A stroke / infarction early warning method and system based on radial artery dynamic monitoring

Abstract

The present invention discloses a stroke / infarction early warning method and system based on dynamic monitoring of radial artery. The present invention adopts wearable devices in pulse data collection, realizes real-time data collection, analysis and early warning, and effectively copes with the sudden characteristics of stroke. At the same time, based on the intelligent system and networking communication function of the wearable device, clinical accompanying symptom data can also be selectively uploaded to ensure the accuracy and timeliness of the data.

Description

Technical Field

[0001] The present invention relates to the field of medical health, and in particular to a stroke / apoplexy early warning method and system based on dynamic monitoring of radial artery. Background Art

[0002] The main method of traditional stroke screening is to go to the hospital for a professional examination to see if the head and neck arteries are narrowed due to blockage. However, this method is more troublesome to implement, the cost is high, and it is not conducive to preventing the occurrence of random strokes. Various blood pressure monitors on the market can only perform random detection of blood pressure information, and have no decisive relationship with the occurrence of stroke (stroke), so it is even more impossible to predict the occurrence of stroke (stroke). Lack of risk assessment of stroke (stroke), and taking measures after the occurrence of stroke (stroke) will bring more serious impacts. For example, the Chinese patent "An algorithm for predicting abnormal blood pressure leading to stroke based on big data and artificial intelligence", the patent with publication number CN111462911A uses pulse pressure and machine language to establish a prediction model to achieve early warning monitoring of stroke / stroke, while the simple pulse pressure monitoring is single, resulting in inaccurate prediction data, increasing the medical costs of the monitored personnel and the waste of medical resources. Summary of the invention

[0003] The purpose of the present invention is to overcome the above-mentioned deficiencies in the prior art and to provide a method and system for early warning of stroke / apoplexy based on dynamic monitoring of radial artery with a reasonable structural design.

[0004] The technical solution adopted by the present invention to solve the above-mentioned problem is: a stroke / stroke early warning method based on radial artery dynamic monitoring, comprising:

[0005] Database establishment: Collect pulse data of the left radial artery of stroke patients before onset of the disease to establish a pulse feature database;

[0006] Collect clinical symptoms before the onset of stroke / stroke patients, obtain core symptoms after processing, and obtain subdivided characteristics of the core symptoms according to the severity of the symptoms and assign scores b. The highest score of the same subdivided characteristic is defined as c, and compile and establish a core symptom library;

[0007] Data analysis: The radial artery in the Cun position runs along the radial flexor carpi tendon and the extensor pollicis longus tendon, along the front edge of the styloid process, around the bottom of the styloid process tip, to the deep surface of the pollicis longus tendon and the extensor pollicis brevis tendon. The deep surface of the radial artery here is the middle depression between the palmar carpal ligament and the scaphoid bone. The pulse data is classified based on the pulse of traditional Chinese medicine. The pulse has flat, stringy, slippery, rapid, slow, weak, and thin pulses. Floating pulse, deep pulse, thin pulse, stringy pulse, solid pulse, weak pulse, and weak pulse all have the risk of stroke. The above pulse data are entered and quantified according to the severity. The collected pulse data are analyzed using machine learning language to obtain the correlation coefficient of the above pulse causing stroke, and obtain A, A ≤1, and obtain the pulse data association probability model. =A 100%;

[0008] The determination of the initial core symptoms can be based on the input of information on symptoms known to affect stroke in current diagnostic experience, thereby compiling and establishing a core symptom library, such as upper limb paralysis, lower limb paralysis, dizziness, headache, dark purple lips, etc., and determining the correlation between the corresponding core symptoms and stroke / stroke using machine learning language analysis, and obtaining the definition of the correlation coefficient B of the corresponding core symptoms causing stroke. The sum of the correlation coefficients of all core symptoms is 1, and the core symptom association probability model is obtained. , where n represents the number of data on the core symptoms that cause stroke, B k represents the correlation coefficient of the core symptoms causing stroke corresponding to the kth group of data, b k represents the core symptom score corresponding to the kth group of data, c k Indicates the highest score allowed for the core symptoms in the data; Data application: Collect pulse data from the left radial artery of the stroke / stroke monitor and core clinical accompanying symptom data, and upload them to the database;

[0009] Set warning thresholds for pulse data association probability and core symptom association probability and establish warning models for pulse data association probability and core symptom association probability; apply the collected data to the warning model to trigger warnings; issue warnings when either the pulse data association probability or the core symptom association probability exceeds the preset threshold; establish a monitor information file in the database system; and push warning information to the monitor, the monitor's guardian, or any one or more designated medical staff via the database application platform; the warning information includes reminder information and health advice or medical advice.

[0010] Further: Use machine learning language to analyze the correlation between the probability of pulse data association and the probability of core symptoms inducing stroke / stroke, and determine the probability coefficient of pulse data association And the core symptom association probability coefficient , obtain the comprehensive stroke probability model ;in , adopt a comprehensive monitoring method, use machine learning voice to determine the impact of the combined association of pulse data association probability and core symptom association probability on stroke / stroke, respectively determine the proportion of the association probability of pulse data and the core symptom association probability in causing stroke / stroke, determine the association probability coefficient, and determine the comprehensive probability model of stroke / stroke through the coefficient. This method is used for comprehensive analysis when both the pulse data association probability and the core symptom association probability are low, so that the risk estimation value is more accurate. At the same time, it can also be used for comprehensive analysis when the pulse data association probability or the core symptom association probability is at high risk, and the risk of stroke / stroke can be estimated by determining its change amplitude.

[0011] Furthermore: the pulse data is monitored and uploaded by wearable devices, the core symptoms are entered by PC, mobile device and any one or more of the above-mentioned wearable devices, and the core clinical accompanying symptoms are entered by medical staff or by risk groups or by guardians of risk groups in an open manner, and the entry form is list selection or a combination of list selection and autonomous description. On the one hand, when entering, list selection is performed, wherein the data information selected in the list is processed, and the autonomously entered information is stored in the system background, and is classified by medical experts in combination with the corresponding patient's stroke / stroke symptoms and added to the list, so that the core symptom library is improved according to the data information.

[0012] Furthermore: when the collected data is processed by the comprehensive probability model of stroke, and the comprehensive probability of stroke obtained exceeds the threshold, the person is identified as a person at risk of stroke. Data collection and comprehensive probability prediction are carried out on the person at risk of stroke within the set time threshold. When a continuous increase in the probability prediction is obtained in multiple time periods, it is identified as the emergency period of stroke, and the application platform pushes early warning information and pushes it to the medical guardian for medical system intervention. Medical examinations are conducted on the risk population and real symptoms are obtained. Emergency medical intervention is carried out when there is a real risk of stroke and a critical state. When the probability prediction obtained in multiple time periods is in a decreasing state, it is identified as the mitigation period of stroke, and the application platform pushes risk reminders and intervention suggestions.

[0013] Further: based on the correlation between the comprehensive stroke probability increase data obtained by the monitored person and the onset time, a time prediction model T is established using machine learning. The above model is used to evaluate and predict the time when a stroke will occur in the risk population, a time threshold is set, and emergency pre-tensioning and emergency medical system intervention are carried out when the set time threshold is reached. Medical examinations are conducted on the risk population and real symptoms are obtained. When there is a real risk of stroke or a critical condition, emergency medical intervention is carried out. At the same time, the time prediction helps the corresponding responsible person to estimate the time, which is of great reference significance for the time coordination of medical treatment and medical system intervention.

[0014] Furthermore: the database is jointly built by hospitals at all levels and data is shared. Hospitals at all levels can call the medical data collected from the archived population under permission conditions to achieve data sharing, which is also conducive to the improvement of database data and further ensures the reliability of monitoring data.

[0015] Further: Setting the core symptom association probability model correction coefficient , where n represents the number of data on the core symptoms that cause stroke, B k The core symptom-induced stroke correlation coefficient corresponding to the kth group of data is represented, and the modified core symptom correlation probability model is obtained. ,Since the sum of the correlation coefficients of all core symptoms is 1, when the core symptoms cannot be entered with comprehensive information, the probability will be inaccurate. Therefore, the core symptom correlation probability model is modified to obtain relatively accurate probability information. Therefore, this method constitutes a system, which is more comprehensive in entering core symptoms and more effective in ensuring the accuracy of data. For example, the core symptoms of upper limb activity, lower limb activity, dizziness, headache, and dark purple lips that cause stroke have correlation coefficients B defined as 0.3, 0.3, 0.1, 0.1, and 0.2, respectively. When the symptom of headache is missing, the scores are 3, 2, 4, and 4, respectively, and the maximum score is 5. % %; , then the corrected probability %

[0016] Further: the terminal for pushing the warning information includes a PC, a mobile device, and any one or more of the above-mentioned wearable devices.

[0017] Further: the pulsation data is the left hand radial artery value.

[0018] A stroke / stroke early warning system based on dynamic monitoring of the radial artery is constructed using the logic of the above method to provide stroke / stroke early warning for people who need dynamic monitoring.

[0019] Compared with the prior art, the present invention has the following advantages and effects:

[0020] (1) The present invention constructs a large database of stroke pathological pulse characteristics and stroke symptom characteristics, and uses machine learning to establish a model that can calculate the probability and approximate time of onset, thereby achieving timely prediction and early warning of the occurrence of the disease in stroke prevention and treatment.

[0021] (2) Hospitals at all levels jointly build and share data, effectively ensuring the data volume, optimizing the data model and ensuring its accuracy, so that professional doctors at all levels can obtain medical data in a timely manner, and providing timely and effective guarantees for stroke prevention and emergency treatment.

[0022] (3) The present invention is more humane and professional in establishing the system, which ensures the diversity and specialization of data while also facilitating the operation of users and medical staff.

[0023] (4) The present invention adopts wearable devices in pulse data collection, realizes real-time data collection, analysis and early warning, and effectively responds to the sudden characteristics of stroke. At the same time, based on the intelligent system and networking communication function of the wearable device, it can also select and upload clinical symptom data to ensure the accuracy and timeliness of the data.

[0024] (5) The system constructed by this method can push warnings to multiple roles according to the setting requirements, and ensure the timeliness of the monitor's response to the warning according to the actual situation of the monitor, which is of great significance for stroke prevention and emergency treatment.

[0025] (6) The present invention uses machine learning language to grasp the correlation between the comprehensive probability increase and decrease data obtained from the monitored person and the onset time, which has important guiding significance for the preparation before the onset of stroke and emergency intervention. DETAILED DESCRIPTION

[0026] The present invention will be further described in detail below by way of examples. The following examples are provided to explain the present invention but the present invention is not limited to the following examples. Example

[0027] This embodiment relates to a stroke / apoplexy early warning method and system based on radial artery dynamic monitoring. The following takes a monitor as an example to illustrate this method. Patients are collected in the hospital. The prospective patient has a potential risk due to headache and is listed as a monitor. Then his information is filed in the database and the wearable device is worn. The early warning method is as follows:

[0028] Method 1: The wearable device collects its pulse and uploads it to the cloud database. The result obtained by using the pulse data association probability model is , exceeding the threshold triggers an early warning;

[0029] Method 2: Upload core symptoms such as headache, dizziness, numbness on one side of the body, etc. and assign points to obtain the core symptom association probability model , exceeding the threshold triggers an early warning;

[0030] Method 3: If any of the uploaded disease characteristics or pulse data exceeds the threshold, the other data will be recorded as an early warning, and the comprehensive probability model of stroke will be used to obtain the comprehensive probability P of causing stroke. If it exceeds the threshold, an early warning will be triggered;

[0031] Method 4: Based on method 3, after the threshold warning is exceeded, multiple data entry is performed within the set time interval;

[0032] For example:

[0033] As above, the combined probability is 78% after measuring again at intervals of 8 to 24 hours.

[0034] As above, the combined probability is 82% after measuring again at intervals of 8 to 24 hours.

[0035] As above, the combined probability is 88% after measuring again at intervals of 8 to 24 hours.

[0036] If the amplitude continues to rise, it can be warned that it is the emergency stage of stroke. The system will automatically warn and notify the relevant medical institutions or the monitored person (or his guardian) to take measures. The time interval system can be adjusted according to the degree of increase in the amplitude.

[0037] On the contrary, for example:

[0038] The combined probability was 72% when the measurements were repeated after an interval of 8 to 24 hours.

[0039] The combined probability was 68% after measuring again at intervals of 8 to 24 hours;

[0040] If its amplitude continues to decrease, it can be clearly determined as the remission period of stroke, and the system will automatically remind and give intervention suggestions. Example

[0041] According to the fourth method mentioned above, the relationship between the probability increase amplitude and time is obtained, and a time prediction model is established using machine learning to obtain a prediction of the predicted onset time of stroke. As the predicted onset time gradually shortens, the measurement interval can be gradually shortened. At the same time, based on the gradually shortened or increased amplitude data of the onset time, the classification of intervention measures can be set, which has important guiding significance for the preparation before the onset of stroke and emergency intervention. At the same time, through this system, it also has important guiding significance for the prevention of stroke / stroke onset and the severity of stroke / stroke after intervention.

[0042] The above contents described in this specification are merely examples of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, as long as they do not deviate from the contents of the present specification or exceed the scope defined by the claims, they shall all fall within the protection scope of the present invention.

Claims

1. A stroke early warning method based on radial artery dynamic monitoring, characterized in that: The steps include: Database establishment: Collect pulse data of stroke patients before onset a and establish a pulse feature database; Collect clinical symptoms before the onset of stroke / stroke patients, obtain core symptoms after processing, and obtain subdivided characteristics of the core symptoms according to the severity of the symptoms and assign scores b. The highest score of the same subdivided characteristic is defined as c, and compile and establish a core symptom library; Data analysis: Use machine learning language to analyze the correlation between pulse data and stroke, and obtain the correlation coefficient A of the corresponding pulse data feature causing stroke, A ≤1, and obtain the pulse data association probability model =A 100%, using machine learning language to analyze the correlation between the corresponding core symptoms and stroke / stroke, and obtain the corresponding core symptom-induced stroke correlation coefficient definition B, the sum of the correlation coefficients of all core symptoms is 1, and the core symptom correlation probability model is obtained ; Data application: Collect pulse data from the left radial artery of the stroke monitor and core clinical symptom data, and upload them to the database; Set the pulse data association probability and core symptom association probability warning thresholds and establish a pulse data association probability and core symptom association probability warning model, apply the collected data to the warning model to trigger the warning, establish a collection population file in the database system, and the database application platform pushes warning information to the monitor, the monitor's guardian, and any one or more designated medical staff; In data analysis, machine learning language is used to analyze the correlation between the probability of pulse data association and the probability of core symptoms inducing stroke / stroke, and determine the probability coefficient of pulse data association And the core symptom association probability coefficient , obtain the comprehensive stroke probability model ;in ; Set the core symptom association probability model correction coefficient , where n represents the number of data on the core symptoms that cause stroke, B k The core symptom-induced stroke correlation coefficient corresponding to the kth group of data is represented, and the modified core symptom correlation probability model is obtained. .

2. The stroke / infarction early warning method based on radial artery dynamic monitoring according to claim 1, characterized in that: The pulse data is monitored and uploaded by wearable devices, the core symptoms are entered by PC, mobile device and any one or more of the above-mentioned wearable devices, and the core clinical accompanying symptoms are entered by medical staff, risk groups or guardians of risk groups in an open manner, and the entry form is list selection or a combination of list selection and self-description, among which the self-described symptoms are classified by medical experts in combination with the corresponding patient's stroke / stroke symptoms and added to the list.

3. The stroke / infarction early warning method based on radial artery dynamic monitoring according to claim 2 is characterized in that: When the collected data is processed by the comprehensive stroke probability model and the comprehensive probability of stroke obtained exceeds the threshold, the person is identified as a person at risk of stroke. Data collection and comprehensive probability prediction are performed on the person at risk of stroke within the set time threshold. When the probability prediction shows a continuous increase in multiple time periods, it is identified as the emergency period of stroke, and an early warning is issued and the medical system intervenes. When the probability prediction shows a decrease in multiple time periods, it is identified as the mitigation period of stroke, and the application platform pushes risk reminders and intervention suggestions.

4. The stroke / infarction early warning method based on radial artery dynamic monitoring according to claim 3 is characterized in that: In data analysis, the correlation between the comprehensive stroke probability increase data obtained from the monitored person and the onset time is used to establish a time prediction model T using machine learning, and a time warning threshold is set. When the predicted time exceeds the threshold, a warning is issued or appropriate medical intervention measures are taken.

5. The stroke / infarction early warning method based on radial artery dynamic monitoring according to claim 1 is characterized by: The database is jointly built by hospitals at all levels and data is shared. Hospitals at all levels can call upon medical data collected from the registered population under permitted conditions.

6. The stroke / injury early warning method based on radial artery dynamic monitoring according to claim 2 is characterized in that: The terminals for pushing warning information include PC, mobile device and any one or more of the above-mentioned wearable devices.

7. The stroke / injury early warning method based on radial artery dynamic monitoring according to claim 1 is characterized by: The pulsation data are the left hand radial artery pulse data.

8. A stroke early warning system based on radial artery dynamic monitoring, characterized in that: The method is composed of the stroke warning method described in any one of claims 1 to 7.