Anti-falling method, anti-falling system and anti-falling foot ring

Abstract

The invention discloses an anti-tumble method, an anti-tumble system and an anti-tumble foot ring, and relates to the technical field of protective equipment, and the method comprises the steps: obtaining the body information of a target user through dynamic body monitoring, collecting road and environment perception data, constructing a tumble database, and extracting tumble and non-tumble data sets; training to obtain an intelligent tumble discriminator, carrying out tumble risk analysis by combining the physical state of the target user and the road and environment information, and generating a discrimination result; and when the judgment result meets the early warning condition, the voice propaganda and education module is activated, and a fall early warning is sent to the user in time. The technical problems that an existing fall early warning system depends on single sensor data, and the accuracy of risk assessment and the timeliness of early warning response are insufficient are solved, the fall risk of the user is accurately predicted and fed back in time through multi-dimensional data fusion analysis, and the user experience is improved. The technical effect of effectively preventing and intervening the falling event is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of protective equipment, and in particular to a fall prevention method, a fall prevention system and a fall prevention foot ring. Background Art

[0002] Early prevention of falls and providing timely risk warnings are important means to protect the health of the elderly and reduce fall accidents. Existing fall prevention technologies mostly monitor the user's physiological state, posture changes or walking style to determine the occurrence of falls. They can automatically alarm when a fall occurs to remind the user or surrounding personnel to intervene. However, most of these traditional methods focus on single physical signal monitoring, such as accelerometers, gyroscopes, etc., and often ignore the impact of the user's surrounding environment and external factors on the risk of falls. They still face problems such as insufficient perception of environmental factors, inaccurate data processing, and lack of personalized real-time feedback. Summary of the invention

[0003] The present application provides a fall prevention method, a fall prevention system and an anti-fall anklet, which are used to solve the technical problems that the existing fall warning system relies on single sensor data, and the accuracy of risk assessment and the timeliness of warning response are insufficient.

[0004] In a first aspect of the present application, a fall prevention method is provided, the method comprising: obtaining target external side perception information of the target user, the target external side perception information comprising road perception information and environmental perception information; establishing a fall database, and extracting a first fall data set and a first non-fall data set from the fall database; performing supervised machine learning on the first fall data set and the first non-fall data set, and training to obtain an intelligent fall discriminator; performing a fall analysis on the target user side information, the road perception information and the environmental perception information by the intelligent fall discriminator to obtain a target discrimination result; when the target discrimination result meets a predetermined warning constraint, activating a voice education module to provide a fall voice warning to the target user.

[0005] In a second aspect of the present application, a fall prevention system is provided, the system comprising: a dynamic body monitoring module, the dynamic body monitoring module is used to perform dynamic body monitoring on a target user to obtain target user side information; an external side perception information acquisition module, the external side perception information acquisition module is used to obtain target external side perception information of the target user, the target external side perception information includes road perception information and environment perception information; a fall data extraction module, the fall data extraction module is used to establish a fall database, and extract a first fall data set and a first non-fall data set from the fall database; an intelligent fall prevention system; a dynamic body monitoring module, the dynamic body monitoring module is used to perform dynamic body monitoring on a target user to obtain target user side information; an external side perception information acquisition module is used to obtain target external side perception information of the target user, the target external side perception information includes road perception information and environment perception information; a fall data extraction module, the fall data extraction module is used to establish a fall database, and extract a first fall data set and a first non-fall data set from the fall database; a smart fall prevention system; a dynamic body monitoring ... A discriminator training module, the intelligent fall discriminator training module is used to perform supervised machine learning on the first fall data set and the first non-fall data set, and train to obtain an intelligent fall discriminator; a fall analysis discriminator module, the fall analysis discriminator module is used to perform fall analysis on the target user side information, the road perception information and the environment perception information through the intelligent fall discriminator to obtain a target discrimination result; a fall voice warning module, the fall voice warning module is used to activate the voice education module to provide a fall voice warning to the target user when the target discrimination result meets the predetermined warning constraints.

[0006] The third aspect of the present application provides an anti-fall anklet, which includes a main body, on which is mounted an intelligent monitoring component and a voice education device, and the intelligent monitoring component includes a user-side monitoring component and an external-side monitoring component.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The present application provides a fall prevention method, a fall prevention system and an anti-fall anklet, which relate to the field of protective equipment technology. By dynamically monitoring the target user's physical information and environmental data, a fall database is constructed and an intelligent fall discriminator is trained. The discriminator analyzes the risk of falls based on the user's physical state, road and environmental information, generates a discrimination result, and when the result meets the warning conditions, a fall warning is issued to the user through a voice education module. The existing fall warning system relies on single sensor data, and the technical problems of insufficient accuracy of risk assessment and timeliness of warning response are solved. The multi-dimensional data fusion analysis is used to accurately predict the user's fall risk and provide timely feedback, achieving the technical effect of effective prevention and intervention of fall events. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 A schematic diagram of a fall prevention method flow chart provided in an embodiment of the present application;

[0011] Figure 2 A schematic diagram of the structure of a fall prevention system provided in an embodiment of the present application.

[0012] Explanation of the accompanying drawings: dynamic body monitoring module 11, external side perception information acquisition module 12, fall data extraction module 13, intelligent fall discriminator training module 14, fall analysis and discrimination module 15, fall voice warning module 16. DETAILED DESCRIPTION

[0013] The present application provides a fall prevention method, a fall prevention system and an anti-fall anklet, which are used to solve the technical problems that the existing fall warning system relies on single sensor data, and the accuracy of risk assessment and the timeliness of warning response are insufficient.

[0014] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0015] It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices.

[0016] Embodiment 1, as Figure 1 As shown, the present application provides a fall prevention method, the method comprising:

[0017] P10: Perform dynamic body monitoring on the target user to obtain target user side information.

[0018] Furthermore, step P10 of the embodiment of the present application further includes:

[0019] P11: Reading predetermined hemodynamic characteristics; P12: Traversing and extracting the target body monitoring information of the target user based on the predetermined hemodynamic characteristics to obtain the target user side information; wherein the predetermined hemodynamic characteristics include blood flow, blood flow velocity, blood pressure and vascular resistance.

[0020] It should be understood that dynamic body monitoring of the target user is performed to obtain information on the target user side. Specifically, the predetermined hemodynamic characteristics of the target user are first read to obtain key physiological parameters related to blood flow, including blood flow, blood flow velocity, blood pressure and vascular resistance. Blood flow refers to the amount of blood flowing through a certain part of the blood vessel per unit time, blood flow velocity is the rate at which blood flows in the blood vessel, blood pressure reflects the pressure of blood on the blood vessel wall, and vascular resistance characterizes the difficulty of blood flow. These hemodynamic characteristics can intuitively reflect the state of the user's circulatory system and provide basic data for monitoring user health and judging whether blood flow is sufficient and stable.

[0021] Furthermore, based on the read hemodynamic characteristics, traversal extraction is performed in combination with the real-time body monitoring information of the target user. The target user's body monitoring information includes heart rate, body temperature, movement status, etc., which are collected in real time by wearing sensors (such as optical sensors, pressure sensors, ultrasonic sensors, etc.). These sensors can accurately measure the target user's physiological data and convert them into digital signals for system analysis and processing. By traversing and extracting these data, the system can comprehensively evaluate the user's current health status and generate side information of the target user, namely his or her physiological condition, activity level, and potential risk of falling. These data will provide necessary support for subsequent fall risk identification.

[0022] In general, by accurately monitoring and analyzing the target user's hemodynamic characteristics and body monitoring data, detailed physiological information can be obtained to help the system understand the user's health status in real time, providing a scientific basis for subsequent fall identification and warning. This process not only enhances the comprehensive understanding of the target user's health status, but also improves the accuracy and timeliness of the fall prevention system's response.

[0023] P20: Acquire target external side perception information of the target user, where the target external side perception information includes road perception information and environment perception information.

[0024] Furthermore, to obtain road perception information, step P20 in the embodiment of the present application further includes:

[0025] P21: Acquire the target external scene where the target user is located, wherein the target external scene includes a target road, and the target road includes a first obstacle; P22: Retrieve the obstacle monitoring strategy to perform multi-dimensional obstacle feature collection on the first obstacle to obtain first obstacle feature information; P23: Construct the road perception information based on the first obstacle feature information.

[0026] Optionally, it is crucial to obtain external perception information of the target user. Such perception information includes road perception information and environment perception information, among which road perception information is particularly critical and is used to assess the walking environment and potential obstacle risks of the user.

[0027] First, it is necessary to determine the external scene where the target user is located, especially the target road where the user is located. The target road refers to the walking path where the user is currently located, which may include public roads, stairs, corridors, etc. At this time, there may be potential first obstacles on the target road, such as steps, cracks, potholes, pedestrians or other objects. These obstacles are the main risk factors for falls, so in this step, it is necessary to fully perceive and evaluate the roads around the user to determine the possible obstacles on the road.

[0028] Next, the obstacle monitoring strategy is called up to collect detailed multi-dimensional features of the first obstacle. Multi-dimensional obstacle feature collection refers to obtaining different attributes of obstacles, such as the distance, height, shape, surface state (such as whether it is smooth, slippery, etc.), position change, etc. of the obstacle through sensors or cameras. These features are very important for assessing the severity of obstacles and their impact on user walking. The system can use tools such as lidar, infrared sensors, and visual recognition technology to capture and analyze the characteristics of obstacles in real time, thereby obtaining comprehensive obstacle information.

[0029] Furthermore, based on the collected multi-dimensional feature information of the first obstacle, the system converts it into road perception information. Road perception information includes a detailed description of all obstacles on the target road and their potential impact on the user's walking. This information not only includes the physical properties of the obstacles, but also covers the location, number and changing trend of the obstacles. By integrating this data, the system can establish a comprehensive perception map of the target road, and then evaluate the safety of the target user's walking path and the potential risk of falling.

[0030] Through the above steps, not only can the target user's road environment be accurately sensed, but obstacles on the road can also be fully identified and detailed obstacle information can be provided. This information will help the system effectively assess the safety of the user's walking environment and provide solid data support for subsequent fall risk analysis and early warning.

[0031] Further, the obstacle monitoring strategy includes a first-class obstacle monitoring strategy and a second-class obstacle monitoring strategy, and the obstacle monitoring strategy is called to collect multi-dimensional obstacle features of the first obstacle to obtain first obstacle feature information. Step P22 of the embodiment of the present application also includes:

[0032] Step A: Determine whether the first obstacle belongs to Class I obstacle; Step B1: If it does, obtain the first obstacle feature information of the first obstacle based on the Class I obstacle monitoring strategy; Step B2: If it does not, obtain the first obstacle feature information of the first obstacle based on the Class II obstacle monitoring strategy.

[0033] In a possible embodiment of the present application, an obstacle monitoring strategy is the key to ensure accurate acquisition of road perception information. In order to formulate appropriate monitoring strategies according to different types of obstacles, it is first necessary to determine which type of obstacle the identified first obstacle belongs to. The types of obstacles can be divided into Class I obstacles and Class II obstacles according to their characteristics and possible risks. Class I obstacles generally refer to three-dimensional obstacles (such as roadside stones, walls, railings, etc.) and some more conspicuous obstacles with larger physical volumes. Such obstacles usually constitute obvious physical obstacles to people's walking paths. Class II obstacles include obstacles located on the road surface, such as water stains, oil stains, slippery roads, etc., which usually do not have a three-dimensional form, but may cause slips or difficulties in walking, especially in slippery environments.

[0034] At this time, sensor technology (such as lidar, visual sensors, ultrasonic sensors, etc.) is used to obtain road information in real time and determine the type of obstacle. For example, the system can classify obstacles based on criteria such as the height of the obstacle, surface characteristics, and whether it will cause a slip.

[0035] If the first obstacle belongs to a class of obstacles (such as three-dimensional obstacles or stone piers, etc.), the system needs to obtain the characteristic information of the obstacle based on a class of obstacle monitoring strategy. Class A obstacles usually have fixed positions, large volumes and heights, so the monitoring strategy usually focuses on capturing information such as their position, size, shape and stability. Through LiDAR or 3D imaging technology, the system can accurately measure the size, position and shape of these obstacles, help build a detailed environmental model, and evaluate the impact of these obstacles on the target user's path.

[0036] In addition, computer vision technology (such as image recognition algorithms) can be used to analyze the appearance characteristics of obstacles and determine whether they are high-risk obstacles. For one type of obstacle, the information obtained by the system includes not only the geometric properties of the obstacle, but also its relative position to the user's path, and whether there is a possibility of movement or change.

[0037] If the first obstacle does not belong to the first category of obstacles, that is, it belongs to the second category of obstacles (such as water stains, oil stains, slippery roads, etc.), the system needs to obtain its characteristic information based on the second category of obstacle monitoring strategy. Second category obstacles usually refer to objects that do not block the walking path, but they may make the road surface slippery or increase the risk of falling. In this case, the system mainly monitors the road surface status through environmental sensors (such as humidity sensors, temperature sensors, infrared sensors, etc.).

[0038] For example, problems such as wetness or oil stains may not be easy to judge through physical changes, but require real-time monitoring of road surface temperature, humidity or surface smoothness. The system can use equipment such as temperature and humidity sensors or infrared thermal imaging technology to obtain the wetness of the road surface in real time and determine whether there is a risk of falling. In addition, through video monitoring and image processing technology, the system can detect surface changes such as water stains and oil stains, and then evaluate road conditions.

[0039] Through the above steps, we can select appropriate monitoring strategies to collect detailed features according to the types and characteristics of obstacles, so as to obtain comprehensive information about obstacles. This information helps to judge the safety of the target user's walking environment in real time and provide accurate data support for subsequent fall risk warnings.

[0040] Further, to obtain environmental perception information, step P20 of the embodiment of the present application further includes:

[0041] P24: Perform time state perception on the target environment to obtain real-time time; P25: Perform location state perception on the target environment to obtain real-time location; P26: Perform spatial state perception on the target environment to obtain real-time spatial data, and the real-time spatial data includes real-time spatial temperature, real-time spatial humidity and real-time spatial wind force; P27: The real-time time, the real-time location, the real-time spatial temperature, the real-time spatial humidity and the real-time spatial wind force together constitute the environmental perception information.

[0042] It should be understood that the purpose of obtaining environmental perception information is to fully understand the state of the target user's environment and then assess the potential risk of falling. Environmental perception information not only includes the perception of the external road environment, but also involves multi-dimensional perception data of time, place and space. This information can help the system accurately judge the environmental conditions of the user.

[0043] First, the time state of the target environment is sensed to obtain the real-time time. Time state perception is mainly to record the moment information of the user's environment. Time information is crucial to determine the impact of environmental changes. For example, as time changes, factors such as lighting conditions, climate conditions, and slippery roads may change, which in turn affects the target user's risk of falling. The system obtains accurate real-time time through the system clock or time sensor, and refers to this information in subsequent analysis.

[0044] Next, the real-time location information is obtained by sensing the location status of the target environment. Location perception is the real-time tracking of the target user's geographic location through technologies such as GPS positioning systems, Bluetooth beacons, or Wi-Fi positioning. Real-time location information is very important for determining the nature of the target user's environment. For example, if the user is in a dimly lit place (such as corridors, stairs, etc.) or in a more complex environment (such as multiple obstacles or changing road conditions), these factors may increase the risk of falling. Therefore, real-time location information helps the system make a more accurate fall risk assessment.

[0045] Next, the spatial state of the target environment is sensed and a series of real-time spatial data is obtained. These spatial data include: real-time spatial temperature, that is, the temperature condition of the environment. Changes in temperature may affect the friction of the road surface (for example, ice may form at low temperatures), which in turn affects the walking stability of the user. Real-time spatial humidity, humidity is an indicator of the degree of humidity in the air. In a humid environment, especially on rainy days or in places with a lot of water and oil stains, the ground is easy to slip, increasing the risk of falling. Real-time spatial wind force, wind force information helps assess the stability of the external environment. Strong winds may affect the balance of users, especially the elderly or patients with limited mobility. These spatial perception data are collected in real time through devices such as temperature and humidity sensors, meteorological sensors, and anemometers. By obtaining these data, the system can evaluate in real time the impact of environmental conditions on the user's walking, especially in complex weather or environments, whether it may cause slips or falls.

[0046] Finally, the real-time time, the real-time location, the real-time spatial temperature, the real-time spatial humidity and the real-time spatial wind are integrated to generate environmental perception information. These environmental perception information will provide basic data for subsequent fall risk analysis. In actual applications, the system combines these data with the user's physiological state and external obstacle information for a comprehensive assessment. For example, if the system detects that the user's environment has a low temperature, high humidity, and strong wind, the system can combine these environmental factors to assess the potential risk of falling and issue an early warning. This process not only enhances the ability of environmental factors to predict falls, but also provides personalized early warning strategies based on real-time environmental conditions.

[0047] P30: Establishing a fall database, and extracting a first fall data set and a first non-fall data set from the fall database.

[0048] Specifically, by collecting a large amount of historical data, a fall database containing various types of fall events is constructed. The database includes various types of fall event data, covering falls in different environments, different user groups and different situations. Data sources may include: physiological monitoring data of users in different environments (such as heart rate, blood pressure, hemodynamics, etc.), environmental perception data (such as temperature and humidity, road conditions, lighting conditions, etc.), user activity data (such as walking speed, gait changes, step stability, etc.), user fall event records (such as the specific time, place and status of the fall). By integrating these data into a unified database, various patterns and causes of falls can be comprehensively analyzed and understood.

[0049] After the fall database is constructed, the first fall data set and the first non-fall data set are extracted from it. These two data sets represent typical samples of fall events and non-fall events, respectively, and will serve as the basic data for subsequent machine learning and pattern recognition.

[0050] Among them, the first fall data set includes all known fall events and their related features. These data record the various factors that lead to falls by analyzing the user's physiological monitoring data, environmental conditions and behavioral patterns. For example, it may include falls due to abnormal hemodynamics, falls when walking on slippery surfaces, or falls caused by uneven roads. The data of each fall event will record in detail the time, location, environmental conditions at the time, and user health data.

[0051] The first non-fall dataset is the non-fall dataset as opposed to the fall dataset, which records the situations where the user did not fall under the same environment and conditions. These data also contain the user's physiological data, environmental information, and behavioral data. By comparing these two sets of data, it is possible to find out which features are more significant when a fall occurs, thereby helping to build a fall prediction model.

[0052] It should be noted that in the process of extracting the fall and non-fall data sets, the original data needs to be cleaned and preprocessed, including removing noise data, filling missing values, standardizing data, etc. For example, for hemodynamic data, it is necessary to ensure the accuracy of the data and remove outliers and extreme data points to ensure the training effect of the machine learning model. In addition, environmental data also needs to be preprocessed, for example, humidity, temperature and other parameters need to be standardized to make them suitable for comparison under different environmental conditions.

[0053] After extracting the fall and non-fall data sets, feature selection is required, that is, selecting features that are highly correlated with the risk of falling from a large amount of raw data. These features can be the user's physiological state (such as blood flow velocity, blood pressure fluctuations), environmental factors (such as slippery roads, temperature and humidity), behavioral characteristics (such as gait, walking speed), etc. Through feature selection, redundant information can be reduced and the focus can be placed on the factors that are most likely to affect falls.

[0054] In addition, data labeling is a key step in machine learning. In this process, it is necessary to accurately label each piece of data as to whether a fall has occurred based on historical records and expert experience. Through data labeling, the system can learn the relationship between different factors and the occurrence of falls, thereby training a reliable fall prediction model.

[0055] P40: Perform supervised machine learning on the first fall data set and the first non-fall data set, and train to obtain an intelligent fall discriminator.

[0056] Furthermore, step P40 of the embodiment of the present application also includes:

[0057] P41: Perform sample enhancement preprocessing on the first fall data set to obtain a first target fall data set; P42: Perform supervised machine learning on the first target fall data set and the first non-fall data set based on the support vector machine principle to obtain the intelligent fall discriminator; wherein, the first target fall data set includes first user-side information, first road perception information and first environment perception information, and the first target fall data set has a fall label, the first non-fall data set includes second user-side information, second road perception information and second environment perception information, and the first non-fall data set has a non-fall label.

[0058] It should be understood that the first fall data set and the first non-fall data set are trained by a supervised machine learning method to obtain an intelligent fall discriminator, so that the system can determine whether the user has a fall risk based on real-time data.

[0059] Specifically, the first fall data set needs to be preprocessed with sample enhancement. Sample enhancement is a common technique in machine learning, especially in the case of an unbalanced data set, it can help improve the accuracy and robustness of the model by generating more training samples. In the construction of the fall data set, there are usually fewer fall events and more non-fall events, which leads to data imbalance and may affect the training effect of the model. The sample enhancement methods include: data transformation, that is, appropriate transformation of existing fall data, such as time translation, rotation, scaling and other operations on sensor data to simulate different environments and behavioral scenarios. Data synthesis, that is, generating new fall data samples by adding noise, interpolation and other methods, so that the model can be trained in a variety of scenarios. Through these methods, a more balanced first target fall data set is generated, which contains richer fall event information, and each data sample has a corresponding fall label, that is, indicating whether the event caused a fall.

[0060] Furthermore, the principle of support vector machine (SVM) is used to perform supervised machine learning on the enhanced first target fall data set and the first non-fall data set, thereby obtaining an intelligent fall discriminator. Support vector machine is a powerful classification algorithm, which is particularly suitable for dealing with binary classification problems. Its basic principle is to map the data into a high-dimensional space, in which a hyperplane is constructed to distinguish fall events from non-fall events. By maximizing the interval between the two types of data points, SVM can find an optimal decision boundary, so that the classification result has high accuracy and robustness.

[0061] Among them, the first target fall data set contains a variety of information related to falls, including first user-side information (such as the user's physiological monitoring data, movement status, etc.), first road perception information (such as road status, obstacle information, etc.), and first environment perception information (such as temperature and humidity, lighting conditions, etc.). This information can reflect the risk of falls that users may face in different environments. The first non-fall data set corresponds to the fall data set. The non-fall data set contains the user's second user-side information, second road perception information, and second environment perception information, and each data sample is marked as no fall has occurred. The existence of these non-fall data helps to provide positive and negative samples for the model, helping the system learn to distinguish between fall and non-fall states.

[0062] During the training process, SVM calculates the optimal hyperplane by inputting the features of the first target fall data set and the first non-fall data set into the model, so that the data points are accurately classified. Each input sample (whether it is a fall or not) is mapped to the feature space, and SVM finds the classification surface that can best distinguish between falls and non-falls based on the boundary maximization strategy.

[0063] In this way, the trained intelligent fall discriminator will be able to judge whether the user is at risk of falling based on the real-time input user-side information, road perception information, and environmental perception information. The discriminator can evaluate the user's behavior and environmental conditions based on the rules and features learned during the training process, and give a real-time prediction result of the fall.

[0064] P50: The intelligent fall discriminator performs a fall analysis on the target user side information, the road perception information and the environment perception information to obtain a target discrimination result.

[0065] Specifically, a trained intelligent fall discriminator is used to perform fall analysis on the target user side information, road perception information and environmental perception information collected in real time, thereby obtaining a target discrimination result.

[0066] First, the target user-side information collected in real time is input into the fall discriminator. The target user-side information includes the user's physiological state data, behavioral action data, etc. For example, the user's heart rate, blood pressure, body position changes, gait characteristics and other physiological information can reflect the user's current physical health status. If the user's hemodynamics is abnormal (such as a sharp drop in blood pressure or abnormal heart rate, etc.), it may indicate a potential risk of falling. In addition, the user's body position changes and movement status (such as walking speed, posture, etc.) can also provide important clues. When the user shows an unstable gait or sudden changes in movement, the risk of falling will also increase.

[0067] Next, the road perception information is used as input. Road perception information includes the physical environment of the road on which the user is located, such as the road state (e.g., flat, rugged), the distribution of obstacles (e.g., stones on the road, three-dimensional obstacles), and slippery areas on the road (e.g., water stains, oil stains, etc.). This information can help determine whether the user is in an environment where it is easy to slip. For example, if there are slippery areas on the road, especially water stains or oil stains, these environmental factors will increase the risk of falling. Therefore, judging the safety of the road and the impact of obstacles is an important part of preventing falls.

[0068] Furthermore, environmental perception information is used as input, namely the time, location, and spatial state of the user's environment. This includes: real-time time, that is, the risk of falling may change in different time periods. For example, at night, the light is dim, and the user may find it difficult to perceive obstacles around them, thereby increasing the probability of falling. Real-time location, depending on the specific location of the user, such as whether it is indoors or outdoors, or whether it is in a complex environment (such as stairs, slippery external roads, etc.), all of which affect the risk of falling. Real-time spatial data, including spatial temperature, humidity, wind force, etc., these factors will affect the user's activity status. High temperature or high humidity environment may cause physical discomfort, while strong winds or extremely cold weather may make it more difficult for users to walk and easily lose balance.

[0069] The above three types of information are input into the intelligent fall discriminator, and a comprehensive analysis is performed based on the rules learned in the model to evaluate whether the user is at risk of falling. The intelligent fall discriminator processes the user's real-time data and combines it with the pre-trained classification model to calculate the probability of falling and compare it with the preset risk threshold. If the analysis results show that the user is at risk of falling, a fall discrimination result is output, indicating that the user has a high probability of falling. If the analysis results show that the user's state is normal and the environmental conditions are safe, the system outputs a non-fall discrimination result, indicating that there is no obvious risk of falling at present.

[0070] P60: When the target identification result meets the predetermined warning constraint, the voice education module is activated to issue a fall voice warning to the target user.

[0071] Optionally, when the target discrimination result meets the predetermined warning constraints, the system will activate the voice education module and issue a fall voice warning to the target user. This step is a key link in the fall warning system, which ensures that the user can receive a warning in time when there is a risk of falling, so as to take preventive measures and reduce the occurrence of actual falls.

[0072] First, the target judgment result (i.e., whether there is a risk of falling) obtained in the previous step is compared with the preset warning constraints. The preset warning constraints refer to the trigger conditions set in the system, which usually include the probability threshold of the risk of falling, time constraints, environmental factors, etc. For example, when the judgment result shows that the user's risk of falling exceeds a certain set threshold (such as the risk of falling exceeds 80%), or the user is in a more dangerous environment (such as on a slippery ground or in a dimly lit area), the warning trigger condition is met and the system will activate the subsequent warning process.

[0073] When the system determines that the target user has a high risk of falling and meets the preset warning conditions, the system will enter the warning state and the voice education module will be activated. The main function of this module is to warn the target user through voice prompts and provide relevant suggestions for anti-fall measures to help users reduce the risk of falling.

[0074] By combining the target discrimination results with the predetermined warning constraints, the voice reminder function in the intelligent fall warning system is realized, ensuring that users are given timely warnings and protective guidance at critical moments, thereby effectively reducing the incidence of fall accidents.

[0075] In summary, the embodiments of the present application have at least the following technical effects:

[0076] This application obtains the target user's physical information through dynamic body monitoring, collects road and environmental perception data, builds a fall database and extracts fall and non-fall data sets, trains an intelligent fall discriminator, combines the target user's physical state, road and environmental information to perform fall risk analysis, and generates a discrimination result; when the discrimination result meets the warning conditions, activates the voice education module and issues a fall warning to the user in a timely manner.

[0077] The technical effect of accurately predicting and promptly feeding back users' fall risks through multi-dimensional data fusion analysis has been achieved, thereby achieving effective prevention and intervention of fall incidents.

[0078] Embodiment 2 is based on the same inventive concept as the anti-fall method in the above embodiment. Figure 2 As shown, the present application provides a fall prevention system, and the system and method embodiments in the present application are based on the same inventive concept. The system includes:

[0079] A dynamic body monitoring module 11, which is used to perform dynamic body monitoring on a target user to obtain target user side information;

[0080] An external side perception information acquisition module 12, the external side perception information acquisition module 12 is used to obtain target external side perception information of the target user, the target external side perception information includes road perception information and environment perception information;

[0081] A fall data extraction module 13, the fall data extraction module 13 is used to build a fall database and extract a first fall data set and a first non-fall data set in the fall database;

[0082] An intelligent fall discriminator training module 14, wherein the intelligent fall discriminator training module 14 is used to perform supervised machine learning on the first fall data set and the first non-fall data set, and train to obtain an intelligent fall discriminator;

[0083] A fall analysis and discrimination module 15, wherein the fall analysis and discrimination module 15 is used to perform a fall analysis on the target user side information, the road perception information and the environment perception information through the intelligent fall discriminator to obtain a target discrimination result;

[0084] The fall voice warning module 16 is used to activate the voice education module to issue a fall voice warning to the target user when the target discrimination result meets the predetermined warning constraint.

[0085] Furthermore, the dynamic body monitoring module 11 is also used to perform the following steps:

[0086] Reading predetermined hemodynamic characteristics; traversing and extracting the target body monitoring information of the target user based on the predetermined hemodynamic characteristics to obtain the target user side information; wherein the predetermined hemodynamic characteristics include blood flow, blood flow velocity, blood pressure and vascular resistance.

[0087] Furthermore, the external side perception information acquisition module 12 is also used to perform the following steps:

[0088] Acquire a target external scene in which the target user is located, wherein the target external scene includes a target road, and the target road includes a first obstacle; retrieve an obstacle monitoring strategy to collect multi-dimensional obstacle features of the first obstacle to obtain first obstacle feature information; and construct the road perception information based on the first obstacle feature information.

[0089] Furthermore, the external side perception information acquisition module 12 is also used to perform the following steps:

[0090] Step A: Determine whether the first obstacle belongs to Class I obstacle; Step B1: If it does, obtain the first obstacle feature information of the first obstacle based on the Class I obstacle monitoring strategy; Step B2: If it does not, obtain the first obstacle feature information of the first obstacle based on the Class II obstacle monitoring strategy.

[0091] Furthermore, the external side perception information acquisition module 12 is also used to perform the following steps:

[0092] The target environment is perceived in time to obtain real-time time; the target environment is perceived in location to obtain real-time location; the target environment is perceived in space to obtain real-time space data, wherein the real-time space data includes real-time space temperature, real-time space humidity and real-time space wind force; the real-time time, the real-time location, the real-time space temperature, the real-time space humidity and the real-time space wind force together constitute the environmental perception information.

[0093] Furthermore, the intelligent fall discriminator training module 14 is also used to perform the following steps:

[0094] The first fall data set is subjected to sample enhancement preprocessing to obtain a first target fall data set; the first target fall data set and the first non-fall data set are subjected to supervised machine learning based on the support vector machine principle to obtain the intelligent fall discriminator; wherein, the first target fall data set includes first user-side information, first road perception information and first environment perception information, and the first target fall data set has a fall label, the first non-fall data set includes second user-side information, second road perception information and second environment perception information, and the first non-fall data set has a non-fall label.

[0095] Embodiment 3, based on the same inventive concept as the anti-fall method in the above embodiment, the present application provides an anti-fall anklet. The anti-fall anklet comprises: a body, the body is equipped with an intelligent monitoring component and a voice education device, and the intelligent monitoring component comprises a user-side monitoring component and an external-side monitoring component.

[0096] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0097] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0098] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A method for preventing falls, characterized in that: include: Perform dynamic body monitoring on the target user to obtain target user side information; Acquire target external side perception information of the target user, wherein the target external side perception information includes road perception information and environment perception information; Establishing a fall database, and extracting a first fall data set and a first non-fall data set from the fall database; Performing supervised machine learning on the first fall data set and the first non-fall data set, and training to obtain an intelligent fall discriminator; Performing a fall analysis on the target user side information, the road perception information and the environment perception information by the intelligent fall discriminator to obtain a target discrimination result; When the target identification result meets the predetermined warning constraint, the voice education module is activated to issue a fall voice warning to the target user.

2. The fall prevention method according to claim 1, characterized in that: Dynamic body monitoring of the target user is performed to obtain target user side information, including: reading predetermined hemodynamic characteristics; Based on the predetermined hemodynamic characteristics, the target body monitoring information of the target user is traversed and extracted to obtain the target user side information; Wherein, the predetermined hemodynamic characteristics include blood flow, blood flow velocity, blood pressure and vascular resistance.

3. The fall prevention method according to claim 1, characterized in that: Acquiring target external side perception information of the target user, wherein the target external side perception information includes road perception information and environment perception information, including: Acquire a target external scene where the target user is located, the target external scene includes a target road, and the target road includes a first obstacle; Retrieving an obstacle monitoring strategy to collect multi-dimensional obstacle features of the first obstacle to obtain first obstacle feature information; The road perception information is constructed based on the first obstacle feature information.

4. The fall prevention method according to claim 3, characterized in that: The obstacle monitoring strategy includes a first-class obstacle monitoring strategy and a second-class obstacle monitoring strategy. The obstacle monitoring strategy is called to collect multi-dimensional obstacle features of the first obstacle to obtain first obstacle feature information, including: Step A: determining whether the first obstacle belongs to a type of obstacle; Step B1: If yes, obtaining the first obstacle feature information of the first obstacle based on the first type of obstacle monitoring strategy; Step B2: If it does not belong to, obtain the first obstacle feature information of the first obstacle based on the second type obstacle monitoring strategy.

5. The fall prevention method according to claim 3, characterized in that: The target external scene also includes the target environment, including: Performing time state perception on the target environment to obtain real-time time; Performing location state perception on the target environment to obtain a real-time location; Performing spatial state perception on the target environment to obtain real-time spatial data, wherein the real-time spatial data includes real-time spatial temperature, real-time spatial humidity and real-time spatial wind force; The real-time time, the real-time location, the real-time spatial temperature, the real-time spatial humidity and the real-time spatial wind force together constitute the environmental perception information.

6. The fall prevention method according to claim 1, characterized in that: Performing supervised machine learning on the first fall data set and the first non-fall data set, and training to obtain an intelligent fall discriminator, including: Performing sample enhancement preprocessing on the first fall data set to obtain a first target fall data set; Based on the support vector machine principle, supervised machine learning is performed on the first target fall data set and the first non-fall data set to obtain the intelligent fall discriminator; Among them, the first target fall data set includes first user-side information, first road perception information and first environment perception information, and the first target fall data set has a fall label, the first non-fall data set includes second user-side information, second road perception information and second environment perception information, and the first non-fall data set has a non-fall label.

7. A fall prevention system, characterized in that: The fall prevention system is used to perform the method according to any one of claims 1 to 6, comprising: A dynamic body monitoring module, which is used to perform dynamic body monitoring on a target user to obtain target user side information; An external side perception information acquisition module, the external side perception information acquisition module is used to obtain target external side perception information of the target user, the target external side perception information includes road perception information and environment perception information; A fall data extraction module, the fall data extraction module is used to build a fall database and extract a first fall data set and a first non-fall data set in the fall database; An intelligent fall discriminator training module, wherein the intelligent fall discriminator training module is used to perform supervised machine learning on the first fall data set and the first non-fall data set, and train to obtain an intelligent fall discriminator; A fall analysis and discrimination module, wherein the fall analysis and discrimination module is used to perform a fall analysis on the target user side information, the road perception information and the environment perception information through the intelligent fall discriminator to obtain a target discrimination result; A fall voice warning module is used to activate the voice education module to provide a fall voice warning to the target user when the target discrimination result meets the predetermined warning constraint.

8. A fall prevention anklet, characterized in that: The anti-fall anklet is used to execute the method described in any one of claims 1 to 6. The anti-fall anklet includes a main body, on which is mounted an intelligent monitoring component and a voice education device, and the intelligent monitoring component includes a user-side monitoring component and an external-side monitoring component.

Images