Machine learning-based combined management equipment for various chronic diseases

Abstract

The invention provides a multi-chronic-disease joint management device based on machine learning. The apparatus includes: a data input unit configured to input personal information of a user and data related to a plurality of chronic diseases; the risk level determination unit is configured to input the personal information of the user and the data related to the various chronic diseases into a risk level model to obtain a risk level corresponding to the user in a plurality of risk levels of the various chronic diseases, and the risk level model is a machine learning model; the management unit is configured to monitor data related to the various chronic diseases and determine personalized control targets suitable for the user for the various chronic diseases according to the personal information of the user and the risk level of the user; and a medication determination unit configured to input the personal information of the user and the data related to the plurality of chronic diseases into a medication model, and determine a personalized medication scheme suitable for the user for the plurality of chronic diseases.

Description

[0001] This application is a divisional application of a patent application with an application date of March 29, 2021, an application number of 202110334308.X, and an invention title of "Device for Joint Management of Multiple Chronic Diseases and Computer-readable Storage Medium". Technical Field

[0002] This application relates to a device for joint management of multiple chronic diseases and a computer-readable storage medium. Background Art

[0003] Currently, the latest epidemiological survey data shows that the prevalence of diabetes is 11.2%, the incidence of hypertension is 27.9%, and the overall prevalence of dyslipidemia is as high as 40.4%. In addition, the latest epidemiological survey data shows that 72% of diabetic patients in China also suffer from hypertension or dyslipidemia or both. The co-occurrence and synergistic effect of the three highs (hypertension, hyperglycemia, hyperlipidemia / dyslipidemia) not only lead to cardiovascular and cerebrovascular events, but also impose a huge economic burden on patients.

[0004] Existing evidence shows that early screening, early diagnosis, and treatment of early risk factors or early diseases can delay or prevent cardiovascular and cerebrovascular events. Early comprehensive intervention for metabolic abnormalities such as hypertension, hyperglycemia, and dyslipidemia helps prevent the occurrence of diseases and even reverse the progression of diseases.

[0005] From the guiding suggestions of various medical guidelines, the three chronic diseases of diabetes, hypertension, and dyslipidemia all require lifestyle interventions to establish a healthy lifestyle so as to achieve the purpose of controlling the progression of diseases and delaying disease complications. However, there is currently no effective way to manage these three chronic diseases as an organic whole.

[0006] Therefore, it is desirable to provide a device for joint management of multiple chronic diseases that can manage multiple chronic diseases including hypertension, hyperglycemia, and hyperlipidemia / dyslipidemia as a whole. Summary of the Invention

[0007] According to an embodiment of the present application, there is provided a device for joint management of multiple chronic diseases based on machine learning, including:

[0008] A data input unit configured to input personal information of a user and data related to multiple chronic diseases;

[0009] A risk level determination unit configured to input the personal information of the user and the data related to multiple chronic diseases into a risk level model to obtain a risk level corresponding to the user among multiple risk levels of multiple chronic diseases, where the risk level model is a machine learning model; and

[0010] The management unit is configured to monitor data related to the multiple chronic diseases and determine personalized control targets for the multiple chronic diseases suitable for the user according to the user's personal information and the user's risk level.

[0011] In some examples, the multiple chronic diseases include hypertension, diabetes, and dyslipidemia.

[0012] In some examples, the user's personal information includes one or more of the following: age, gender, body mass index (BMI), long-term place of residence, eating habits and preferences, underlying diseases, allergy history, occupation, daily activity level, exercise hobbies; and

[0013] Data related to the multiple chronic diseases includes one or more of the following: user physical sign data including detection data of hypertension, diabetes, and dyslipidemia, medical history data of hypertension, diabetes, and dyslipidemia, cardiovascular risk factors, medical history data of other diseases, patient medication situation data, drug instruction manuals, drug indications, drug contraindications, drug incompatibility taboos, drug side effects, drug inventory, drug prices.

[0014] In some examples, the risk level model is obtained by model training using a user data set and a medical guideline data set, where the user data set includes personal information of multiple users, data related to multiple chronic diseases, and data on risk levels, and the medical guideline data set includes medical diagnosis and treatment guidelines, expert consensus, medical journals, medical textbooks, medical dictionaries, doctor diagnosis data in drug instruction manuals, and patient medication situation data related to multiple diseases.

[0015] In some examples, the risk level model uses one or more of the input user's personal information and data related to multiple chronic diseases as risk factors and dynamically evaluates the user's risk level according to the number of risk factors.

[0016] In some examples, the risk level model evaluates the user's risk level as a high level in response to an increase in the number of risk factors and evaluates the user's risk level as a low level in response to a decrease in the number of risk factors.

[0017] In some examples, the control targets include control targets for multiple physical sign parameters related to one or more of the user's blood glucose, blood pressure, and blood lipids, and

[0018] The management unit is also configured to set the values of the control targets for multiple physical sign parameters of users with a high risk level lower than the values of the control targets for multiple physical sign parameters of users with a low risk level according to the user's risk level; and

[0019] The management unit also sets a risk threshold for each of multiple chronic diseases according to the user's risk level.

[0020] In some examples, the management device further includes:

[0021] A reporting unit configured to output a monitoring report for the user's multiple chronic diseases according to the control objective.

[0022] In some examples, the reporting unit outputs the monitoring report at a predetermined frequency or at a preset time.

[0023] In some examples, the reporting unit outputs a risk warning signal in response to the monitoring result indicating that one or more of the multiple chronic diseases exceed the risk threshold.

[0024] In some examples, the risk level determination unit is further configured to use one or more of the input user's personal information and data related to multiple chronic diseases as risk factors to evaluate and / or predict the user's cardiovascular disease risk level.

[0025] In some examples, the management device further includes:

[0026] A medication determination unit configured to input the user's personal information and data related to multiple chronic diseases into a medication model to determine a personalized medication plan for the user for multiple chronic diseases, and the medication model is a machine learning model.

[0027] In some examples, the medication model is obtained by training the model using a user dataset and a medical guideline dataset, where the user dataset includes the personal information of multiple users, data related to multiple chronic diseases, and data on medication records, and the medical guideline dataset includes medical guidelines, expert consensus, medical journals, medical textbooks, medical dictionaries, drug instructions, doctor diagnosis data, and patient medication situation data related to multiple diseases.

[0028] In some examples, the medication model uses one or more of the input user's personal information and data related to multiple chronic diseases as recommendation factors and dynamically adjusts the user's medication plan according to the used recommendation factors.

[0029] In some examples, the medication model analyzes the side effects of a predetermined drug for one of multiple chronic diseases on other chronic diseases and adjusts the user's medication plan to recommend drugs that have no side effects on other chronic diseases.

[0030] In some examples, the medication model analyzes the drug prices of multiple predetermined drugs for one or more of multiple chronic diseases and the income information of the user, and adjusts the user's medication plan to recommend predetermined drugs that match the user's income information.

[0031] In some examples, the management device further includes:

[0032] A reporting unit configured to output reminder information for reminding the user to take medicine according to the medication plan.

[0033] In some examples, the reporting unit outputs reminder information at a predetermined frequency or at a preset time.

[0034] In some examples, the input unit further receives problem data or keyword data input by the user,

[0035] The reporting unit outputs a corresponding answer according to the input problem data or keyword data, using the preset answers, the outputs determined by the risk level determination unit, or the outputs determined by the medication determination unit.

[0036] In some examples, the management device further includes:

[0037] An education unit configured to input the personal information of the user and data related to multiple chronic diseases into an education model, and determine a personalized education plan for multiple chronic diseases suitable for the user, where the education model is a machine learning model.

[0038] In some examples, the education model is obtained by training the model using a user dataset and a medical media dataset, where the user dataset includes the personal information of multiple users, data related to multiple chronic diseases, disease course data, disease control conditions, and data on reading habits, and the medical media dataset includes patient teaching content related to multiple diseases, which includes medical videos, medical audios, medical animations, medical comics, and medical popular science articles.

[0039] In some examples, the education model uses one or more of the personal information of the input user and data related to multiple chronic diseases as recommendation factors, and dynamically adjusts the user's education plan according to the used recommendation factors.

[0040] In some examples, the education model analyzes the user's underlying diseases, patient medication conditions, allergy history, age, and disease course data, and outputs positive treatment education content and psychological counseling content for multiple chronic diseases.

[0041] In some examples, the medication model analyzes the user's underlying diseases, patient medication conditions, allergy history, age, and the user's reading habits, and outputs patient teaching content suitable for the user.

[0042] In some examples, the management device further includes:

[0043] An including unit, configured to respond to problem data or keyword data input by the user through the input unit, and output a corresponding answer by using preset answers, outputs determined by a risk level determination unit, outputs determined by a medication determination unit, and / or outputs determined by an education unit.

[0044] In some examples, the management device further includes:

[0045] A diet behavior intervention unit, configured to input the user's personal information and data related to multiple chronic diseases into a diet model, and determine a personalized diet plan for multiple chronic diseases suitable for the user, where the diet model is a machine learning model.

[0046] In some examples, the diet model is obtained by training the model using a user data set and a medical diagnosis and treatment guideline data set, where the user data set includes personal information of multiple users, data related to multiple chronic diseases, data on daily activity levels, and data on seasons, and the medical diagnosis and treatment guideline data set includes nutrition and dietary guidelines related to multiple diseases.

[0047] In some examples, the diet model uses one or more of the input user's personal information and data related to multiple chronic diseases as recommendation factors, and dynamically adjusts the user's diet behavior plan according to the used recommendation factors.

[0048] In some examples, the diet behavior plan includes a recipe recommendation plan, and

[0049] The diet model analyzes the user's gender, age, weight, long-term place of residence, eating habits and preferences, medical history, season, daily activity level data, medication data, and nutrition and dietary guidelines related to multiple diseases, and outputs a recipe suitable for the user.

[0050] In some examples, the input unit further obtains an image of the user's food, and the diet model identifies the food and estimates the calorie value of the food.

[0051] In some examples, the diet behavior plan includes production guidelines for one or more specific diets suitable for multiple chronic diseases and / or guiding information for obtaining a specific diet.

[0052] In some examples, the management device further includes:

[0053] A reporting unit, configured to output information for suggesting the intake amount for the user, reminder information for warning against taboo foods, reminder information for warning against excessive intake, reminder information for suggesting the amount of exercise that needs to be increased, and statistical information on the intake amount within a predetermined time period, according to the determined diet behavior plan.

[0054] In some examples, the management device further includes:

[0055] An exercise behavior intervention unit, configured to input the personal information of the user and data related to multiple chronic diseases into an exercise model, and determine a personalized exercise plan for multiple chronic diseases suitable for the user, where the exercise model is a machine learning model.

[0056] In some examples, the exercise model is obtained by training the model using a user dataset and a medical treatment guideline dataset, where the user dataset includes the personal information of multiple users, data related to multiple chronic diseases, data on exercise hobbies, exercise data for the current day and the recent period, and data on seasons, and the medical treatment guideline dataset includes exercise guidelines related to multiple diseases.

[0057] In some examples, the exercise model uses one or more of the input personal information of the user and data related to multiple chronic diseases as recommendation factors, and dynamically adjusts the user's exercise plan according to the used recommendation factors.

[0058] In some examples, the exercise model analyzes the user's BMI, gender, age, underlying diseases, occupation, exercise hobbies, and exercise guidelines related to multiple diseases, and outputs an exercise plan suitable for the user.

[0059] In some examples, the exercise model also inputs the user's food intake, recent patient monitoring data, and drug intake to dynamically adjust the user's exercise plan.

[0060] In some examples, the management device further includes:

[0061] A reporting unit, configured to output information indicating that the exercise amount meets the standard, information indicating the energy consumed within a predetermined time, information indicating the change in body fat or body weight or BMI, reminder information for warning against inappropriate exercise types, and information for obtaining external exercise resources, according to the determined exercise plan.

[0062] In another embodiment, a multiple chronic disease joint management device based on machine learning is provided, including:

[0063] A processor; and

[0064] A memory, configured to store program instructions executable by the processor, and when the program instructions are executed by the processor, the management device is caused to execute:

[0065] Input the personal information of the user and data related to multiple chronic diseases;

[0066] Input the personal information of the user and data related to multiple chronic diseases into a risk level model to obtain a risk level corresponding to the user among multiple risk levels of multiple chronic diseases, where the risk level model is a machine learning model; and

[0067] Monitor the data related to the multiple chronic diseases, and determine personalized control objectives for the multiple chronic diseases suitable for the user according to the personal information of the user and the risk level of the user.

[0068] In another embodiment, a computer-readable storage medium is provided, storing program instructions executable by a processor, and when the program instructions are executed by the processor, perform:

[0069] Input the personal information of the user and data related to multiple chronic diseases;

[0070] Input the personal information of the user and data related to multiple chronic diseases into a risk level model to obtain a risk level corresponding to the user among multiple risk levels of multiple chronic diseases, where the risk level model is a machine learning model; and

[0071] Monitor the data related to the multiple chronic diseases, and determine personalized control objectives for the multiple chronic diseases suitable for the user according to the personal information of the user and the risk level of the user.

[0072] Therefore, the multiple chronic diseases joint management device and the computer-readable storage medium according to the embodiments of the present application can manage multiple chronic diseases including hypertension, hyperglycemia, and hyperlipidemia / dyslipidemia as a whole. In addition, the multiple chronic diseases joint management device according to the embodiments of the present application can also give personalized and accurate recommendations for the patient's energy intake, conveniently provide the user with food / drug incompatibilities, and provide diverse presentation methods for various monitoring data, diet behavior plans, education plans, and exercise behavior plans, facilitating the user to conveniently understand their own condition and increasing the user's usage stickiness. Brief Description of the Drawings

[0073] Figure 1 It is a functional configuration block diagram of a multiple chronic diseases joint management device according to the first embodiment of the present application;

[0074] Figure 2 It is a training flowchart of a risk level model of a management device according to the first embodiment of the present application;

[0075] Figure 3 It is a functional configuration block diagram of a combined management device for multiple chronic diseases according to the second embodiment of the present application;

[0076] Figure 4 It is a training flow chart of the medication model of the management device according to the second embodiment of the present application;

[0077] Figure 5 It is a functional configuration block diagram of a combined management device for multiple chronic diseases according to the third embodiment of the present application;

[0078] Figure 6 It is a training flow chart of the education model of the management device according to the third embodiment of the present application;

[0079] Figure 7 It is a functional configuration block diagram of a combined management device for multiple chronic diseases according to the fourth embodiment of the present application;

[0080] Figure 8 It is a training flow chart of the diet model of the management device according to the fourth embodiment of the present application;

[0081] Figure 9 It is a functional configuration block diagram of a combined management device for multiple chronic diseases according to the fifth embodiment of the present application;

[0082] Figure 10 It is a training flow chart of the exercise model of the management device according to the fifth embodiment of the present application;

[0083] Figure 11 It is a functional configuration block diagram of a combined management device for multiple chronic diseases according to the sixth embodiment of the present application; and

[0084] Figure 12 It is a functional configuration block diagram of a combined management device for multiple chronic diseases according to the seventh embodiment of the present application. Detailed implementation manners

[0085] Next, a combined management device for multiple chronic diseases according to an embodiment of the present application will be described in detail with reference to the accompanying drawings.

[0086] <The first embodiment>

[0087] Next, a combined management device for multiple chronic diseases based on machine learning according to the first embodiment of the present application will be described in detail in combination with Figure 1-2 With the development of powerful data storage, computing platforms, and mobile Internet, as well as the large-scale explosion and rapid electronic digitization of medical data, machine learning technology has been widely applied in the medical field.

[0088] ​

[0089] Figure 1 It is a functional configuration block diagram of a combined management device for multiple chronic diseases according to the first embodiment of the present application. The combined management device for multiple chronic diseases can be an electronic device such as a smartphone, a laptop computer, a desktop computer, a server device, etc.

[0090] As Figure 1 shown, the management device 100 according to the first embodiment of the present application includes: a data input unit 101, a risk level determination unit 102, a management unit 103, and a reporting unit 104.

[0091] The data input unit 101 is used to input the personal information of the user and data related to multiple chronic diseases. Specifically, the data input unit 101 may include a camera for taking pictures of the user's medical records to input personal information, detection data and medical history data of multiple chronic diseases, etc. The data input unit 101 may also include a touch screen and / or a keyboard and / or a mouse, and can input personal information, detection data and medical history data of multiple chronic diseases, etc. The data input unit 101 may also include a microphone, and can input personal information, detection data and medical history data of multiple chronic diseases, etc. The data input unit 101 may also include a communication unit for connecting to the hospital's database, and can input personal information, detection data and medical history data of multiple chronic diseases, etc. The data input unit 101 may include a wearable device for obtaining the personal information of the user, detection data and medical history data of multiple chronic diseases, etc.

[0092] For example, the personal information of the user includes age, gender, body mass index (BMI), long-term place of residence, eating habits and preferences, underlying diseases, allergy history, occupation, daily activity level, sports hobbies, and so on.

[0093] In addition, the data related to multiple chronic diseases includes, for example, one or more of the following: user physical sign data including detection data of hypertension, diabetes, and dyslipidemia, medical history data of hypertension, diabetes, and dyslipidemia, cardiovascular risk factors, medical history data of other diseases, patient medication situation data, drug instructions, drug indications, drug contraindications, drug incompatibility taboos, drug side effects, drug inventory, drug prices.

[0094] In the present application, the three high diseases (hypertension, diabetes, and dyslipidemia) are mainly used as multiple chronic diseases for description.

[0095] The risk level determination unit 102 is used to input the personal information of the user and the data related to multiple chronic diseases into a risk level model, and obtain a risk level corresponding to the user among multiple risk levels of multiple chronic diseases. In this embodiment, the risk level model is a machine learning-based model. It will be referred to later Figure 2Describe the training process of the risk level model.

[0096] The management unit 103 is used to monitor the data related to the multiple chronic diseases, and determine the personalized control objectives for the multiple chronic diseases suitable for the user according to the personal information of the user and the risk level of the user determined by the risk level determination unit 102. The control objectives include, for example, the control objectives of the user's blood sugar, blood pressure, and blood lipid.

[0097] The reporting unit 104 is used to output a monitoring report for the multiple chronic diseases of the user according to the control objectives determined by the management unit 103.

[0098] Next, the training process of the risk level model will be described with reference to Figure 2 Describe the training process of the risk level model.

[0099] Step S201: Obtain the user dataset and the medical guideline dataset.

[0100] The user dataset includes, for example, the personal information of multiple users, the data related to multiple chronic diseases, and the data of risk levels.

[0101] The medical guideline dataset includes, for example, medical treatment guidelines related to multiple diseases, expert consensus, medical journals, medical textbooks, medical dictionaries, drug instructions, doctor diagnosis data, and patient medication situation data.

[0102] Step S202: Use the user dataset and the medical guideline dataset as training data to input into a machine learning model for training, and obtain a risk level model that stratifies users.

[0103] The machine learning model can utilize various technologies, such as technologies based on the Hadoop big data structure, machine learning technologies based on machine learning algorithm models, and deep learning technologies based on neural networks.

[0104] The core of the Hadoop platform includes a distributed storage and computing framework for the storage backup and computing requirements of massive data. The Hadoop platform uses the distributed file system HDFS for data backup and storage. An HDFS cluster consists of a NameNode and several DataNodes. The NameNode, as the main server, manages the namespace of the data file system and the access operations of PC or mobile users to the data. The DataNodes in the cluster manage the data transmitted from user terminals for storage. Data files are divided into several data blocks and stored on the DataNodes. The NameNode performs namespace operations on the file system, such as opening, closing, renaming files or directories, etc., and is responsible for mapping data blocks to specific DataNodes. The DataNode is responsible for processing data read and write requests from the file system client and creating, deleting, and replicating data blocks under the unified scheduling of the NameNode.

[0105] The computing framework of the Hadoop platform adopts the MapReduce data processing model. The working process of MapReduce is divided into two stages: the Map stage and the Reduce stage. Each stage has Key / value pairs as input and output, and their types can be artificially selected. First, the first Map function divides different data stored in HDFS into different data clusters according to a predetermined category, so that the data in a single cluster all belong to the same category, and the value of this category is used as the correct value in the calibration of this cluster model; the second Map function combines the data in each cluster with the predetermined data; and in the Reduce stage, the output of the second Map function is directly used as input, and the calibration model algorithm is called, and the result obtained is the calibration model parameters for each user terminal.

[0106] On the other hand, algorithm models for machine learning can use, for example, the LinearRegression algorithm, the Support Vector Machine (SVM) algorithm, the K-Nearest Neighbors (KNN) algorithm, the Logistic Regression algorithm, the Decision Tree algorithm, the K-Means algorithm, the Random Forest algorithm, the Naive Bayes algorithm, the Dimensional Reduction algorithm, and the GradientBoosting algorithm.

[0107] The machine learning model extracts parameters such as age, gender, body mass index (BMI) from the user's personal information as labels, and extracts multiple diagnostic criteria for various chronic diseases specified in medical guidelines, as well as multiple diagnostic criteria for various chronic diseases involved in medical journals, expert consensus, medical textbooks, medical dictionaries, drug instructions, doctor's diagnostic data, and patient medication data. It matches according to the number of labels and multiple diagnostic criteria to determine the user's risk level.

[0108] By using the data of a large number of users to train the risk level model, the final risk level model is obtained. In addition, it is also possible to automatically obtain updated data of medical treatment guidelines, expert consensus, medical journals, medical textbooks, medical dictionaries, drug instructions, doctor's diagnostic data, and patient medication data, so as to retrain the risk level model to improve the accuracy of the risk level model.

[0109] In one example, in step S201: By inputting the user data set, including the user's personal age, gender, body mass index (BMI), native place, underlying diseases, allergy history, occupation, daily activity level, exercise hobbies, eating habits, prescription data, etc., analyze the potential data associations or relationships existing with the phased test data. The test data includes the numerical fluctuations of hypertension, diabetes, and dyslipidemia.

[0110] The data association model can be divided into multiple types such as two-dimensional, multi-dimensional, 1:1, 1:N, and even N:N. Its association functions include linear and non-linear. For non-linear relationships, the data relationship can be directly displayed with an image.

[0111] For example, through data analysis, it is found that among diabetic patients, in the group of people who like to eat porridge or rice rolls, most of the fasting blood glucose values are greater than 10 mmol / L. "Porridge" and "rice rolls" have an approximately proportional relationship with "blood glucose value".

[0112] In step S202: Select the "Dependency" input and output data from step S201 and sequentially try the following algorithm models.

[0113] 1) Linear regression algorithm;

[0114] 2) Support vector machine algorithm;

[0115] 3) K-nearest neighbor algorithm;

[0116] 4) Logistic regression algorithm;

[0117] 5) Decision tree algorithm;

[0118] 6) Naive Bayes algorithm;

[0119] Compare the correct prediction probabilities of the above algorithm results and select one or more most suitable algorithm models.

[0120] For example, using other noise data such as "congee" and "rice noodle rolls" as input parameters, the output results "blood glucose values" of two algorithm models are finally fitted to be greater than 10 mmol / L. The two algorithm models are the "support vector machine algorithm" and the "K-nearest neighbor algorithm".

[0121] In step S202, "black box" data is also used for prediction to improve model iteration, and finally a multi-algorithm model is used to achieve machine learning.

[0122] For example, using foods similar to "congee" and "rice noodle rolls" (such as calories, sugar content, etc.) as new input parameters, allowing the system to calculate blood glucose values by itself, and for the same type of situation, the system provides reasonable diet and other suggestions.

[0123] The above only shows an example of machine learning. In this application, the machine learning model can use one or more of the input personal information of the user and data related to multiple chronic diseases as risk factors to dynamically evaluate the risk level of the user.

[0124] For example, the risk level model can use age, gender, weight, whether there are other combined diseases, etc. as risk factors (i.e., labels), and dynamically evaluate the risk level of the user according to the number of labels.

[0125] The risk level model evaluates the user's risk level as a high level in response to an increase in the number of labels, and evaluates the user's risk level as a low level in response to a decrease in the number of labels.

[0126] For example, the control objectives include control objectives for multiple physical sign parameters related to one or more of the user's blood glucose, blood pressure, and blood lipids.

[0127] The management unit 103 sets the values of the control objectives for multiple physical sign parameters of users with a high risk level to be lower than the values of the control objectives for multiple physical sign parameters of users with a low risk level according to the user's risk level.

[0128] The management unit 103 can also set a danger threshold for each of multiple chronic diseases according to the user's risk level.

[0129] In addition, the reporting unit 104 can output a monitoring report for multiple chronic diseases of the user according to the control objectives determined by the management unit 103.

[0130] For example, the reporting unit 104 can output a monitoring report at a predetermined frequency or at a preset time.

[0131] The reporting unit 104 can also combine control objectives to present a system monitoring report. The report can be presented, for example, in the form of a line chart, a dot chart, a table, or a voice broadcast. The user can automatically switch the presentation form of the report according to personal preferences.

[0132] The report can include, for example, a blood pressure line chart, the number of times blood glucose reaches the standard / the number of measurements, Time in Range (TIR), predicted glycated hemoglobin, blood lipids (including total cholesterol, low-density lipoprotein, high-density lipoprotein, triglycerides).

[0133] In response to the monitoring results indicating that one or more of multiple chronic diseases exceed the risk threshold, the reporting unit 104 outputs a risk warning signal. For example, the reporting unit 104 can send a risk warning signal to the user's relatives and friends or doctor.

[0134] In addition, the risk level determination unit 102 can also use one or more of the input personal information of the user and the data related to multiple chronic diseases as risk factors to evaluate and / or predict the user's cardiovascular disease risk level. The reporting unit 104 can also provide a warning signal about the risk of having cardiovascular disease or other diseases.

[0135] Therefore, the multiple chronic disease joint management device according to the first embodiment of the present application can manage multiple chronic diseases including hypertension, hyperglycemia, and hyperlipidemia / dyslipidemia as a whole in a unified manner. It can provide a variety of presentation methods for various monitoring data, which is convenient for users to understand their own conditions and increases the user's stickiness.

[0136] <Second Embodiment>

[0137] The following will be combined with Figure 3-4 Describe in detail the multiple chronic disease joint management device based on machine learning according to the second embodiment of the present application.

[0138] The difference between the multiple chronic disease joint management device 300 according to the second embodiment and the multiple chronic disease joint management device 100 according to the first embodiment is that a medication determination unit 305 is additionally added. For the sake of brevity of description, the repeated description of the same unit components as those of the multiple chronic disease joint management device 100 is omitted here.

[0139] The medication determination unit 305 inputs the personal information of the user and the data related to multiple chronic diseases into a medication model to determine a personalized medication plan for the user for multiple chronic diseases, and the medication model is a machine learning model.

[0140] In the combined management of multiple chronic diseases, the medication regimen is different from that for a single chronic disease. Existing medical guidelines usually only give recommendations on medication regimens for a single disease, without considering the side effects of certain drugs applicable to one disease on other diseases when multiple diseases coexist.

[0141] The medication model used by the medication determination unit 305 takes into account the mutual influence between multiple chronic diseases and is implemented through a machine learning model.

[0142] Next, the training process of the medication model will be described with reference to Figure 4 Describe the training process of the medication model.

[0143] Step S401: Obtain a user data set and a medical guideline data set.

[0144] The user data set includes, for example, personal information of multiple users, data related to multiple chronic diseases, and data on medication records.

[0145] The medical guideline data set includes, for example, medical treatment guidelines related to multiple diseases, medical journals, medical textbooks, medical dictionaries, drug instruction manuals, doctor diagnosis data, and patient medication situation data.

[0146] Step S402: Input the user data set and the medical guideline data set as training data into the machine learning model for training to obtain a medication model for recommended regimens for different drugs.

[0147] The machine learning model extracts parameters such as age, underlying diseases, and allergy history in the user's personal information as labels, extracts multiple medication recommendations for multiple chronic diseases specified in the medical guidelines, medication recommendations in medical journals, medical textbooks, medical dictionaries, drug instruction manuals, and multiple medication recommendations for multiple chronic diseases involved in the patient medication situation data, and matches according to the number of labels and the multiple medication recommendations to determine the user's medication recommendations.

[0148] The architecture of the machine learning model in this embodiment is basically the same as that of the machine learning model in the first embodiment, and its detailed description is omitted here.

[0149] By training the medication model using data of a large number of users, a final medication model is obtained. In addition, updated data of medical treatment guidelines, medical journals, medical textbooks, medical dictionaries, drug instruction manuals, doctor diagnosis data, and patient medication situation data can also be automatically obtained to retrain the medication model to improve the accuracy of medication recommendations.

[0150] For example, the medication model uses one or more of the input user's personal information and data related to multiple chronic diseases as recommendation factors, and dynamically adjusts the user's medication plan according to the used recommendation factors.

[0151] Specifically, the medication model can analyze the side effects of a prescribed medication for one of multiple chronic diseases on other chronic diseases, and adjust the user's medication plan to recommend a medication that has no side effects on other chronic diseases.

[0152] For example, a certain medication recommended for ordinary diabetic patients has the side effect of causing elevated blood pressure. In this case, if the user's physical sign data indicates that the user also has hypertension, the medication model does not recommend this medication to the user, but selects a medication that can lower blood sugar without causing elevated blood pressure to recommend to the user. For example, the medication model first selects a hypoglycemic medication that also has the effect of lowering blood pressure, and secondly recommends a medication that can lower blood sugar without causing elevated blood pressure.

[0153] In another example, the medication model can also analyze the drug prices of multiple prescribed medications for one or more of multiple chronic diseases and the user's income information, and adjust the user's medication plan to recommend a prescribed medication that matches the user's income information.

[0154] For example, if the medication model determines through analyzing the user's income information that the user's income is relatively low, it recommends medications with relatively low prices and relatively low effects to the user. For example, the medication model sets the priority of price higher than the treatment effect, so that when the treatment effects are similar, the medication model recommends medications with relatively low prices.

[0155] On the other hand, if the medication model determines through analyzing the user's income information that the user's income is relatively high, it recommends medications with relatively high prices and better effects to the user. For example, the medication model sets the priority of the treatment effect higher than the price, so as to recommend the medication with the best effect to the user.

[0156] The difference between the reporting unit 306 in the management device 300 according to the second embodiment and the reporting unit 104 in the first embodiment is that, in addition to the functions of the reporting unit 104, the reporting unit 306 can also output reminder information for reminding the user to take medicine according to the medication plan.

[0157] The reporting unit 306 outputs reminder information at a predetermined frequency or at a preset time. For example, the reporting unit 306 can remind the user to take medicine through an alarm clock, a text message, or a family greeting, so that the user's medication compliance can reach more than 90%.

[0158] The reporting unit 306 can also report medication records, such as insulin dose reduction curves, pharmacoeconomic bills (savings amount, quality-adjusted life years (QALY), etc.).

[0159] In addition, the reporting unit 306 can also incorporate an AI consultation function. The input unit 101 can receive problem data or keyword data input by the user, and then the reporting unit 306 outputs corresponding answers based on the input problem data or keyword data, using pre-set answers, outputs determined by the risk level determination unit, or outputs determined by the medication determination unit. For example, the user can input diabetes drug incompatibilities, and the reporting unit 306 can provide information on diabetes drug incompatibilities to the user according to the keywords input by the user.

[0160] The reporting unit 306 can also be used for doctors to view examination results online, patients' medication conditions, and dispensing medications (subsequent to offline hospital examinations), facilitating family doctors or lower-level specialists to answer inquiries about basic disease problems, etc.

[0161] The reporting unit 306 can also respond to the user's query operation, provide instruction manuals for common medications for various chronic diseases, provide medication tips (such as drug contraindications, drug incompatibilities, etc.), and so on.

[0162] Therefore, the multi-chronic disease joint management device according to the second embodiment of the present application can manage multiple chronic diseases including hypertension, hyperglycemia, and dyslipidemia as a whole. It can conveniently provide drug incompatibilities to users, provide diverse presentation methods for various monitoring data to users, facilitate users to conveniently understand their own conditions, and increase user stickiness.

[0163] <Third Embodiment>

[0164] The following will be combined with Figure 5-6 Describe in detail the machine learning-based multi-chronic disease joint management device according to the third embodiment of the present application.

[0165] The multi-chronic disease joint management device 500 according to the third embodiment is different from the multi-chronic disease joint management device 100 according to the first embodiment in that an education unit 505 is additionally added. For the sake of brevity in description, the repeated description of the same unit components as those of the multi-chronic disease joint management device 100 is omitted here.

[0166] The education unit 505 inputs the user's personal information and data related to multiple chronic diseases into an education model to determine a personalized education plan for multiple chronic diseases suitable for the user, and the education model is a machine learning model.

[0167] In the combined management of multiple chronic diseases, the education of users needs to be carried out in combination with the different characteristics of multiple chronic diseases.

[0168] The medication model used by the education unit 505 takes into account the different characteristics between multiple chronic diseases and is implemented through a machine learning model.

[0169] Next, the training process of the medication model will be described with reference to Figure 6 Describe the training process of the medication model.

[0170] Step S601: Obtain a user data set and a medical media data set.

[0171] The user data set includes, for example, personal information of multiple users, data related to multiple chronic diseases, disease course data (first diagnosis, within half a year, within one year, more than one year), disease control conditions (stable condition, poor control), and data on reading habits.

[0172] The medical media data set includes, for example, patient teaching content related to multiple diseases, which includes medical videos, medical audio, medical animations, medical comics, and medical popular science articles.

[0173] Step S602: Input the user data set and the medical media data set as training data into the machine learning model for training to obtain an education model for recommended solutions for different disease conditions.

[0174] The machine learning model extracts parameters such as underlying diseases, patient medication conditions, allergy history, age, and disease course data from the user's personal information as labels, and extracts multiple educational recommendations for multiple chronic diseases stipulated in medical guidelines, educational recommendations in medical journals and medical textbooks, and matches the number of labels with multiple educational recommendations to determine the user's educational recommendations.

[0175] The architecture of the machine learning model in this embodiment is basically the same as that of the machine learning model in the first embodiment, and its detailed description is omitted here.

[0176] By using the data of a large number of users to train the education model, the final education model is obtained. In addition, the management device can also perform data docking with domestic and foreign medical authoritative websites and large libraries, so that the education model can also automatically obtain updated data on medical treatment guidelines, expert consensus, medical journals, and medical textbooks to retrain the medication model to improve the accuracy of educational recommendations.

[0177] For example, the education model uses one or more of the input user's personal information and data related to multiple chronic diseases as recommendation factors and dynamically adjusts the user's education plan according to the used recommendation factors.

[0178] Specifically, the education model analyzes the user's underlying diseases, patient medication conditions, allergy history, age, and disease course data, and outputs positive treatment education content and psychological counseling content for multiple chronic diseases.

[0179] For example, for newly diagnosed diabetes patients, videos on psychological counseling and diabetes treatment education are recommended. For patients with both diabetes and hypertension, videos on diabetes and hypertension treatment education and more matters that should be noted are recommended.

[0180] In another example, the medication model analyzes the user's underlying diseases, patient medication conditions, allergy history, age, and the user's reading habits, and outputs patient teaching content suitable for the user.

[0181] For example, if the education model determines through analysis that the user has read a diabetes diet recipe and the content the user is interested in is how to prevent and control, the education model outputs more content on how to prevent and control diabetes to the user.

[0182] The difference between the reporting unit 506 in the management device 500 according to the second embodiment and the reporting unit 104 in the first embodiment is that, in addition to the functions of the reporting unit 104, the reporting unit 506 can also respond to question data or keyword data input by the user through the input unit, and use pre-set answers, outputs determined by the risk level determination unit, outputs determined by the medication determination unit, and / or outputs determined by the education unit to output corresponding answers.

[0183] In addition, the reporting unit 506 can also embed a consultation and interaction platform. The input unit 101 can receive question data or keyword data input by the user, and then the reporting unit 506 outputs corresponding answers according to the input question data or keyword data, using pre-set answers, outputs determined by the risk level determination unit, or outputs determined by the medication determination unit. For example, the user can input a diabetes diet recipe, and the reporting unit 506 can provide information on the diabetes diet recipe to the user according to the keyword input by the user.

[0184] Therefore, the multiple chronic disease joint management device according to the third embodiment of the present application can manage multiple chronic diseases including hypertension, hyperglycemia, and dyslipidemia as a whole. Content that helps users learn disease prevention and treatment and receive psychological counseling can be conveniently provided to users, and various monitoring data can be presented to users in a diversified manner, facilitating users to conveniently understand their own conditions and increasing user stickiness.

[0185] <Fourth Embodiment>

[0186] Next, it will be combined with Figure 7-8Describe in detail a machine learning-based combined management device for multiple chronic diseases according to the fourth embodiment of the present application.

[0187] The combined management device 700 for multiple chronic diseases according to the fourth embodiment is different from the combined management device 100 for multiple chronic diseases according to the first embodiment in that a dietary behavior intervention unit 705 is additionally added. For the sake of brevity of description, the repeated description of the same unit components as those of the combined management device 100 for multiple chronic diseases is omitted here.

[0188] The dietary behavior intervention unit 705 inputs the personal information of the user and data related to multiple chronic diseases into a dietary model, and determines a personalized dietary plan for multiple chronic diseases suitable for the user. The dietary model is a machine learning model.

[0189] In the combined management of multiple chronic diseases, dietary management is a very important link. Existing medical guidelines usually only give suggestions on dietary plans for a single disease, without considering that when multiple diseases coexist, some foods suitable for one disease are not suitable for other diseases. In addition, patients of different ages and regions have completely different requirements for diet in different seasons. For example, most elderly patients like light foods, and patients in Sichuan and Hunan like spicy foods, etc. In this case, if a personalized dietary plan can be recommended according to the user's situation and dietary management can be carried out in the dietary way the user likes, it will be more helpful for the user to establish a healthy lifestyle.

[0190] The dietary model used by the dietary behavior intervention unit 705 takes into account the dietary characteristics of different patients and is implemented through a machine learning model.

[0191] Next, the training process of the medication model will be described with reference to Figure 8 Describe the training process of the medication model.

[0192] Step S801: Obtain a user dataset and a medical treatment guideline dataset.

[0193] The user dataset includes the personal information of multiple users, data related to multiple chronic diseases, data on daily activity levels, data on current and recent exercise, and data on seasons.

[0194] The medical treatment guideline dataset includes nutrition and dietary guidelines related to multiple diseases. For example, the latest domestic and foreign treatment guidelines for three highs (such as "Dietary Guidelines for Chinese Residents", "Chinese Dietary Reference Intakes for Nutrients", "Dietary Guidance for Diabetic Patients", "Chinese Diabetes Medical Nutrition Therapy Guidelines", etc.).

[0195] Step S802: Use the user data set and the medical treatment guideline data set as training data to input into a machine learning model for training, and obtain a diet model for recommended solutions for different users.

[0196] The machine learning model extracts parameters such as gender, age, physical signs, medical history, and daily activities in the user's personal information as labels, extracts multiple diet recommendations for various chronic diseases specified in the medical treatment guidelines, and matches according to the number of labels and the multiple diet recommendations to determine the user's diet recommendation.

[0197] The architecture of the machine learning model in this embodiment is basically the same as that of the machine learning model in the first embodiment, and its detailed description is omitted here.

[0198] By using the data of a large number of users to train the diet model, the final diet model is obtained. In addition, the medical treatment guidelines can be automatically obtained to retrain the diet model to improve the accuracy of diet recommendations.

[0199] For example, the diet model uses one or more of the input user's personal information and data related to various chronic diseases as recommendation factors, and dynamically adjusts the user's diet plan according to the used recommendation factors.

[0200] Specifically, the diet behavior plan includes a recipe recommendation plan. The diet model can analyze the user's gender, age, weight, long-term place of residence, eating habits and preferences, medical history, season, daily activity level data and medication data, as well as the nutrition and dietary guidelines related to various diseases, and output recipes suitable for the user.

[0201] For example, a light recipe can be recommended for elderly ordinary diabetes patients. For young ordinary diabetes patients, it is recommended to try to quit smoking and drinking. If drinking is necessary, avoid drinking on an empty stomach and do not exceed 25g of alcohol.

[0202] In addition, the input unit 101 can also take pictures of the user's food, and the diet model identifies the food and estimates the calorie value of the food.

[0203] The diet behavior plan can include production guidelines for one or more specific diets suitable for various chronic diseases and / or guiding information for obtaining specific diets. For example, video scenes of diabetes meal production guidelines, nutrient content quick reference books for foods, diabetes meal takeaway services, etc.

[0204] The difference between the reporting unit 706 in the management device 700 according to the fourth embodiment and the reporting unit 104 in the first embodiment is that in addition to the function of the reporting unit 104, the reporting unit 706 can also output reminder information to remind the user of the diet according to the diet plan.

[0205] For example, in the case where the willingness of patients to independently upload their diet is relatively low, the reporting unit 706 can encourage users by prompting them to upload daily diet photos through clock-in and giving corresponding incentives (such as obtaining "willpower").

[0206] The reporting unit 706 can also provide personalized daily / meal intake recommendations for users. For elderly users and users in second- and third-tier cities who are not familiar with the calorie concept model, etc., calorie concept replacement can be carried out. For example, when making diet recommendations, commonly used measurement units in daily life can be provided to users in the form of text or pictures as quantity references, such as half a potato, half a bowl of rice, etc.

[0207] In addition, the reporting unit 706 can also output information suggesting the intake amount for users, reminder information for taboo foods, reminder information for excessive intake, reminder information for the amount of exercise that needs to be increased, and statistical information on the intake amount within a predetermined time period, according to the determined diet behavior plan.

[0208] For example, for diabetic patients, it is prompted that they cannot eat high-sugar foods such as chocolate.

[0209] In addition, for users with excessive energy, a diet and exercise remedy plan for the next step can be prompted. For example, according to the currently monitored blood sugar, blood pressure, and blood lipid indicators of the user, combined with the amount of food consumed and the amount of exercise and work intensity of the user today, the recommended amount of food for this meal is suggested for the user. If it is found that the amount exceeds this recommended amount, the user is reminded to eat less for this meal. For example: only eat half of the meat dishes, or 2 / 3, etc., or the intake amount required for the next meal, or the amount of exercise that needs to be increased today.

[0210] In addition, for example, a diet report can be generated at the end of each week to analyze the food intake situation this week, whether the calorie intake exceeds the upper limit, and give recommendations for the next week's recipe.

[0211] The reporting unit 706 can provide regular recipe recommendations, including the calorie and GI of ingredients, etc. In addition, it can also provide recommendations for the next week's recipe. Combining the basic information input by the user, the disease monitoring data this week, and the diet and exercise situation, a personalized recommendation for the food combination of three meals a day for the next week is provided, including staple foods, vegetables, meat, poultry, eggs, salt, etc., and the recommended cooking methods are also given. The recommended recipe for the next week is pushed at the end of each week. For example

[0212] The reporting unit 706 can provide personalized customized recipes. For example, after the user manually enters the ingredients, according to the current disease status, eating habits, and professional characteristics of the patient, the recommended cooking methods and seasonings suitable for patients with high blood pressure, high blood sugar, and high blood lipids are given.

[0213] Therefore, the multiple chronic disease joint management device according to the fourth embodiment of the present application can manage multiple chronic diseases including hypertension, hyperglycemia, and dyslipidemia as a whole. The diet recommendation plan can be conveniently provided to the user, and personalized and accurate recommendations can be given for the user's energy input amount. Diversified presentation methods can be provided for various monitoring data of the user, facilitating the user to conveniently understand their own condition and increasing the user's usage stickiness.

[0214] <Fifth Embodiment>

[0215] The following will be combined with Figure 9-10 Describe in detail the multiple chronic disease joint management device based on machine learning according to the fifth embodiment of the present application.

[0216] The difference between the multiple chronic disease joint management device 900 according to the fifth embodiment and the multiple chronic disease joint management device 100 according to the first embodiment is that a sports behavior intervention unit 905 is additionally added. For the sake of brevity of description, the repeated description of the same unit components as those of the multiple chronic disease joint management device 100 is omitted here.

[0217] The sports behavior intervention unit 905 inputs the user's personal information and data related to multiple chronic diseases into a sports model, and determines a personalized sports plan for multiple chronic diseases suitable for the user. The sports model is a machine learning model.

[0218] In the joint management of multiple chronic diseases, sports management is a very important link. Existing medical guidelines usually only give simple suggestions for a single disease. For example, do not exercise strenuously, but do not consider different user groups for how to specifically monitor and manage sports. Patients of different ages have completely different requirements for sports in different seasons. For example, most elderly patients like to sit still or take a slow walk, while young patients hope for strenuous ball games, etc. In this case, if a personalized sports plan can be recommended according to the user's situation and sports management can be carried out in the sports way the user likes, it will be more helpful for the user to establish a healthy lifestyle.

[0219] The sports model used by the sports behavior intervention unit 905 takes into account the sports characteristics of different patients and is implemented through a machine learning model.

[0220] Next, the training process of the medication model will be described with reference to Figure 10 Describe the training process of the medication model.

[0221] Step S1001: Obtain a user dataset and a medical treatment guideline dataset.

[0222] The user data set includes personal information of multiple users, data related to various chronic diseases, data on sports hobbies, and data on seasons.

[0223] The medical treatment guideline data set includes exercise guidelines related to various diseases.

[0224] Step S1002: Input the user data set and the medical treatment guideline data set as training data into a machine learning model for training to obtain an exercise model for different users' recommended exercise plans.

[0225] The machine learning model extracts parameters such as BMI, gender, age, underlying diseases, occupation, and sports hobbies in the user's personal information as labels, extracts multiple exercise recommendations for various chronic diseases specified in the medical treatment guidelines, and matches them according to the number of labels and the multiple exercise recommendations to determine the user's exercise recommendations.

[0226] The architecture of the machine learning model in this embodiment is basically the same as that of the machine learning model in the first embodiment, and its detailed description is omitted here.

[0227] By using the data of a large number of users to train the exercise model, the final exercise model is obtained. In addition, the medical treatment guidelines can be automatically obtained to retrain the exercise model to improve the accuracy of exercise recommendations.

[0228] For example, the exercise model can analyze the user's BMI, gender, age group, underlying diseases, occupational characteristics (light / medium / heavy physical strength), and sports hobbies (including exercise time, exercise venue, exercise form, etc.) and output an exercise method suitable for the user.

[0229] For example, for elderly ordinary hypertensive patients, a slow walk with a small amount of exercise can be recommended. For young ordinary diabetic patients, appropriate badminton exercise can be recommended, etc.

[0230] In addition, the input unit 101 can also collect data such as the user's cardiopulmonary function evaluation results, sports hobbies, and living habits using wearable devices. The user is allowed to synchronize the physical fitness test data of external apps such as IOS Health, WeChat Sports, and Keep. The synchronized content includes, for example, the number of daily steps and the real-time heart rate change during exercise.

[0231] The difference between the reporting unit 906 in the management device 900 according to the fifth embodiment and the reporting unit 104 in the first embodiment is that in addition to the function of the reporting unit 104, the reporting unit 906 can also output reminder information to remind the user to exercise according to the exercise plan.

[0232] For example, personalized daily exercise volume and exercise method recommendations can be provided (Kcal compliance rate greater than 80% or exercise 5 days a week, 30 minutes a day), and the daily required exercise volume can be adjusted by combining the daily food intake, recent monitoring data of the patient, and drug intake, and a personalized weekly exercise plan can be given. At the same time, different exercise items with the same consumption amount can be provided as alternative plans (form + duration + intensity).

[0233] The reporting unit 906 can also provide users with personalized exercise reports. For example, a comprehensive report is generated at the end of each week to analyze the exercise situation of this week, whether the calorie consumption value reaches the expectation, and give exercise suggestions for the next week. It can also provide a pie chart of the exercise volume compliance, a total calorie consumption bill (which can be presented in the form of daily, weekly, quarterly, and annual), and changes in body fat / weight / BMI.

[0234] In addition, the reporting unit 906 can also provide types of exercise that are not recommended. For example, for elderly patients, strenuous exercise is not recommended.

[0235] In addition, the reporting unit 906 can also link to health coaches on external platforms and link to exercise communities. Set functions such as punching in and exercise ranking. For example, randomly match users with the same three high levels and individual information to challenge the daily exercise volume. The winner can obtain additional willpower, etc.

[0236] In addition, the reporting unit 906 can also push a library of exercise teaching videos in stages.

[0237] Therefore, the multi-chronic disease combined management device according to the fifth embodiment of the present application can manage multiple chronic diseases including hypertension, hyperglycemia, and dyslipidemia as a whole. The exercise recommendation plan can be conveniently provided to users, and various monitoring data can be presented to users in a diversified manner, facilitating users to conveniently understand their own conditions and increasing user stickiness.

[0238] <Sixth Embodiment>

[0239] Next, reference will be made to Figure 11 Describe in detail the multi-chronic disease combined management device based on machine learning according to the sixth embodiment of the present application.

[0240] The multi-chronic disease combined management device 1100 according to the sixth embodiment includes all the units of the multi-chronic disease combined management device according to the first to fifth embodiments. It should be noted that all the units of the multi-chronic disease combined management device according to the first to fifth embodiments can be combined in any way to form a new management device.

[0241] Each unit of the combined chronic disease management device 1100 according to the sixth embodiment is the same as the unit of the combined chronic disease management device of the first to fifth embodiments, and its detailed description is omitted here.

[0242] Next, the operation process of using the combined chronic disease management device 1100 will be described in the form of a case.

[0243] For example, the information obtained through the input unit 101 of the combined chronic disease management device 1100 includes: Patient A is male, 35 years old, and the basic personal information includes BMI: 26 kg / m 2 , diagnosed with type 2 diabetes in the hospital 3 years ago, without atherosclerotic cardiovascular disease. Currently, metformin 500 mg is taken orally once or twice a day, and the blood sugar has been controlled stably. A severe hypoglycemia occurred 2 days ago, and the random blood sugar was 2.9 mmol / L. The physical sign monitoring data uploaded today is: HbA1c: 8.2%, fasting blood sugar: 9.3 mmol / L, blood pressure 136 / 85 mmHg.

[0244] The risk level determination unit 102 of the combined chronic disease management device 1100 determines that:

[0245] 1. Label 1: Type 2 diabetes;

[0246] 2. Label 2: Male;

[0247] 3. Label 3: Age < 60 years old;

[0248] 4. Label 4: Without atherosclerotic cardiovascular disease;

[0249] 5. Label 5: History of severe hypoglycemia.

[0250] Combining Labels 1 to 4, the risk level determination unit 102 determines that the risk level is a medium - low risk level.

[0251] The management unit 103 recommends the control targets as:

[0252] HbAlc < 7.0%, fasting blood sugar: 4.4 - 7.0 mmol / L, non - fasting blood sugar: < 10.0 mmol / L, blood pressure < 130 / 80 mmHg, total cholesterol < 4.5 mmol / L, HDL - C > 1.0 mmol / L, triglyceride < 1.7 mmol / L, LDL - C < 2.6 mmol / L, BMI ≤ 24 kg / m 2 .

[0253] However, combining Label 5, the patient has a history of severe hypoglycemia, and the management unit 103 recommends the control targets as:

[0254] HbAlc<8%, fasting blood glucose: 4.4-7.0mmol / L, non-fasting blood glucose<10mmol / L, blood pressure<130 / 80mmHg, total cholesterol (TC)<4.5mmol / L.

[0255] When the two recommended control targets are inconsistent, a relatively strict control target can be recommended. Therefore, the following guidance suggestions are given.

[0256] 1) Management unit 103 recommends the following control targets: HbAlc <7.0%, fasting blood glucose: 4.4-7.0mmol / L, non-fasting blood glucose: <10.0mmol / L, blood pressure <130 / 80mmHg, total cholesterol <4.5mmol / L, HDL-C >1.0mmol / L, triglycerides <1.7mmol / L, LDL-C <2.6mmol / L, BMI ≤24 kg / m 2 .

[0257] 2) Monitoring recommendations: Monitor fasting or 2 hours postprandial blood sugar 2-4 times a week.

[0258] 3) Dietary intervention unit 705 recommends: Reduce total daily dietary calories by at least 400-500 kcal; energy provided by carbohydrates should account for 50%-65% of total energy; recommended protein intake is about 0.8 g•kg -1 •d -1 ; Saturated fatty acid intake accounts for less than 30% of total fatty acid intake; Salt intake is <6g / d.

[0259] 4) Exercise recommendations in dietary behavior intervention unit 905: Refer to exercise recommendations in the 2017 CDS guidelines P11: At least 150 minutes per week (such as 5 days of exercise per week, 30 minutes every day) of moderate intensity (50%-70% of maximum heart rate, a little effort during exercise, faster heartbeat and breathing but not rapid) aerobic exercise.

[0260] 5) Medication determination unit 305 recommends medication suggestions: Try to avoid using hypoglycemic drugs that cause hypoglycemia, such as insulin, sulfonylureas, etc.

[0261] 6) Management Unit 103 recommends follow-up: For patients whose blood sugar is well controlled and reaches the target, it is recommended to measure HbA1c twice a year;

[0262] 7) Management Unit 103 recommends other suggestions: Try to quit smoking and drinking; if you need to drink, avoid drinking on an empty stomach; do not drink more than 25g of alcohol (15g of alcohol is equivalent to 350ml of beer, 150ml of wine or 45ml of distilled spirits). Do not drink more than twice a week.

[0263] Therefore, the multi-chronic disease joint management device according to the sixth embodiment of the present application can manage multiple chronic diseases including hypertension, hyperglycemia, and dyslipidemia as a whole. The diet / exercise recommendation plan can be conveniently provided to the user, personalized and accurate recommendations can be given for the patient's energy intake, the food / drug incompatibility can be conveniently provided to the user, and various monitoring data can be presented to the user in a diversified manner, facilitating the user to conveniently understand their own condition and increasing the user's stickiness.

[0264] Figure 12 is a block diagram showing a multi-chronic disease joint management device according to the seventh embodiment of the present disclosure.

[0265] Refer to Figure 12 , the electronic device 1200 may include a processor 1201 and a memory 1202. The processor 1201 and the memory 1202 can both be connected through a bus 1203. The electronic device 1200 can be an electronic device such as a smart phone, a laptop computer, a desktop computer, or a server device.

[0266] The processor 1201 can perform various actions and processes according to the programs stored in the memory 1202. Specifically, the processor 1201 can be an integrated circuit chip with signal processing capabilities. The above-mentioned processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc., and can be of the X86 architecture or the ARM architecture.

[0267] The memory 1202 stores computer instructions, which implement the operations of the above-mentioned multi-chronic disease combined management device 1100 when executed by the processor 1201. The memory 1202 can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. The non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct rambus random access memory (DR RAM). It should be noted that the memories of the methods described herein are intended to include but not be limited to these and any other suitable types of memories.

[0268] The present disclosure also provides a computer-readable storage medium, on which computer instructions are stored, and when the computer instructions are executed by a processor, the operations of the above-mentioned multi-chronic disease combined management device 1100 can be implemented. Similarly, the computer-readable storage medium in the embodiments of the present disclosure can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. It should be noted that the computer-readable storage media described herein are intended to include but not be limited to these and any other suitable types of memories.

[0269] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order from that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0270] In general, the various example embodiments of the present disclosure can be implemented in hardware or dedicated circuits, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while other aspects can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of the embodiments of the present disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatus, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuits or logic, general hardware or a controller or other computing device, or some combination thereof.

[0271] The example embodiments of the present invention described in detail above are merely illustrative and not restrictive. Those skilled in the art should understand that various modifications and combinations can be made to these embodiments or their features without departing from the principles and spirit of the present invention, and such modifications should fall within the scope of the present invention.

Claims

1. A device for the combined management of multiple chronic diseases based on machine learning, comprising: A data input unit configured to input personal information of a user and data related to multiple chronic diseases, wherein the multiple chronic diseases include hypertension, diabetes, and dyslipidemia; A risk level determination unit configured to input the personal information of the user and the data related to multiple chronic diseases into a risk level model to obtain a risk level corresponding to the user among multiple risk levels of multiple chronic diseases, and the risk level model is a machine learning model. Wherein the training process of the risk level model includes: Obtaining a user data set and a medical guideline data set, and analyzing potential data associations or relationships existing between the user data set and stage detection data, and the stage detection data includes fluctuations in hypertension, diabetes, and dyslipidemia values; Inputting the user data set and the medical guideline data set as training data into a machine learning model for training to obtain a risk level model for stratifying users. Among them, input and output data pairs with potential data associations or relationships are selected to test multiple algorithm models, and one or more algorithm models are selected for the risk level model based on the correct prediction probability of the results of the algorithm models; and A management unit configured to monitor data related to the multiple chronic diseases, and determine personalized control targets for the multiple chronic diseases suitable for the user according to the personal information of the user and the risk level of the user. The control targets include control targets for multiple physical sign parameters of the user related to one or more of blood glucose, blood pressure, and blood lipids. The management unit is further configured to set the values of the control targets for multiple physical sign parameters of users with a high risk level to be lower than the values of the control targets for multiple physical sign parameters of users with a low risk level according to the risk level of the user; and A medication determination unit configured to input the personal information of the user and the data related to multiple chronic diseases into a medication model to determine a personalized medication plan for the multiple chronic diseases suitable for the user. The medication model is a machine learning model. Among them, the medication model analyzes the side effects of a predetermined drug for one of the multiple chronic diseases on other chronic diseases, and adjusts the user's medication plan to recommend drugs without side effects on other chronic diseases.

2. The management device according to claim 1, wherein The personal information of the user includes one or more of the following: age, gender, height, weight, body mass index (BMI), long-term place of residence, eating habits and preferences, underlying diseases, allergy history, occupation, daily activity level, and sports hobbies; And The data related to multiple chronic diseases includes one or more of the following: user physical sign data including detection data of hypertension, diabetes, and dyslipidemia, medical history data of hypertension, diabetes, and dyslipidemia, cardiovascular risk factors, medical history data of other diseases, patient medication situation data, drug instructions, drug indications, drug contraindications, drug incompatibility taboos, drug side effects, drug inventory, and drug prices.

3. The management device according to claim 2, wherein The user data set includes personal information of multiple users, data related to various chronic diseases, and data on risk levels. The medical guideline data set includes medical treatment guidelines, expert consensus, medical journals, medical textbooks, medical dictionaries, drug instructions, doctor diagnosis data, and doctor-patient medication data related to various diseases.

4. The management device according to claim 3, wherein, The risk level model uses one or more of the input personal information of the user and the data related to various chronic diseases as risk factors, and dynamically evaluates the risk level of the user according to the number of risk factors.

5. The management device according to claim 4, wherein, The risk level model evaluates the risk of the user as a high level in response to an increase in the number of risk factors, and evaluates the risk level of the user as a low level in response to a decrease in the number of risk factors.

6. The management device according to claim 5, wherein the management unit further sets a danger threshold for each of the various chronic diseases according to the risk level of the user.

7. The management device according to claim 6, further comprising: A reporting unit configured to output a monitoring report for the various chronic diseases of the user according to the control target.

8. The management device according to claim 7, wherein, The reporting unit outputs the monitoring report at a predetermined frequency or at a preset time.

9. The management device according to claim 8, wherein, The reporting unit outputs a danger warning signal in response to the monitoring result indicating that one or more of the various chronic diseases exceed the danger threshold.

10. The management device according to claim 1, wherein, The risk level determination unit is further configured to use one or more of the input personal information of the user and the data related to various chronic diseases as risk factors to evaluate and / or predict the cardiovascular disease risk level of the user.

Images