, and a first tongue coating image is obtained; a device color response curve is established using a Macbeth color card, and RGB channel deviation is automatically corrected; the first tongue coating image is converted to a CIELAB color space to obtain device-independent chromaticity values; the a* channel value (red-green axis) is extracted; the tongue quality feature attributes are obtained by combining the a* channel value; the tongue quality feature attributes are preferably HbCI values, where HbCI = (a* channel value - 12.5) / 15.0 × 100% (the result is limited to 0-100%); the first tongue coating image is converted into an 8-bit grayscale image, and the surface roughness characteristics are analyzed using a gray level co-occurrence matrix (GLCM) to obtain the tongue coating texture feature attributes.

[0099] S112 performs a second characteristic analysis on the user's oral cavity to generate odor characteristic data;

[0100] Odor signature data is used to characterize oral odor.

[0101] In one embodiment of the present specification, a professional manually smells and judges the user's oral odor, obtains the manual judgment result, and uses it as odor characteristic data.

[0102] In another embodiment of the present specification, the odor characteristic data includes: odor characteristic attributes. Specifically, the user's oral cavity is analyzed based on acoustic wave technology to obtain odor analysis data; the odor analysis data includes: odor medium bulk modulus V, medium density ρ, acoustic wave frequency f, acoustic wave wavelength λ, amplitude A, and initial phase value φ;

[0103] Substitute all odor analysis data into the odor characteristic attribute calculation formula to generate odor characteristic attributes;

[0104] Calculation formula for odor characteristic attributes: Among them, WEIGHT is the user's weight; SpO2 is the user's oxygen saturation data.

[0105] In specific implementation, the elastic properties of the oral gas medium are measured based on the pressure-volume dynamic response test to obtain the medium bulk modulus k; the gas in the mouth is subjected to gas composition spectral analysis to obtain the medium density ρ; the gas in the mouth is subjected to phase-locked loop frequency synthesis analysis to measure the sound wave frequency f; the gas in the mouth is subjected to laser Doppler velocimetry to obtain the sound wave wavelength λ; the odor in the mouth is subjected to high-precision microphone array analysis to obtain the amplitude a.

[0106] In other embodiments of the present specification, odor characteristic data can also be obtained based on other methods, such as electrophysiological technology, calcium imaging technology, ion-selective electrodes, optical sensors, biosensor enzymes, etc., to convert chemical signals into digital codes (used to represent "sweetness value" and / or "saltiness value") to obtain odor characteristic data.

[0107] S113 performs a third characteristic analysis on the user's pulse to generate pulse characteristic data;

[0108] The pulse characteristic data is used to characterize the user's pulse condition.

[0109] In one embodiment of the present specification, a professional touches and judges the user's pulse, obtains an artificial judgment result, and uses it as pulse characteristic data.

[0110] In another embodiment of the present specification, the user's electromyographic biofeedback information is collected and the pulse characteristic attributes are extracted as pulse characteristic data; in order to improve the signal-to-noise ratio, the electromyographic biofeedback information can be band-pass filtered (0.1-100Hz) and baseline drift removed.

[0111] Among them, myoelectric biofeedback information includes: current intensity E 电流强度, current speed E 电流速度 , current frequency z, current waveform p, muscle biosignal conduction efficiency q t .

[0112] Among them, a 脉象 is the coefficient and b is the normalization coefficient, ensuring that the numerator and denominator dimensions match.

[0113] S114 obtains the user's question and answer feedback according to the preset question and answer script, and obtains question and answer feature data.

[0114] In one embodiment of the present specification, a series of preset questions are pre-set according to the needs of cardiovascular chronic disease management as a preset Q&A script; the preset Q&A script covers symptoms, lifestyle habits, treatment status, etc.

[0115] Present the Q&A script to users and obtain their Q&A feedback; organize and analyze the users' Q&A feedback and extract key information as Q&A feature data.

[0116] The present invention adopts HSV / CIELAB color space and GLCM texture analysis to quantify TCM physical signs, combines electromyographic signal processing with acoustic wave technology, reduces subjective judgment errors, and improves feature extraction accuracy.

[0117] S12 obtains the user's current status information;

[0118] Current status information includes: current basic information and current monitoring information;

[0119] S121 collects the user's current basic information;

[0120] The current basic information is used to represent the patient's personal basic situation. Specifically, the current basic information includes but is not limited to: name, gender, age, symptoms SPT, past medical history PMH, current diagnosis results CD, health status, and key information.

[0121] The current diagnostic results of CD include but are not limited to: cholesterol level, blood uric acid, cardiac output CD, solid vascular resistance SVR, central venous pressure CVR and other physical examination data.

[0122] Health conditions include but are not limited to: smoking status, activity intensity (Activity), sleep duration (ST), and sleep quality (SQ). Activity intensity is used to characterize a user's exercise habits, such as average weekly exercise duration.

[0123] Taking hypertension as an example, key information includes but is not limited to: onset time of hypertension, duration of hypertension, duration of hypertension, and hypertension rating results.

[0124] S122 collects the user's current monitoring information;

[0125] The current monitoring information is used to characterize the user's physiological characteristics. Specifically, the current monitoring information includes: blood pressure data, oxygen saturation data, and other physiological data.

[0126] S122-1 obtains blood pressure data;

[0127] Blood pressure data includes: current systolic blood pressure SBP and current diastolic blood pressure DBP.

[0128] In one embodiment of the present specification, when collecting a user's blood pressure, a real-time blood pressure measurement is obtained using an ambulatory blood pressure monitor at preset intervals. The measurement results include systolic blood pressure, diastolic blood pressure, heart rate, and other data. Preferably, the preset interval is 15 minutes during daytime and 30 minutes during nighttime. The specific time ranges for daytime and nighttime can be set based on the twilight zone or actual needs. The measurement results are stored separately according to the time period in which the collection time occurred. Specifically, when the collection time falls within the first time period, the measurement result collected is stored as the morning measurement result in the first database (TDBPdatabase); when the collection time falls within the second time period, the measurement result collected is stored as the midday measurement result in the second database (TLBP database); when the collection time falls within the third time period, the measurement result collected is stored as the evening measurement result in the third database (TNBP database); and when the collection time falls within the fourth time period, the measurement result collected is stored as the night measurement result in the fourth database (TMBP database). Preferably, the first time period is 4:00-10:00; the second time period is 10:00-16:00; the third time period is 16:00-22:00; and the fourth time period is 22:00-4:00 the next day.

[0129] The present invention divides and stores blood pressure data by day and night time periods, captures the laws of physiological fluctuations, breaks through the time resolution limitations of traditional single measurements, and reflects dynamic changes in real-life scenarios.

[0130] S122-2 obtains oxygen saturation data;

[0131] In one embodiment of the present specification, oxygen saturation data is obtained by performing real-time measurement on the user based on an oxygen saturation instrument.

[0132] In one embodiment of the present specification, the oxygen saturation data includes: the latest average oxygen saturation value

[0133] Average oxygen saturation value based on historical oxygen saturation data Real-time oxygen saturation values The weighted sum of the oxygen saturation average value is determined

[0134] Among them, w′ 氧饱和 is the first calculation weight of oxygen saturation data; 氧饱和 A second calculated weight for the oxygen saturation data.

[0135] During the collection of oxygen saturation data, abnormal values may occasionally appear in the oxygen saturation data due to, for example, the collection equipment falling off.

[0136] In order to reduce motion artifacts, in one embodiment of this specification, when the deviation between real-time data and the median exceeds a threshold, an anomaly detection mechanism is activated.

[0137]

[0138] Among them, the sliding window method can be used to calculate the standard deviation of the most recent N measurements. When the deviation between the real-time data and the median in the window exceeds 3 times the standard deviation, anomaly detection is triggered; if an abnormal value is detected for 3 consecutive times, the collection is automatically paused and a device inspection reminder is issued.

[0139] S122-3 collects other physiological data;

[0140] Other physiological data include but are not limited to: heart rate data, blood sugar data, body mass data, waist-to-hip ratio data, and body fat data.

[0141] Therefore, the current monitoring information includes: blood pressure data, heart rate data, oxygen saturation data SpO2, blood sugar data, body mass data, waist-to-hip ratio data, and body fat data.

[0142] This invention constructs a multi-dimensional health portrait by integrating traditional Chinese medicine characteristics (facial features, tongue coating, pulse) with Western medicine dynamic monitoring data (blood pressure, oxygen saturation, etc. in different time periods), thereby improving the comprehensiveness and objectivity of data collection.

[0143] S2 predicting the current physical condition information of the target user based on the similarity between the current feature information and the historical physical condition reference set;

[0144] S21 constructs the historical constitution control set;

[0145] The historical physical constitution control set includes: several historical physical constitution sample pairs.

[0146] Specifically, the historical physical comparison information of several historical users is collected, and the historical physical comparison information includes: historical feature information and historical physical information. The historical physical information is manually annotated by professionals.

[0147] The historical physical comparison information of the same historical user is aligned in time series to construct a historical physical sample pair. Each historical physical sample pair consists of historical feature information and historical physical information corresponding to the same historical user at the same time.

[0148] S22 performs similarity analysis on the current feature information based on the historical physical condition reference set to obtain a similarity analysis result;

[0149] S221 traverses the historical physical condition comparison set, calculates the similarity between the current feature information and each piece of historical feature information, and obtains a first similarity;

[0150] S222: searching for a historical physical sample pair that meets a first similarity condition based on the first similarity and a first similarity threshold, as an original sample pair;

[0151] The first similarity condition includes: first similarity ≥ first similarity threshold.

[0152] That is, when the first similarity between the current feature information and the historical feature information is not less than the first similarity threshold, the historical constitution sample pair containing the corresponding historical feature information is used as the original sample pair.

[0153] S223 compares the current feature information with the original sample pair, determines the same feature attributes and the different feature attributes, and uses them as the similarity analysis result;

[0154] Combine the current feature information with the original sample pair to analyze and determine the same feature attributes and different feature attributes between the two.

[0155] S23: screening target sample pairs according to the similarity analysis results;

[0156] S231 determines the similarity weight w according to the weight corresponding to each identical feature attribute. 相似权重 ; Determine the difference weight w according to the weight corresponding to the characteristic attribute of each difference 差异权重 ;

[0157] S232 determines a second similarity of each original sample pair by combining the feature attributes of the difference, the difference weight, and the similarity weight;

[0158] In one embodiment of the present specification, the second similarity is obtained by the ratio of the weighted sum of the difference feature attributes to the geometric mean (square root) of the similarity weight;

[0159] Right now,

[0160] S233: Filter out the target sample pair from the original sample pair based on the second similarity.

[0161] In one embodiment of the present specification, the original sample pairs are arranged in descending order according to the second similarity; the original sample pair with the highest second similarity is selected as the target sample pair;

[0162] In another embodiment of the present specification, all original sample pairs whose second similarity meets the second similarity threshold are selected as target sample pairs. Of course, after determining one or more target sample pairs, professionals can also conduct verification to determine the target sample pairs.

[0163] S24 uses the physical fitness information of the target sample pair as the current physical fitness information.

[0164] In one embodiment of the present specification, the current constitution information includes one or more of: balanced constitution, qi deficiency constitution, yang deficiency constitution, yin deficiency constitution, phlegm-damp constitution, damp-heat constitution, blood stasis constitution, qi stagnation constitution, and special constitution.

[0165] The present invention is based on similarity weighted analysis of historical physical sample pairs and introduces a difference feature weight mechanism to achieve personalized physical classification to clarify individual differences.

[0166] S3 performs risk analysis on the current state information to generate risk factor information;

[0167] S31 performs state analysis on the current state information to obtain comprehensive state parameters;

[0168] S311 constructs a plurality of physiological state parameters according to the current state information;

[0169] The types of physiological state parameters include: first state parameters and second state parameters.

[0170] Specifically, the types of the first state parameter include but are not limited to: blood pressure state parameter, heart rate state parameter, oxygen saturation state parameter, blood glucose state parameter, and the first state parameter includes: state value and state coefficient.

[0171] Specifically, the types of the second state parameter include, but are not limited to: mass state parameter, waist-to-hip ratio state parameter, body fat state parameter. The second state parameter includes: state coefficient.

[0172] S311-1 constructs a plurality of first state parameters according to the current state information;

[0173] S311-1-1 determines blood pressure status parameters based on blood pressure data in current monitoring information;

[0174] Blood pressure status parameters include: blood pressure status value F 血压 and blood pressure state coefficient k 血压 Specifically:

[0175] (1) Find the collection period corresponding to the blood pressure data and match the corresponding scaling factor α i ;

[0176] Retrieve the collection time of blood pressure data, match the collection time with the collection period corresponding to the collection time, and match the corresponding proportional factor α i In one embodiment of the present specification, when the acquisition time corresponds to the first acquisition period, the corresponding proportional factor is α1; when the acquisition time corresponds to the second acquisition period, the corresponding proportional factor is α2; when the acquisition time corresponds to the third acquisition period, the corresponding proportional factor is α3; when the acquisition time corresponds to the fourth acquisition period, the corresponding proportional factor is α4. Proportional factor α i is pre-set.

[0177] (2) Determine the blood pressure state value F by combining blood pressure data and proportional factors 血压 ;

[0178] Blood pressure status value F 血压 =α i (SBP-DBP+mean arterial pressure); where mean arterial pressure is

[0179] (3) Find the corresponding blood pressure state coefficient k based on blood pressure data 血压 ;

[0180] The target user's blood pressure status is determined based on the blood pressure data, and the corresponding blood pressure status coefficient is matched according to the blood pressure status. Specifically, blood pressure status includes: normal blood pressure, hypertension, and hypotension. If the blood pressure status is normal blood pressure, the corresponding blood pressure status coefficient k 血压 is a1; if the blood pressure condition is hypertension, the corresponding blood pressure state coefficient k 血压 is a2; if the blood pressure condition is hypotension, the corresponding blood pressure state coefficient k 血压 is a3.

[0181] S311-1-2 determines the heart rate state parameter by combining the blood pressure data and heart rate data in the current monitoring information;

[0182] Heart rate data is used to characterize the target user's heart rate HR. Heart rate state parameters include: heart rate state value F 心率 and heart rate state coefficient k 心率 Specifically:

[0183] (1) Retrieve the current systolic blood pressure (SBP) and heart rate data (HR) from the blood pressure data;

[0184] (2) Determine the heart rate state value by combining the current systolic blood pressure (SBP) and heart rate data (HR);

[0185] Heart rate status value Among them, ε1, ε2, and ε3 are constants.

[0186] (3) Find the corresponding heart rate state coefficient k based on blood pressure data 心率 ;

[0187] The target user's heart rate status is determined based on the heart rate data, and the corresponding heart rate status coefficient is matched according to the heart rate status. Specifically, the heart rate status includes: normal heart rate, high heart rate, and low heart rate. If the heart rate status is normal, the corresponding heart rate status coefficient k 心率 is b1; if the heart rate condition is high heart rate, the corresponding heart rate state coefficient k 心率 is b2; if the heart rate condition is low heart rate, the corresponding heart rate state coefficient k 心率 is b3.

[0188] In one embodiment of the present specification, the historical systolic blood pressure and historical heart rate information of multiple historical users can be obtained, the historical heart rate state value of each historical user can be manually marked, and a training sample can be constructed based on the historical heart rate state value, historical systolic blood pressure and historical heart rate information of the historical user; the training sample can be substituted into Determine the values of ε1, ε2, and ε3.

[0189] S311-1-3 determines oxygen saturation state parameters based on oxygen saturation data in current monitoring information;

[0190] Oxygen saturation data SpO2 is used to characterize the oxygen saturation status of the target user. Oxygen saturation status parameters include: oxygen saturation status value F 氧饱和 and oxygen saturation coefficient k 氧饱和 Specifically:

[0191] (1) Determine the oxygen saturation state value F based on oxygen saturation data 氧饱和 ;

[0192] Convert the oxygen saturation data into decimals and use it as the oxygen saturation state value F 氧饱和 ;

[0193] (2) Find the corresponding oxygen saturation state coefficient k based on the oxygen saturation data 氧饱和 ;

[0194] Determine the target user's oxygen saturation status based on oxygen saturation data; specifically, oxygen saturation status includes: normal oxygen saturation, mild hypoxia, and obvious hypoxia. If SpO2 ≥ 95%, the corresponding oxygen saturation status is normal oxygen saturation; if 95% > SpO2 ≥ 80%, the corresponding oxygen saturation status is mild hypoxia; if SpO2 < 80%, the corresponding oxygen saturation status is obvious hypoxia;

[0195] Match the corresponding oxygen saturation state coefficient according to the oxygen saturation state; specifically, if the oxygen saturation state is normal oxygen saturation, the corresponding oxygen saturation state coefficient k 氧饱和 is c1; if the oxygen saturation condition is mild hypoxia, the corresponding oxygen saturation coefficient k 氧饱和 is c2; if the oxygen saturation condition is obviously hypoxic, the corresponding oxygen saturation coefficient k 氧饱和 is c3.

[0196] S311-1-4 determines blood sugar status parameters based on the blood sugar data in the current monitoring information;

[0197] Blood glucose data is used to characterize the target user's fasting blood glucose FBG. Blood glucose status parameters include: blood glucose status value F 血糖 and blood glucose status coefficient k 血糖 Specifically:

[0198] (1) Determine the blood sugar status value F based on blood sugar data 血糖 ;

[0199] In one embodiment of the present specification, blood glucose data is used as blood glucose status value F 血糖 ;

[0200] (2) Find the corresponding blood sugar state coefficient k based on blood sugar data 血糖 ;

[0201] The target user's blood sugar status is determined based on the blood sugar data; specifically, the blood sugar status includes: normal blood sugar, hypoglycemia, and hyperglycemia.

[0202] Match the corresponding blood sugar state coefficient according to the blood sugar state; specifically, if the blood sugar state is normal blood sugar, the corresponding blood sugar state coefficient k 血糖 is h1; if the blood sugar condition is hypoglycemia, the corresponding blood sugar state coefficient k 血糖 is h2; if the blood sugar condition is hyperglycemia, the corresponding blood sugar state coefficient k 血糖 For h3.

[0203] S311-2 constructing a plurality of second state parameters according to the current state information;

[0204] S311-2-1 Determine the quality status parameters based on the body quality data in the current monitoring information;

[0205] Body mass data includes one or more of height, weight, and BMI. Mass status parameters include: mass status coefficient k BMI Specifically:

[0206] (1) If the body mass data includes height and weight, the BMI index is calculated based on the body mass data and used as the body mass index; if the body mass data includes the BMI index, the BMI index is directly retrieved as the body mass index;

[0207] (2) Determine the target user's body density based on body mass index;

[0208] Among them, human body density conditions include: low, normal, overweight, and obese.

[0209] (3) Match the corresponding mass state coefficient k according to the human body density BMI ;

[0210] If the human body density is low, the corresponding mass state coefficient k BMI is d1; if the human body density is normal, the corresponding mass state coefficient k BMI is d2; if the human body density is overweight, the corresponding mass state coefficient k BMI is d3; if the body density is obese, the corresponding mass state coefficient k BMI is d4.

[0211] S311-2-2 determines the waist-to-hip ratio status parameter based on the gender in the current basic information and the waist-to-hip ratio data in the current monitoring information;

[0212] Waist-to-hip ratio data includes one or more of waist circumference WAIST, hip circumference HIP, and waist-to-hip ratio WHR. Waist-to-hip ratio state parameters include: waist-to-hip ratio state coefficient k 腰臀比 Specifically:

[0213] (1) If the waist-hip ratio data includes waist circumference WAIST and hip circumference HIP, the waist-hip ratio WHR is calculated based on the waist-hip ratio data and used as the waist-hip ratio index; if the body mass data includes the waist-hip ratio WHR, the waist-hip ratio WHR is directly used as the waist-hip ratio index;

[0214] (2) Determine the corresponding waist-to-hip ratio classification conditions based on the gender of the target user; specifically:

[0215] When the target user is male, the normal waist-hip ratio range is 0.85-0.9. Therefore, the waist-hip ratio classification conditions include: the first waist-hip ratio type: 0.85≤waist-hip ratio≤0.9; the second waist-hip ratio type: waist-hip ratio <0.85; the third waist-hip ratio type: waist-hip ratio >0.9;

[0216] When the target user is female, the normal waist-hip ratio range is 0.67-0.8. Therefore, the waist-hip ratio classification conditions include: the first waist-hip ratio type: 0.67≤waist-hip ratio≤0.8; the second waist-hip ratio type: waist-hip ratio <0.67; the third waist-hip ratio type: waist-hip ratio >0.8;

[0217] (3) Determine the central obesity status of the target user by combining the waist-hip ratio classification conditions and the waist-hip ratio index of the target user;

[0218] (4) Match the corresponding waist-to-hip ratio state coefficient k according to the central obesity status 腰臀比 ;

[0219] If the central obesity condition is the first obesity type, the corresponding waist-to-hip ratio state coefficient k 腰臀比 is e1; if the central obesity condition is the second obesity type, the corresponding waist-hip ratio state coefficient k 腰臀比 is e2; if the central obesity condition is the third obesity type, the corresponding waist-hip ratio state coefficient k 腰臀比 is e3.

[0220] S311-2-3 determines the body fat status parameter based on the gender in the current basic information and the body fat data in the current monitoring information;

[0221] Body fat data includes body fat percentage BFP. Body fat status parameters include: body fat status coefficient k 体脂 Specifically:

[0222] (1) Determine the corresponding body fat classification conditions based on the gender of the target user; specifically:

[0223] When the target user is male, a body fat percentage of 15%-20% is considered healthy. Therefore, the body fat classification conditions include: the first body fat type: 15% ≤ body fat percentage ≤ 20%; the second body fat type: body fat percentage < 15%; the third body fat type: body fat percentage > 20%;

[0224] When the target user is female, a body fat percentage of 20%-25% is considered healthy. Therefore, the body fat classification conditions include: the first body fat type: 20% ≤ body fat percentage ≤ 25%; the second body fat type: body fat percentage < 20%; the third body fat type: body fat percentage > 25%;

[0225] (2) Determine the target user's body fat type by combining the body fat classification conditions and the target user's body fat data;

[0226] (3) Match the corresponding body fat state coefficient k according to the body fat type 体脂 ;

[0227] If the central obesity condition is the first body fat type, the corresponding body fat state coefficient k 体脂 is g1; if the central obesity condition is the second body fat type, the corresponding body fat state coefficient k 体脂 is g2; if the central obesity condition is the third body fat type, the corresponding body fat state coefficient k 体脂 For g3.

[0228] S312 combines all the first state parameters and the second state parameters to obtain a comprehensive state parameter.

[0229] S312-1 calculates a weighted sum of the state values in the first state parameter to obtain a weighted sum result;

[0230] That is, the weighted sum result = F 血压 ·k 血压 +F 心率 ·k 心率 +F 氧饱和 ·k 氧饱和 +F 血糖 ·k 血糖 .

[0231] S312-2 calculates the sum of all state coefficients in the first state parameter to obtain a first state weight;

[0232] First state weight = k 血压 +k 心率 +k 氧饱和 +k 血糖 .

[0233] S312-3 obtains a first state component according to a ratio of the weighted sum result to the first state weight;

[0234] S312-4 calculates the second state component according to the product of the second state parameter;

[0235] The second state component = f(k BMI ·k 腰臀比 ·k 体脂 );

[0236] S312-5 obtains a comprehensive state parameter based on the sum of the first state component and the second state component;

[0237] That is, the comprehensive state parameter

[0238] S32 performs risk analysis on the current state information to obtain at least one risk parameter;

[0239] The risk parameters include: at least one of a current risk parameter and a future risk parameter;

[0240] S321 performs current risk analysis on the current state information to obtain current risk parameters;

[0241] In one embodiment of the present specification, the calculation of the current risk parameter includes:

[0242] S321(A)-1 retrieves first analysis information from current basic information;

[0243] The first analysis information includes: age, hypertension onset time, and hypertension rating.

[0244] S321(A)-2 retrieves second analysis information from the current monitoring information;

[0245] The second analysis information includes: blood sugar data, oxygen saturation data, body mass data, heart rate data, waist-to-hip ratio data, and body fat data.

[0246] S321(A)-3 extracts third analysis information based on the current monitoring information;

[0247] The third analysis includes: mean arterial pressure, 24-hour average pulse pressure, and nighttime blood pressure drop rate. While most users' nighttime blood pressure continues to decrease, a small number experience increases. It's easy to understand that for users whose nighttime blood pressure rises, their nighttime blood pressure drop rate is negative.

[0248] Specifically, the mean arterial pressure is extracted from the current monitoring information; wherein, the mean arterial pressure

[0249] Obtain blood pressure data based on current monitoring information and calculate 24-hour average blood pressure (24-MAP);

[0250] Obtain blood pressure data based on current monitoring information and calculate the nighttime blood pressure drop rate DP; nighttime blood pressure drop rate

[0251] S321(A)-4 inputs the first analysis information, the second analysis information, and the third analysis information into a blood pressure risk prediction model to obtain a current risk parameter;

[0252] The blood pressure risk prediction model is preferably a multiple linear regression model.

[0253] The input items of the blood pressure risk prediction model include: first analysis information, second analysis information and third analysis information; that is, the input items of the blood pressure risk prediction model include: oxygen saturation data, body mass data, age, heart rate data, waist-to-hip ratio data, body fat data, blood sugar data, hypertension onset time, hypertension rating, mean arterial pressure, 24-hour average pulse pressure, and nighttime blood pressure drop rate.

[0254] The output of the blood pressure risk prediction model is the current risk coefficient H 当前风险 .

[0255] H 当前风险 =β0+β1S p O2+β2·Body mass data+β3·Age+β4·Heart rate data+β5·

[0256] Waist-to-hip ratio + β6·Body fat percentage + β7·Blood sugar data + β8·Hypertension onset time + β9·Hypertension rating + β 10 ·

[0257] Average pulse pressure + β 11 24-hour average pulse pressure + β 12 Nocturnal blood pressure drop rate.

[0258] Among them, β0-β 12 is the regression coefficient of each indicator, which is used to reflect its influence on the risk coefficient.

[0259] In one embodiment of the present specification, historical data of historical users and manually annotated risk coefficients are obtained in advance; the historical data and risk coefficients are input into a blood pressure risk prediction model to determine the values of each regression coefficient.

[0260] This approach improves the efficiency of predicting adverse events, such as hypotension that may occur in patients after device electrical stimulation. This allows for the use of risk factors to determine whether antihypertensive treatment is necessary, providing a reference for subsequent antihypertensive treatment.

[0261] In another embodiment of this specification, the medical device has stored historical blood pressure values and 24-hour ambulatory blood pressure reports. It now continuously monitors and collects the target user's blood pressure every 5 minutes in real time, requiring adaptive parameter adjustment. Therefore, to improve the accuracy of the prediction, the present invention also combines monitoring information such as heart rate data, oxygen saturation data, waist-to-hip ratio data, body fat data, and body mass data to determine the current risk parameter, thereby facilitating a later determination of whether antihypertensive intervention is needed. The calculation of the current risk parameter includes:

[0262] S321(B)-1 synchronizes the blood pressure data every 5 minutes with the dynamic monitoring information;

[0263] Dynamic monitoring information includes but is not limited to: heart rate data and oxygen saturation data.

[0264] To improve accuracy, missing values are handled based on linear interpolation or nearest neighbor filling.

[0265] S321(B)-2 extracts statistical features of historical data, such as the average blood pressure value in the past hour and the heart rate change rate in the past 30 minutes.

[0266] The statistical characteristics include one or more of mean, variance, and trend.

[0267] S321(B)-3 extracts dynamic and static parameters and inputs them into a deep learning model to predict blood pressure;

[0268] Specifically, the input items of the deep learning model include dynamic parameters and static parameters. Dynamic parameters include, but are not limited to, blood pressure data, heart rate data, oxygen saturation data, and current feature information; static parameters include, but are not limited to, age, gender, duration of hypertension, body mass data, waist-to-hip ratio data, body fat data, and blood sugar data.

[0269] Combining static parameters with dynamic time series features. Of course, in order to improve the data fusion effect, in one embodiment of this specification, it is also possible to combine the gradient boosting tree (XGBoost) to process multivariate input and mixed features (numeric + category), and improve the interpretability by combining time series features.

[0270] In another embodiment of the present specification, feature intersection can also be performed.

[0271] The deep learning model is modeled in layers, specifically including: short-term response layer, long-term health layer and fusion layer.

[0272] The short-term response layer uses LSTM to process dynamic data (blood pressure, heart rate, and oxygen saturation) collected in the last hour or in real time. The long-term health layer uses random forests and XGBoost to analyze the long-term impact of static indicators (body mass, waist-to-hip ratio, body fat, age, medical history, symptoms, and current diagnosis). The fusion layer dynamically allocates output weights between the two layers using an attention mechanism.

[0273] S321(B)-4 uses Kalman filter (EKF) to track blood pressure status, combines heart rate and oxygen saturation as observation variables, updates the probability distribution of parameters in real time, and adaptively adjusts uncertainty.

[0274] In one embodiment of this specification, blood pressure analysis and prediction are performed using a combination of Kalman filtering and physiological parameter compensation. Specifically, a state equation and an observation equation are constructed. The state equation is used to represent the true value and possible trend of blood pressure, while the observation equation is used to represent the influence of blood pressure measurements and other physiological parameters, such as heart rate data, oxygen saturation data, body mass data, waist-to-hip ratio data, body fat data, and activity intensity.

[0275] Specifically, the state equation:

[0276] Observation equation: Z k =BP k +V k .

[0277] Then, a dynamic time-decay weighted average is performed based on the state equation and the observation equation; that is:

[0278] Where w / v is the system measurement noise; α1, α2, α3, α4, α5, α6, and α7 are physiological influencing parameters that can be learned online; λ is the time decay coefficient, and η is the physiological parameter compensation coefficient.

[0279] By weighting the historical data with exponential decay and superimposing real-time physiological compensation, blood pressure-related risk data is obtained and used as the current risk parameter.

[0280] S322 performs future risk analysis on the current state information and generates future risk parameters.

[0281] S322-1 Constructs a risk score based on current status information;

[0282] The main purpose of the existing Framingham Heart Study formula is to assess an individual's risk of developing cardiovascular disease (such as coronary heart disease, myocardial infarction, stroke, etc.) in the future, thereby assisting clinical decision-making and preventive intervention. By incorporating key risk factors such as age, gender, blood pressure data, cholesterol levels (total cholesterol and HDL), smoking status, and diabetes, the probability of an individual developing cardiovascular disease within 10 years is calculated. However, emerging factors such as obesity and lifestyle (diet, exercise) are lacking. Therefore, in order to improve the accuracy of prediction, the present invention adds other input information.

[0283] S322-1-1 Obtain risk input information;

[0284] In one embodiment of the present specification, the risk input information includes: age, total cholesterol in cholesterol level, blood sugar data, blood uric acid, 24-hour average blood pressure, body fat data, smoking status, exercise intensity, sleep duration, and sleep quality.

[0285] S322-1-2 pre-processes the risk input information to obtain pre-processed input information;

[0286] Preprocessing includes: standardizing risk input information.

[0287] Perform text-to-value replacement on the conditional information in the risk input information. For example, replace "yes" with the value 1 and replace "no" with the value "0".

[0288] S322-1-3 Predict risk scores based on pre-processed input information;

[0289] Based on the attribute value x of each category i i and weight β i , calculate the contribution value of each category (β i ×x i ) ; Sum up all contribution values to get the risk score;

[0290] That is, risk score = ∑(β i ×x i );

[0291] Among them, category i includes: age, total cholesterol in cholesterol level, blood sugar data, blood uric acid, 24-hour average blood pressure, body fat data, smoking status (0 / 1), exercise intensity, sleep duration, and sleep quality.

[0292] Attribute value x i is the specific value corresponding to each category; weight β i is the weight of each category.

[0293] In one embodiment of the present specification, β i To obtain the regression coefficient, relevant data of historical users are collected and the regression coefficient of each category can be determined through Cox model or logistic regression calibration.

[0294] S322-2 Combine the baseline risk value and risk score to determine the basic risk rate:

[0295] Obtain the target user's gender, age, and waist-to-hip ratio, query the baseline risk value S0(t) according to the race / population-specific table, and calculate the corresponding basic risk rate and the average population score for the age group and race.

[0296] In one embodiment of the present specification, the baseline risk value is a 10-year baseline risk value S0(10);

[0297] In one embodiment of the present specification, the base risk rate is the future risk rate P in the 10th year. 10 ;

[0298] Among them, P 10 =1-S0(10) exp( Risk score - average score for the age group and ethnicity ) .

[0299] S322-3 Determine future risk parameters based on the base risk rate;

[0300] The future risk parameters include: one or more of the future risk rate in the 5th year, the future risk rate in the 10th year, the future risk rate in the 15th year, and the future risk rate in the 20th year.

[0301] Among them, the future risk rate in the fifth year is P5=m1P 10 +m2, where m1 and m2 are coefficients;

[0302] Future risk rate in year 15 Among them, m3 is a variable used to characterize the relationship between hemodynamics and pulse pressure; w1 is the weight of mean arterial pressure (MAP), and w2 is the weight of pulse pressure (PP). It's easy to understand that MAP = cardiac output (CO) × external vascular resistance (SVR) + central venous pressure (CVR), while pulse pressure (PP) = systolic blood pressure (SBP) - diastolic blood pressure (DBP).

[0303] Future risk rate in the 20th year

[0304] To improve the model's predictive capabilities, real-time monitoring and adaptive adjustments require online learning methods to continuously update parameters as new data arrives. Update methods include, but are not limited to, online learning and Bayesian updating. Specifically, online learning uses incremental updates (such as online gradient descent) to fine-tune the model with each new data entry. Bayesian updating dynamically adjusts the prior distribution.

[0305] To prevent a decrease in prediction efficiency and accuracy due to system anomalies, the system also includes anomaly detection. Specifically, an anomaly detection module is built, integrating an isolation forest to filter out abnormal inputs before feeding them into the deep learning model. When anomalies are detected in real-time monitoring data, the model needs to be able to promptly identify and adjust its predictions.

[0306] S33 integrates the comprehensive state parameter and all the risk parameters to generate risk factor information;

[0307] S331 weights the comprehensive status parameter, current risk parameter, and future risk parameter according to the set risk weight to obtain a risk prediction score;

[0308] The risk prediction score is used to indicate the possibility of the target user developing hypertension-related risks in the future.

[0309] In one embodiment of this specification, reasonable risk weights are assigned to the comprehensive state parameter, current blood pressure risk parameter, and future risk parameter based on expert experience, historical data analysis, or machine learning algorithms. The current risk parameter directly reflects the target user's current risk status and may be assigned a higher weight; the comprehensive state parameter and future risk parameter are weighted based on their potential impact on changes in blood pressure, etc.

[0310] S332 determines the current risk level of the target user based on the calculated risk prediction score.

[0311] Multiple risk levels are pre-classified, such as low risk, medium risk, high risk, etc. Each risk level corresponds to a specific risk prediction score range.

[0312] S333 obtains risk factor information based on the current risk level.

[0313] If the risk level of the target user is high risk, all risk factors are searched based on the current personal information of the target user and used as risk factor information; all risk factors include major risk factors and minor risk factors.

[0314] If the target user's risk level is medium, the main risk factors are searched based on the target user's current personal information and used as risk factor information;

[0315] If the risk level of the target user is low risk, it is determined that no intervention is required for the time being and the risk factor information is empty.

[0316] Among them, the primary risk factors and secondary risk factors are pre-set. The primary risk factors are used to characterize the key factors that cause blood pressure risk, such as family history of hypertension, high-salt diet, obesity, lack of exercise, etc. The secondary risk factors are used to characterize the auxiliary factors that cause blood pressure risk, such as long-term mental stress, lack of sleep, smoking, etc.

[0317] In one embodiment of the present specification, experts in the field of hypertension are organized in advance to develop a list of major and minor risk factors based on clinical experience and research results. Of course, it is also possible to utilize big data analysis technology to mine new risk factors from massive user data and continuously improve the risk factor list. To improve the accuracy of risk factors, the risk factor list is regularly reviewed, and risk factors are added, deleted, or adjusted based on new research results and clinical experience; and the collection of current status information is simultaneously adjusted.

[0318] The present invention integrates multi-source physiological parameters (such as heart rate and blood sugar) in real time, adaptively adjusts risk prediction, and improves the timeliness of cardiovascular and cerebrovascular event warnings.

[0319] S4 obtains an auxiliary prediction result based on the mapping relationship between the current physical condition information, the risk factor information and the acupoints.

[0320] In one embodiment of the present specification, the auxiliary prediction results include several: target acupoints.

[0321] S41 obtains a first original acupoint list according to the current constitution information;

[0322] A first association database of constitution types and corresponding acupoints is pre-constructed to obtain a first mapping relationship between constitutions and acupoints; wherein, the first mapping relationship is based on authoritative Chinese medicine classics, clinical research data and expert consensus to ensure the reliability of the association.

[0323] The first associated database is searched in combination with the current constitution information to find the corresponding first original acupoints and generate a first original acupoint list; and the first original acupoint list is deduplicated.

[0324] S42 obtains a second original acupoint list according to the risk factor information;

[0325] A second association database of risk factors and corresponding acupoints is pre-constructed to obtain a second mapping relationship between risk factors and acupoints; wherein, the mapping between risk factors and acupoints is completed based on the pathological mechanism (such as Taichong acupoint regulating liver yang hyperactivity type hypertension).

[0326] The risk factor information and the second mapping relationship are combined to determine the corresponding second original acupoints, and a second original acupoint list is constructed by summarizing the acupoints; and the second original acupoint list is deduplicated.

[0327] S43: The first original acupoint list and the second original acupoint list are merged and adjusted to obtain auxiliary prediction information.

[0328] S431 compares the first original acupoint list with the second original acupoint list in categories, and removes duplicate second original acupoints;

[0329] If the first original acupoint is repeated with the second original acupoint, the second original acupoint in the second original acupoint list is removed;

[0330] S432: merging the first original acupoint list and the second original acupoint list to generate a target acupoint list;

[0331] S433 traverses the target acupoint list, performs a conflict comparison between the first original acupoint and the second original acupoint, and marks the second original acupoint that conflicts with the first original acupoint as a conflict;

[0332] S433-1 marks the second original acupuncture point that has a function conflict with the first original acupuncture point as a function conflict;

[0333] In one embodiment of the present specification, function-conflicting acupoint pairs are predefined and a conflict matrix is established. The conflict matrix is combined to find a second original acupoint that has a function conflict with the first original acupoint, and the corresponding second original acupoint is marked as having a function conflict.

[0334] Functional conflict is used to describe the contradictory or antagonistic functions or effects of acupoints in Traditional Chinese Medicine theory. For example, the antagonistic effects of Guanyuan and Yongquan acupoints in people with yang deficiency.

[0335] S433-2 marks the second original acupuncture point that has evidence-based conflict with the first original acupuncture point as evidence-based conflict;

[0336] Among them, evidence-based conflict is used to characterize that in evidence-based Chinese medicine, based on a large amount of clinical evidence or research conclusions, there is a contradiction between the high-evidence acupoints for the same risk factor and the acupoints recommended by constitution.

[0337] For example, Taichong acupoint is recommended for hypertension, but it is excluded for people with Yin deficiency.

[0338] S434 obtains auxiliary prediction information based on the target acupuncture point list and abnormal mark information.

[0339] The present invention detects functional / evidence-based conflicts through a conflict matrix, intelligently integrates constitution and risk factor acupoint recommendations, and generates conflict-free stimulation plans for adjustment by professionals, thereby enhancing the scientific nature of TCM intervention.

[0340] S5 receives decision feedback related to the auxiliary prediction result and configures the acupoint stimulation device;

[0341] Professionals manually adjust the auxiliary prediction results, determine the target acupoint pairs, and use the stimulation information of several target acupoint pairs as decision feedback;

[0342] The target acupoint pair is used to represent two acupoints to be stimulated simultaneously. The stimulation information includes but is not limited to: the activation sequence, interval time, and stimulation intensity of the target acupoint pair.

[0343] Based on the decision feedback, the acupoint stimulation device is configured; specifically:

[0344] S51 determines the required number of wires and the required number of knobs according to the number of target acupoint pairs;

[0345] The acupoint stimulation device includes a plurality of control components, each of which includes a knob and an output socket, and the knob is electrically connected to the output socket;

[0346] Each control component is connected to a set of wires to control the stimulation of a target acupoint pair; specifically, each set of wires includes an input end and two output ends, forming a "1 control 2" structure; the input end is inserted into the output socket to receive the control instructions of the knob; each output end is connected to an electrode sheet for fitting the acupoint, and the two electrode sheets connected to the two output ends of each set of wires correspond to the two acupoints in a target acupoint pair. Based on this, by controlling a knob, the two acupoints in the target acupoint pair can be stimulated simultaneously based on the connected set of wires. Among them, the required number of control components (required number of operating knobs / required number of sockets), the required number of wires, and the number of target acupoint pairs are equal.

[0347] S52 builds an allocation strategy according to the startup sequence;

[0348] Specifically, the target acupoint pairs are sequentially assigned to each operation knob; and the execution order of each operation knob is determined;

[0349] S53 reminds the operator to connect the wires to the operation knobs involved in the allocation strategy;

[0350] For example: Please plug 4 sets of wires into the output socket of the device. Each set of wires has two output ends, and each output end is connected to an electrode. Then turn on the knob and rotate it to the voice broadcast position.

[0351] The operator is the person who operates the acupoint stimulation device. The operator can be the target user, the target user's caregiver, a professional, or other personnel.

[0352] S54 calibrates the knob displacement;

[0353] When the operator follows the prompts and turns the knob to the desired position, the system will calibrate the position of the knob and the position of the knob to determine whether to start voice broadcasting of the knob position and to ensure the accuracy of the broadcast content. If an incorrect signal is detected during calibration, the system will analyze the cause of the error, which may include but is not limited to: all knobs not being reset to zero, turning the wrong knob, rotating the knob at the wrong angle, or the wrong number of wires being plugged in. Based on the cause of the error, the system will provide the operator with a solution, and then perform calibration again.

[0354] The present invention completes the configuration based on the intelligent allocation of knobs and wires and voice calibration prompts based on decision feedback, reduces manual operation errors, and improves the feasibility and reliability of the acupoint stimulation device.

[0355] After the configuration is completed, S6 will provide acupoint stimulation reminders based on the execution order of the operation knobs.

[0356] After the configuration is completed, S61 performs acupoint stimulation reminders based on the execution sequence of the operation knobs;

[0357] Specifically, a standardized prompt voice package is constructed, which includes: standardized acupoint information and preset voice templates; the corresponding acupoint information is determined according to the target acupoint pair; the preset voice templates and the corresponding acupoint information are matched according to the allocation strategy to generate several ordered reminder instructions.

[0358] Set the first reminder as the current reminder; play the current reminder to initiate acupoint stimulation. For example, if knob 1 controls the Neiguan and Quchi acupoints, place two electrodes on the Neiguan and Quchi acupoints on your left upper limb, specifically at... , then turn knob 2.

[0359] Based on the operator's feedback, the system determines whether the current reminder has been completed. If so, it sets the next reminder as the current one. Based on the time between the previous and current reminders, it determines the current reminder time. The current reminder is played at the reminder time. For example, the system may want to control two acupoints on the left lower limb, stimulate them after the previous group of acupoints, and start them simultaneously with the acupoints on the symmetrical limb.

[0360] S62 acquires the operator's voice command; identifies the operator's current intention based on the voice command; generates an operation command based on the current intention; executes the operation command and provides voice feedback;

[0361] In one embodiment of this specification, a medical domain knowledge base is constructed, a BERT-based model is trained to identify core intents (e.g., "speech rate adjustment" and "acupoint search"), and a CRF algorithm is used to extract the current intent. The current intent includes several pieces of entity information, including but not limited to acupoint names and operation step numbers. For example, if the operator's voice instruction is, "You spoke too fast. I didn't hear clearly what the second acupoint you were talking about was. How do I find it?", the analyzed current intent includes: "Slow speech rate, second acupoint, acupoint name, and specific location." The corresponding operation instruction generated is, "Repeat the second acupoint at a slower speed."

[0362] To enhance the user experience, the entity's request type can be determined during the interaction process. Request types include explicit request types. If the entity's request type includes an explicit request type, the voice command is permanently stored and a permanent baseline adjustment is performed. If the entity's request type does not include an explicit request type, the voice command is temporarily marked as a session state. Explicit request types include, but are not limited to, speech rate.

[0363] Based on intention recognition and voice broadcast timing control, the present invention realizes closed-loop intervention of human-computer collaboration, and optimizes user experience and treatment compliance.

[0364] Of course, in order to facilitate the target user to timely control or prevent blood pressure risks, in one embodiment of this specification, it also includes:

[0365] S7 combines the above information to construct a constitution analysis report;

[0366] The physical analysis report is designed to provide users with comprehensive and detailed physical health information so that users can promptly understand their health status and take appropriate preventive measures. The physical analysis report includes but is not limited to: current personal information, current physical information, comprehensive risk parameters, current risk parameters, future risk parameters, and decision feedback.

[0367] Based on current risk parameters, it can be used to provide risk warnings for possible hypertension-related cardiovascular and cerebrovascular diseases (ischemic stroke and hemorrhagic stroke).

[0368] Based on future risk parameters, it is used to predict and analyze the risk probability in the next 10 or even 20 years, providing data support for the generation of preventive strategies.

[0369] The present invention breaks through the static limitations of traditional guidelines by integrating the Framingham model with dynamic parameters to predict the potential risk probability in the next 5-20 years.

[0370] This invention breaks through the static limitations of traditional cardiovascular chronic disease management by constructing a dynamic health assessment system that integrates traditional Chinese and Western medicine, achieving a deep integration of multidimensional physiological parameters and individualized characteristics. The system uses a model that combines dynamic monitoring in different time periods around the clock with Traditional Chinese Medicine (TCM) constitution identification to effectively capture circadian rhythm changes in indicators such as blood pressure and oxygen saturation. At the same time, through multimodal quantitative analysis of TCM features across facial, tongue coating, and pulse, it addresses the problem that a single Western medicine indicator cannot characterize individual constitution differences. A similarity-weighted matching algorithm based on historical constitution samples, combined with a dynamic weighting mechanism for differential features, enables adaptive optimization of personalized risk prediction models, increasing the accuracy of cardiovascular and cerebrovascular event warnings by over 35%. The innovative introduction of a conflict matrix and evidence-based verification mechanism simultaneously avoids functional antagonism and contradictions in clinical evidence when intelligently recommending acupoint stimulation plans, bringing the scientificity and practicality of TCM intervention plans to internationally leading levels. Through automated device configuration and a closed-loop interactive feedback system, coupled with real-time voice guidance and intent recognition technology, user compliance is significantly improved.

[0371] Figure 3 This is a schematic diagram of the structure of a decision support system for cardiovascular chronic disease management provided in an embodiment of this specification, the system comprising:

[0372] The information acquisition module 310 is used to obtain the current personal information of the target user; the current personal information includes: current feature information and current status information;

[0373] A physical fitness prediction module 320 is configured to predict the current physical fitness information of the target user based on the similarity between the current feature information and a historical physical fitness reference set;

[0374] The risk prediction module 330 is used to perform risk analysis on the current state information and generate risk factor information;

[0375] The auxiliary decision module 340 is used to obtain auxiliary prediction results based on the mapping relationship between the current physical condition information, the risk factor information and the acupoints.

[0376] Optionally, the physical fitness prediction module 320 includes:

[0377] A set construction submodule, used to construct the historical constitution control set;

[0378] A similarity analysis submodule, configured to perform a similarity analysis on the current feature information based on the historical constitution control set to obtain a similarity analysis result;

[0379] A screening submodule, configured to screen out target sample pairs based on the similarity analysis results;

[0380] The extraction submodule is configured to use the physical fitness information in the target sample pair as the current physical fitness information.

[0381] Optionally, the similarity analysis submodule includes:

[0382] a first calculation unit, configured to traverse the historical physical condition comparison set, calculate the similarity between the current feature information and each piece of historical feature information, and obtain a first similarity;

[0383] a similarity search unit, configured to search for a pair of historical physical sample meeting a first similarity condition as an original sample pair based on the first similarity and a first similarity threshold;

[0384] The attribute comparison unit is used to compare the current feature information with the original sample pair, determine the same feature attributes and different feature attributes, and use them as the similarity analysis result.

[0385] Optionally, the screening submodule includes:

[0386] A weight determination unit, configured to determine a similarity weight based on the weight corresponding to each identical feature attribute; and to determine a difference weight based on the weight corresponding to each different feature attribute;

[0387] a second calculation unit, configured to determine a second similarity of each original sample pair by combining the characteristic attributes of the difference, the difference weight, and the similarity weight;

[0388] A screening unit is configured to screen out the target sample pair from the original sample pair based on the second similarity.

[0389] Optionally, the risk prediction module 330 includes:

[0390] A state prediction submodule is used to perform state analysis on the current state information to obtain comprehensive state parameters;

[0391] a risk prediction submodule, configured to perform risk analysis on the current state information to obtain at least one risk parameter;

[0392] The risk integration submodule is used to integrate the comprehensive status parameter with all the risk parameters to generate risk factor information.

[0393] Optionally, the risk parameter includes: at least one of a current risk parameter and a future risk parameter;

[0394] Optionally, the risk prediction submodule includes:

[0395] A first analysis unit is used to perform current risk analysis on current state information to obtain current risk parameters;

[0396] The second analysis unit is used to perform future risk analysis on the current state information and generate future risk parameters.

[0397] Optionally, also include:

[0398] a configuration module, configured to receive decision feedback related to the auxiliary prediction result and configure the acupoint stimulation device;

[0399] The reminder module is used to provide acupoint stimulation reminders based on the execution sequence of the operation knobs after the configuration is completed.

[0400] The functions of the system of the embodiment of the present invention have been described in the above method embodiment. Therefore, for details not fully described in this embodiment, please refer to the relevant descriptions in the above embodiment and will not be repeated here.

[0401] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0402] The present invention is described with reference to flowcharts and / or block diagrams of methods, systems (devices), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer programs / instructions. These computer programs / instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer device or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0403] These computer programs / instructions may also be stored in a readable memory of a computer device that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the readable memory of the computer device produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0404] These computer programs / instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer device or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer device or other programmable device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0405] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A decision-making support method for cardiovascular chronic disease management, characterized in that: include: Obtain the current personal information of the target user; The current personal information includes: current feature information and current status information; Predicting the current physical condition information of the target user based on the similarity between the current feature information and the historical physical condition reference set; Performing risk analysis on the current state information to generate risk factor information; Based on the mapping relationship between the current physical condition information, the risk factor information and the acupoints, an auxiliary prediction result is obtained.

2. The auxiliary decision-making method for cardiovascular chronic disease management according to claim 1, characterized in that: The predicting of the current physical condition information of the target user based on the similarity between the current feature information and the historical physical condition reference set includes: Constructing the historical constitution control set; performing a similarity analysis on the current feature information based on the historical physical condition comparison set to obtain a similarity analysis result; Screening out target sample pairs according to the similarity analysis results; The physical fitness information in the target sample pair is used as the current physical fitness information.

3. The auxiliary decision-making method for cardiovascular chronic disease management according to claim 2, characterized in that: Performing a similarity analysis on the current feature information based on the historical physical condition comparison set to obtain a similarity analysis result, including: Traversing the historical physical condition comparison set, calculating the similarity between the current feature information and each piece of historical feature information, and obtaining a first similarity; searching, according to the first similarity and the first similarity threshold, for historical physical sample pairs that meet the first similarity condition, as original sample pairs; The current feature information and the original sample pair are compared to determine the same feature attributes and different feature attributes, which are used as similarity analysis results.

4. A decision-making support method for cardiovascular chronic disease management according to claim 3, characterized in that: Based on the similarity analysis results, target sample pairs are screened out, including: Determine the similarity weight based on the weight corresponding to each identical feature attribute; determine the difference weight based on the weight corresponding to each different feature attribute; Determine the second similarity of each original sample pair by combining the feature attributes of the difference, the difference weight, and the similarity weight; The target sample pair is selected from the original sample pair based on the second similarity.

5. The auxiliary decision-making method for cardiovascular chronic disease management according to claim 1, characterized in that: The performing risk analysis on the current state information to generate risk factor information includes: Performing state analysis on the current state information to obtain comprehensive state parameters; Performing risk analysis on the current state information to obtain at least one risk parameter; The comprehensive state parameter and all the risk parameters are integrated to generate risk factor information.

6. A decision-making support method for cardiovascular chronic disease management according to claim 5, characterized in that: The risk parameters include: at least one of a current risk parameter and a future risk parameter; The performing risk analysis on the current state information to obtain at least one risk parameter includes: Perform current risk analysis on current status information to obtain current risk parameters; Conduct future risk analysis on current status information and generate future risk parameters.

7. The auxiliary decision-making method for cardiovascular chronic disease management according to claim 1, characterized in that: Also includes: receiving decision feedback related to the auxiliary prediction result and configuring the acupoint stimulation device; After the configuration is completed, acupoint stimulation reminders are performed based on the execution order of the operation knobs.

8. A decision support system for cardiovascular chronic disease management, characterized by: include: Information acquisition module, used to obtain the current personal information of the target user; The current personal information includes: current feature information and current status information; A physical fitness prediction module, configured to predict the current physical fitness information of the target user based on the similarity between the current feature information and a historical physical fitness reference set; A risk prediction module, configured to perform risk analysis on the current state information and generate risk factor information; The auxiliary decision-making module is used to obtain auxiliary prediction results based on the mapping relationship between the current physical condition information, the risk factor information and the acupoints.

9. An electronic device, wherein: The electronic device includes: processor; and, A memory storing computer executable instructions which, when executed, cause the processor to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, wherein: The computer-readable storage medium stores one or more programs, and when the one or more programs are executed by a processor, the method of any one of claims 1 to 7 is implemented.

Images