Chronic disease health management system based on health behavior change

Abstract

The invention relates to the technical field of health, in particular to a chronic disease health management system based on health behavior change, which comprises a user operation unit used for uploading patient chronic disease data, behavior data and user videos of a user; the data analysis unit is used for determining a health information confirmation mode according to the physical sign data expression value and the diet image level value, namely performing health information search according to the chronic disease data of the patient or performing health information search according to the behavior data and the effective reference image value; the data optimization unit is used for responding to a preset screening condition so as to determine an information screening mode according to the content deviation degree of the health recommendation information to be screening according to a browsing effective reference value or screening according to a user preference value; the data screening unit is used for executing an information screening mode; the recommendation matching efficiency of the health information is improved.

Description

Technical Field

[0001] The present invention relates to the field of health technologies, and particularly to a chronic disease health management system based on health behavior change. Background Art

[0002] With the booming development of medical informatization, information-based means such as electronic prescriptions, medical staff platforms, smart medical terminals, and disease management systems have, to a certain extent, promoted the integration and convenience of medical information. Among them, the disease management system can realize the ability to recommend feasible health management information corresponding to the behavior data of patients. However, due to the strong particularity of patients' personal reading preferences and personal living habits, the matching degree of the recommended information often fails to meet the actual needs. Therefore, how to improve the recommended matching degree of health management information for chronic disease patients is a technical problem that needs to be solved urgently by those skilled in the art.

[0003] Chinese Patent Publication No. CN115982460A discloses a personalized recommendation method, system, and medium for health popular science information. The information collection step: collecting users' health information; the information processing step: processing the collected users' health information to obtain the feature tags of the users; the matching step: matching the feature tags of the users with popular science materials to form a candidate push library; the push step: pushing the content of the candidate push library according to preset rules; the collection step: collecting the operation behavior data of users on the pushed content; the tag improvement step: improving the feature tags of the users according to the operation behavior data and returning to execute the matching step. In the above technical solution, although it is considered to improve the feature tags of users according to the operation behavior data, the operation behavior data is fixed as the exposure times, click times, like behavior, comment behavior, share behavior, and stay duration, which cannot be adjusted according to the actual situation and is difficult to meet the effective matching of the recommended information. Summary of the Invention

[0004] For this reason, the present invention provides a chronic disease health management system based on health behavior change to overcome the problem that the matching degree of existing health recommendation information is difficult to meet the actual needs.

[0005] To achieve the above object, the present invention provides a chronic disease health management system based on health behavior change, including:

[0006] A user operation unit for uploading the patient's chronic disease data, behavior data, and user videos of the user;

[0007] A data analysis unit connected to the user operation unit for determining the health information confirmation method as searching for health information according to the patient's chronic disease data or searching for health information according to the behavior data and the effective reference image value based on the physical sign data performance value and the dietary image level value;

[0008] A data optimization unit, which is connected to the user operation unit and the data analysis unit, and is configured to respond to a preset screening condition to determine the information screening method as screening according to the browsing effective reference value or screening according to the user preference value based on the content deviation degree of the health recommendation information;

[0009] A data screening unit, which is connected to the data optimization unit and the user operation unit, and is configured to execute the information screening method. When screening according to the browsing effective reference value, it detects the browsing depth and browsing frequency of each health recommendation information to determine the browsing effective reference value. When screening according to the user preference value, it detects the information architecture difference degree of each health recommendation information to determine the user preference value.

[0010] Further, the data analysis unit determines the health information confirmation method according to the physical sign data performance value and the diet image level value;

[0011] If the physical sign data performance value is less than or equal to the preset physical sign data performance value and the diet image level value is greater than the preset diet image level value, the health information confirmation method is to search for health information according to the patient's chronic disease data;

[0012] If the physical sign data performance value is greater than the preset physical sign data performance value or the diet image level value is less than or equal to the preset diet image level value, the health information confirmation method is to search for health information according to the behavior data and the effective reference image value.

[0013] Further, the detection process of the data analysis unit for the diet image level value includes:

[0014] Detecting the standard frame similarity and block diversity of the user video;

[0015] Determining whether to perform frame compensation according to the standard frame similarity and block diversity;

[0016] Under the preset execution condition, determining the diet image level value according to the color distribution degree and color richness of the frame to be analyzed;

[0017] The preset execution condition is that the standard frame similarity is less than or equal to the preset standard frame similarity and the block diversity is greater than the preset block diversity or the frame compensation is completed.

[0018] Further, the data analysis unit responds to the preset compensation condition to perform frame compensation on the standard frame, including:

[0019] Detecting the dynamic change value of the standard frame, recording the standard frame with the dynamic change value greater than the preset dynamic change value as the compensation reference frame, extracting video frames for the video interval between the compensation reference frames, and determining the number of video frames extracted in the video interval according to the dynamic change average value;

[0020] The preset compensation condition is that the similarity of the standard frame is greater than the preset similarity of the standard frame and the block diversity is less than or equal to the preset block diversity.

[0021] Further, the data optimization unit responds to the preset screening condition and determines the information screening method according to the content deviation degree of the health recommendation information;

[0022] If the content deviation degree is less than the preset content deviation degree, the information screening method is to screen according to the browsing effective reference value;

[0023] If the content deviation degree is greater than or equal to the preset content deviation degree, the information screening method is to screen according to the user preference value;

[0024] The preset screening condition is that the number of health recommendation information obtained by searching is greater than the preset number of health recommendation information.

[0025] Further, when the data screening unit screens according to the browsing effective reference value, it detects the browsing depth and browsing frequency of each health recommendation information to determine the browsing effective reference value;

[0026] Select the health recommendation information of the recommended quantity in the order from large to small of the browsing effective reference value, and record it as the recommendable information;

[0027] The browsing effective reference value has a positive correlation with both the browsing depth and the browsing frequency.

[0028] Further, the data screening unit is provided with a browsing depth detection strategy. Among them, if the graphic bias coefficient of the health recommendation information is greater than the preset graphic bias coefficient, the data screening unit determines the invalid image according to the image relevance, and determines the browsing depth according to the reading duration of the invalid image;

[0029] If the graphic bias coefficient of the health recommendation information is less than or equal to the preset graphic bias coefficient, the data screening unit determines the browsing depth according to the browsing stay duration.

[0030] Further, the graphic bias coefficient is determined according to the image distribution coefficient and the image distribution ratio;

[0031] The graphic bias coefficient has a negative correlation with the image distribution coefficient;

[0032] The graphic bias coefficient has a positive correlation with the image distribution ratio.

[0033] Further, when the data screening unit screens according to the user preference value, it detects the information architecture difference degree of each health recommendation information to determine the user preference value;

[0034] Select the health recommendation information of the recommended quantity in the order from large to small according to the user preference value, which is recorded as the recommendable information.

[0035] Further, the chronic disease health management system further includes a database for storing health recommendation information.

[0036] Compared with the prior art, the beneficial effect of the present invention is that in the technical solution of the present invention, the data analysis unit determines the health information confirmation method according to the physical sign data performance value and the dietary image level value, and reflects whether the user's behavior data can meet the requirements of effective information search through the physical sign data performance value and the dietary image level value, and correspondingly selects different health information confirmation methods, so that the health information confirmation method can meet the actual application scenario, thereby improving the subsequent health information search efficiency.

[0037] Further, in the present invention, the data analysis unit responds to a preset compensation condition to perform frame number compensation for the standard frame, avoiding the problem of large analysis errors of the dietary image level value caused by the influence of the video shooting method or the contingency of video frame extraction, thereby improving the accuracy of the dietary image level value and further improving the determination accuracy of the health information confirmation method.

[0038] Further, in the present invention, the data optimization unit responds to a preset screening condition to determine the information screening method according to the content deviation degree of the health recommendation information, avoiding the problem of redundant recommended information caused by a large number of health recommendation information. At the same time, the content deviation degree reflects the difference degree of the text content between the health recommendation information, and adaptively selects different information screening methods, so that the information screening method is more in line with the actual text state, thereby improving the matching degree of subsequent information recommendation.

[0039] Further, in the present invention, the data screening unit is provided with a browsing depth detection strategy, and the distribution state of the image and text of the health recommendation information is reflected through the graphic and text deviation coefficient of the health recommendation information, so as to calculate the browsing depth specifically, improve the accuracy of the browsing depth, and further improve the matching degree of subsequent information recommendation. Description of the Drawings

[0040] Figure 1 It is the unit connection diagram of the chronic disease health management system based on health behavior change of the present invention;

[0041] Figure 2 It is the flowchart of determining the health information confirmation method according to the physical sign data performance value and the dietary image level value of the present invention;

[0042] Figure 3 It is the flowchart of determining the information screening method according to the content deviation degree of the health recommendation information of the present invention. Detailed Embodiment

[0043] In order to make the objectives and advantages of the present invention more clear and understandable, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0044] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0045] It should be noted that in the description of the present invention, the terms indicating directions or positional relationships such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0046] In addition, it should also be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0047] Please refer to Figures 1 to 3 As shown, the present invention provides a chronic disease health management system based on health behavior change, including:

[0048] A user operation unit for uploading the patient's chronic disease data, behavior data, and user videos of the user;

[0049] A data analysis unit connected to the user operation unit for determining the health information confirmation method as searching for health information based on the patient's chronic disease data or searching for health information based on the behavior data and the effective reference image value according to the physical sign data performance value and the dietary image level value;

[0050] A data optimization unit connected to the user operation unit and the data analysis unit for responding to a preset screening condition to determine the information screening method as screening according to the browsing effective reference value or screening according to the user preference value according to the content deviation degree of the health recommendation information;

[0051] A data screening unit, which is connected to the data optimization unit and the user operation unit, is used to execute an information screening method. When screening according to the browsing effective reference value, it detects the browsing depth and browsing frequency of each health recommendation information to determine the browsing effective reference value. When screening according to the user preference value, it detects the information architecture difference degree of each health recommendation information to determine the user preference value;

[0052] The user operation unit is also used to display the recommendable information, and the user can choose whether to like the displayed recommendable information. The specific implementation of the like behavior can be achieved but is not limited to double-clicking the recommendable information or clicking the pre-set like option.

[0053] The user data uploaded by the user operation unit is the data of patient users. Patient users can track and obtain behavior data by wearing a smart bracelet. The chronic disease data of patients is obtained by the patient users through medical examinations and then uploaded by themselves after obtaining the doctor's test results.

[0054] This invention applies a number of historical records, which are stored in a database. Any historical record records the work records of the chronic disease health management system for any period of time and records a qualified mark. The qualified mark shows whether the amount of historical records meets the user's needs. Whether it meets the user's needs can be determined according to the recommendation matching effect. For example, a user satisfaction survey can be set, and the satisfaction is used as a judgment index. If the satisfaction is within the acceptable range, the recommendation matching effect is qualified, that is, the historical record meets the user's needs. The user mentioned in the setting of the preset value in this invention is the management user of the chronic disease health management system.

[0055] It should be noted that the preset threshold in this invention can be obtained but is not limited to through human experience or through training and learning of historical records by a deep learning model. In the specific implementation of this invention, for the confirmation method of any preset threshold, the detection value corresponding to the preset threshold of the historical record that meets the user's needs is extracted. For example, for the preset diet image level value, its corresponding detection value is the diet image level value. The diet image level values of the historical records that meet the user's needs are extracted and data cleaning is performed to remove the outliers. The method for confirming the outliers can be the Z-Score method or the IQR method. The average value of the diet image level values after removing the outliers is recorded as the value of the preset diet image level value, and the preset value can be optimized through a deep learning model. This is easy for those skilled in the art to understand and will not be elaborated here.

[0056] Specifically, the data analysis unit determines the health information confirmation method according to the physical sign data performance value and the diet image level value;

[0057] If the manifestation value of the physical sign data is less than or equal to the preset manifestation value of the physical sign data and the dietary image level value is greater than the preset dietary image level value, the health information confirmation method is to search for health information based on the patient's chronic disease data;

[0058] If the manifestation value of the physical sign data is greater than the preset manifestation value of the physical sign data or the dietary image level value is less than or equal to the preset dietary image level value, the health information confirmation method is to search for health information based on the behavior data and the effective reference image value.

[0059] Among them, the confirmation method of the manifestation value of the physical sign data is to detect the exercise time when the exercise intensity of the patient is greater than the preset exercise intensity, calculate the absolute value of the difference between the exercise time and the preset exercise time, and record it as the manifestation value of the physical sign data. The exercise intensity can be detected by, but not limited to, a smart bracelet. When the patient exercises, the heart rate detected by the smart bracelet is recorded as the exercise intensity. The value of the preset exercise intensity is set in advance by a health doctor. It can be understood that the doctor can set the recommended preset exercise intensity according to his own experience and the actual physical condition of the patient. This is easy to understand for those skilled in the art and will not be elaborated here. In the specific implementation of the present invention, the preset exercise intensity is 50% of the maximum heart rate of the patient, and the maximum heart rate of the patient is 208-(0.7×age). The value of the preset manifestation value of the physical sign data is 10% of the preset exercise intensity.

[0060] Specifically, the detection process of the dietary image level value by the data analysis unit includes:

[0061] Detect the standard frame similarity and block diversity of the user video;

[0062] Determine whether to perform frame compensation according to the standard frame similarity and block diversity;

[0063] Under the preset execution conditions, determine the dietary image level value according to the color distribution degree and color richness of the frame to be analyzed;

[0064] The preset execution conditions are that the standard frame similarity is less than or equal to the preset standard frame similarity and the block diversity is greater than the preset block diversity, or the frame compensation is completed.

[0065] The completion of the frame compensation means that all the compensation frames have been extracted.

[0066] Among them, the user video is a video of food taken by the user himself / herself. A number of video frames are evenly collected from the user video and recorded as standard frames. The number of collected standard frames is determined according to the duration of the user video. The longer the duration of the user video, the greater the number of standard frames. In the specific implementation of the present invention, a standard frame is extracted every 10 s. The method for confirming the similarity of the standard frames is to detect the sub-similarity between adjacent standard frames, and the average value of the sub-similarities is recorded as the standard frame similarity. Among them, for the sub-similarity between any two adjacent standard frames, the detection method is to use a pre-trained deep learning model to extract the feature vectors of the two standard frames, and then calculate the cosine similarity between the feature vectors, and this cosine similarity is recorded as the sub-similarity. This is easy to understand for those skilled in the art and will not be elaborated here. The method for confirming the block diversity is to detect the sub-region diversity of each standard frame, and the average value of the sub-region block diversities is recorded as the block diversity. For a single reference frame, the method for confirming the corresponding sub-block diversity is to select several pixel blocks in the reference frame and record them as seed points, and perform regional segmentation on each seed point in a random order. For a single seed point, detect the pixel blocks adjacent to the seed point, merge the pixel blocks that meet the merging condition with the seed point and record them as the current region, and continue to detect the pixel blocks adjacent to the current region. When there are no pixel blocks that meet the merging condition, stop the detection and record the current region as a segmentation region. The merging condition is that the pixel value difference from the seed point is less than the preset pixel difference and has not been included in the segmentation region. The selection of the seed point can be manual selection or random selection or uniform selection through computer software. This is easy to understand for those skilled in the art and will not be elaborated here. The preset pixel difference is 10. Detect the average pixel value of each segmentation region. The calculation formula for the sub-region block diversity H is:

[0067]

[0068] Among them, Hi is the average pixel value of the i-th segmentation region, H0 is the average value of the average pixel values of all segmentation regions, n is the number of segmentation regions, and i = 1, 2, 3,..., n.

[0069] Both the standard frame and the compensation frame are regarded as frames to be analyzed. The confirmation method for the segmentation area of the compensation frame is the same as that of the standard frame. The color richness of the frame to be analyzed = the number of segmentation areas of all frames to be analyzed / the preset number of segmentation areas. The color distribution degree is the average value of the proportion of the segmentation area of each frame to be analyzed. For a single frame to be analyzed, the corresponding proportion of the segmentation area = the total area of the segmentation area / the area of the frame to be analyzed. The dietary image level value = the color distribution degree + the color richness. In the specific implementation of the present invention, the preset number of segmentation areas is the sum of the number of seed points of each frame to be analyzed × 50%. The value of the preset dietary image level value is the average value of the dietary image level values corresponding to the historical records that meet the user's needs, denoted as the preset dietary image level value.

[0070] Specifically, the data analysis unit responds to the preset compensation condition and performs frame number compensation for the standard frame, including:

[0071] Detect the dynamic change value of the standard frame, mark the standard frame with a dynamic change value greater than the preset dynamic change value as the compensation reference frame, extract video frames for the video interval between the compensation reference frames, and the number of video frames extracted in the video interval is determined according to the average dynamic change value;

[0072] The preset compensation condition is that the standard frame similarity is greater than the preset standard frame similarity and the block diversity is less than or equal to the preset block diversity.

[0073] For a single standard frame, the confirmation method for its corresponding dynamic change value is to extract the first 5 video frames and the last 5 video frames in its video time sequence respectively, detect the sub-similarity between adjacent video frames, extract the minimum sub-similarity, the dynamic change value = 1 / the minimum sub-similarity, the preset dynamic change value = 2, the video interval is the corresponding video segment between adjacent compensation reference frames, the number of video frames extracted in the video interval has a positive correlation with the average dynamic change value, and the extracted video frames are marked as compensation frames. In the specific implementation of the present invention, the number of video frames extracted in the video interval = the average dynamic change value × 1.5, and the average dynamic change value is the average value of the dynamic change values of the compensation reference frames corresponding to the video interval.

[0074] Specifically, the data optimization unit responds to the preset screening condition and determines the information screening method according to the content deviation degree of the health recommendation information;

[0075] If the content deviation degree is less than the preset content deviation degree, the information screening method is to screen according to the browsing effective reference value;

[0076] If the content deviation degree is greater than or equal to the preset content deviation degree, the information screening method is to screen according to the user preference value;

[0077] The preset screening condition is that the number of health recommendation messages obtained by searching is greater than the preset number of health recommendation messages.

[0078] The confirmation method of content deviation degree is to extract the effective keywords of each health recommendation message, and detect the number of effective keywords whose occurrence times are less than the preset occurrence times, which is recorded as the content deviation degree. The occurrence times are the number of health recommendation messages in which the effective keywords appear. In the specific implementation of the present invention, the preset occurrence times are 30% of the total number of health recommendation messages, and the preset content deviation degree is 50% of the total number of effective keywords. The effective keywords are set by the user in advance, and the effective keywords are keywords related to the treatment of chronic diseases, such as "chronic diseases", "chronic disease diet", and "kidney metabolic syndrome".

[0079] Specifically, when the data screening unit performs screening according to the browsing effective reference value, it detects the browsing depth and browsing frequency of each health recommendation message to determine the browsing effective reference value;

[0080] Select the health recommendation messages with the recommended quantity in the order from large to small according to the browsing effective reference value, which are recorded as the recommendable messages;

[0081] The browsing effective reference value has a positive correlation with both the browsing depth and the browsing frequency.

[0082] Browsing effective reference value = browsing depth × α1 + browsing frequency × α2, where α1 is the first weight coefficient and α2 is the second weight coefficient. In the implementation of the present invention, α1 = 0.6 and α2 = 0.2. The values of the first weight coefficient and the second weight coefficient can be determined through historical experience. For example, if the browsing depth is more important for reflecting the user's interest, then the value of α1 is set to be larger, or it can be obtained through a deep learning model. It can be understood that historical data can be collected, including the browsing depth, browsing frequency, and target variable (such as the number of likes of health recommendation messages), and a regression model is used to fit the data to obtain the regression coefficients of the browsing depth and browsing frequency, and the regression coefficients are normalized to be recorded as the corresponding weight coefficients.

[0083] For a single health recommendation message, its corresponding browsing frequency is the number of times it is browsed within the most recent single monitoring period, and the duration of the monitoring period is 1 week.

[0084] Specifically, the data screening unit is provided with a browsing depth detection strategy. Among them, if the graphic and text bias coefficient of the health recommendation message is greater than the preset graphic and text bias coefficient, the data screening unit determines the invalid images according to the image relevance, and determines the browsing depth according to the reading duration of the invalid images;

[0085] If the graphic and text bias coefficient of the health recommendation message is less than or equal to the preset graphic and text bias coefficient, the data screening unit determines the browsing depth according to the browsing stay duration.

[0086] Specifically, the graphic and text bias coefficient is determined according to the image distribution coefficient and the image distribution ratio;

[0087] The graphic and text bias coefficient has a negative correlation with the image distribution coefficient;

[0088] The graphic and text bias coefficient has a positive correlation with the image distribution ratio.

[0089] Graphic and text bias coefficient = (image distribution ratio / preset image distribution ratio) - (image distribution coefficient / preset image distribution coefficient). The value of the preset graphic and text bias coefficient is the average of the graphic and text bias coefficients of historical records that meet user requirements. The confirmation method of the image distribution ratio is to detect the image area and the content area in the health recommendation information. Image distribution ratio = image area / content area. The content area is the area of the smallest rectangle that can include the text and images in the health recommendation information. The confirmation method of the image distribution coefficient is to detect the average value of the number of words between adjacent images in each reading order, which is denoted as the image distribution coefficient. In the specific implementation of the present invention, the preset image distribution ratio is 40% of the content area, and the preset image distribution coefficient is 25% of the number of words.

[0090] For any image in the health recommendation information, the confirmation method of its image relevance is as follows: Use a pre-trained convolutional neural network to extract image features, map the image features to a fixed-length vector (such as 2048 dimensions), use a pre-trained natural language processing model to extract text features, map the text features to a fixed-length vector (such as 768 dimensions), and calculate the cosine similarity between the image feature vector and the text feature vector, which is denoted as the image relevance. It can be understood that the value range of the cosine similarity is [-1, 1]. Images with an image relevance less than 0 are denoted as invalid images. When determining the browsing depth according to the reading duration of invalid images, browsing depth = total reading duration - reading duration of invalid images. The total reading duration is the reading duration of the user for the health recommendation information where the invalid image is located, and the reading duration of the invalid image is the duration when the corresponding invalid image is fully displayed within the display interface; when determining the browsing depth according to the browsing stay duration, browsing depth = browsing stay duration. For a single health recommendation information, its corresponding browsing stay duration is the duration used by the user to browse this health recommendation information.

[0091] Specifically, when the data screening unit performs screening according to the user preference value, it detects the information architecture difference degree of each health recommendation information to determine the user preference value;

[0092] Select the recommended number of health recommendation information in descending order of the user preference value, which is denoted as the recommendable information.

[0093] Among them, historical health recommendation information reflecting user preferences is statistically analyzed. The historical health recommendation information is stored in a database. It can be understood that the present invention determines user preferences by recording whether the user gives a like to the historical health recommendation information. If the user records a like behavior for a piece of historical health recommendation information, then this piece of historical health recommendation information is taken as the historical health recommendation information reflecting user preferences. The number of images and the total number of words of each piece of historical health recommendation information are detected, and the average value of the number of images is calculated and recorded as the reference number of images, and the average value of the total number of words is recorded as the reference number of words. For a single health recommendation information, its information architecture difference degree = (the number of images of this health recommendation information - the reference number of images) + (the total number of words of this health recommendation information - the reference number of words). There is a negative correlation between the user preference value and the information architecture difference degree. In the specific implementation of the present invention, the user preference value = 1 / the information architecture difference degree. If the information architecture difference degree is 0, there is no need to calculate the user preference value and the corresponding health recommendation information is recorded as recommendable information.

[0094] Specifically, the chronic disease health management system further includes a database for storing health recommendation information.

[0095] The health recommendation information is information including user videos, behavior data, physical index data, and health suggestions uploaded by patients or doctors. It should be noted that the health recommendation information can be successfully uploaded only after the consent of the patient and the review of the doctor.

[0096] When searching for health information based on the patient's chronic disease data, health recommendation information with the absolute value of the difference between the physical index data in the database and the corresponding physical index data of the user being less than a first preset difference is extracted and sent to the data optimization unit;

[0097] When searching for health information based on behavior data and the effective reference image value, health recommendation information with the absolute value of the difference between the effective reference image value in the database and the corresponding effective reference image value of the user being less than a second preset difference and the absolute value of the difference between the behavior data and the corresponding behavior data of the user being less than a third preset difference is extracted and sent to the data optimization unit; for the health reference information of a single user, its corresponding effective reference image value is the average value of H0 of each standard frame in the corresponding user video. The values of the first preset difference, the second preset difference, and the third preset difference can be set by the user according to the actual scenario. It can be understood that the greater the user's requirement for the information matching accuracy, the smaller the values of the first preset difference and the second preset difference.

[0098] The patient's chronic disease data includes but is not limited to several physical index data, such as blood pressure, left ventricular ejection fraction, serum creatinine, blood sugar, uric acid, and body weight;

[0099] The behavior data includes the exercise time and exercise heart rate of each exercise.

[0100] So far, the technical solution of the present invention has been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0101] The above are only the preferred embodiments of the present invention and are not used to limit the present invention; for those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A chronic disease health management system based on health behavior change, characterized in that, Including: A user operation unit for uploading the patient's chronic disease data, behavioral data, and user videos of the user; A data analysis unit connected to the user operation unit, used to determine the health information confirmation method as searching for health information based on the patient's chronic disease data or searching for health information based on the behavioral data and the effective reference image value according to the performance value of the physical sign data and the dietary image level value; A data optimization unit connected to the user operation unit and the data analysis unit, used to respond to a preset screening condition to determine the information screening method as screening according to the browsing effective reference value or screening according to the user preference value based on the content deviation degree of the health recommendation information; A data screening unit connected to the data optimization unit and the user operation unit, used to execute the information screening method. When screening according to the browsing effective reference value, it detects the browsing depth and browsing frequency of each health recommendation information to determine the browsing effective reference value. When screening according to the user preference value, it detects the information architecture difference degree of each health recommendation information to determine the user preference value.

2. The chronic disease health management system based on health behavior change according to claim 1, wherein The data analysis unit determines the health information confirmation method according to the performance value of the physical sign data and the dietary image level value; If the performance value of the physical sign data is less than or equal to the preset performance value of the physical sign data and the dietary image level value is greater than the preset dietary image level value, the health information confirmation method is to search for health information according to the patient's chronic disease data; If the performance value of the physical sign data is greater than the preset performance value of the physical sign data or the dietary image level value is less than or equal to the preset dietary image level value, the health information confirmation method is to search for health information according to the behavioral data and the effective reference image value.

3. The chronic disease health management system based on health behavior change according to claim 2, characterized in that, The detection process of the data analysis unit for the dietary image level value includes: Detecting the standard frame similarity and block diversity of the user video; Determining whether to perform frame compensation according to the standard frame similarity and block diversity; Under the preset execution condition, determining the dietary image level value according to the color distribution degree and color richness of the frame to be analyzed; The preset execution condition is that the standard frame similarity is less than or equal to the preset standard frame similarity and the block diversity is greater than the preset block diversity or the frame compensation is completed.

4. The chronic disease health management system based on health behavior change according to claim 3, wherein, The data analysis unit responds to the preset compensation condition and performs frame compensation for the standard frame, including: Detecting the dynamic change value of the standard frame, marking the standard frame with a dynamic change value greater than the preset dynamic change value as the compensation reference frame, and extracting video frames for the video interval between the compensation reference frames. The number of video frames extracted from the video interval is determined according to the average dynamic change value; The preset compensation condition is that the standard frame similarity is greater than the preset standard frame similarity and the block diversity is less than or equal to the preset block diversity.

5. The chronic disease health management system based on health behavior change according to claim 4, characterized in that, The data optimization unit responds to the preset screening condition and determines the information screening method according to the content deviation degree of the health recommendation information; If the content deviation degree is less than the preset content deviation degree, the information screening method is to screen according to the browsing effective reference value; If the content deviation degree is greater than or equal to the preset content deviation degree, the information screening method is to screen according to the user preference value; The preset screening condition is that the number of health recommendation information obtained by searching is greater than the preset number of health recommendation information.

6. The chronic disease health management system based on health behavior change according to claim 5, characterized in that, When the data screening unit performs screening according to the browsing effective reference value, it detects the browsing depth and browsing frequency of each health recommendation information to determine the browsing effective reference value; Select the health recommendation information with the recommended quantity in the order from largest to smallest of the browsing effective reference value, and record it as the recommendable information; The browsing effective reference value has a positive correlation with both the browsing depth and the browsing frequency.

7. The chronic disease health management system based on health behavior change according to claim 6, characterized in that, The data screening unit is provided with a browsing depth detection strategy. Among them, if the graphic bias coefficient of the health recommendation information is greater than the preset graphic bias coefficient, the data screening unit determines the invalid images according to the image relevance, and determines the browsing depth according to the reading duration of the invalid images; If the graphic bias coefficient of the health recommendation information is less than or equal to the preset graphic bias coefficient, the data screening unit determines the browsing depth according to the browsing stay duration.

8. The chronic disease health management system based on health behavior change according to claim 7, characterized in that, The graphic bias coefficient is determined according to the image distribution coefficient and the image distribution ratio; The graphic bias coefficient has a negative correlation with the image distribution coefficient; The graphic bias coefficient has a positive correlation with the image distribution ratio.

9. The chronic disease health management system based on health behavior change according to claim 8, wherein, When the data screening unit performs screening according to the user preference value, it detects the information architecture difference degree of each health recommendation information to determine the user preference value; Select the health recommendation information with the recommended quantity in the order from largest to smallest of the user preference value, and record it as the recommendable information.

10. The chronic disease health management system based on health behavior change according to claim 9, characterized in that, It further includes a database for storing health recommendation information.

Images