Intelligent anti-falling shoes

Abstract

The invention belongs to the technical field of intelligent shoes, and particularly relates to an intelligent anti-falling shoe. The system is composed of a setting module, a recognition module, a reminding and judging module, an uploading and storing module and a positioning and calling module, the setting module inputs basic information of a shoe wearer, and activity areas, activity time and emergency contacts are set in a personalized mode; the recognition module is integrated with a high-precision sensor, accurately recognizes road surface accumulated water, obstacles and road surface unevenness, and automatically sends a prompt to a medical worker when a shoe wearer exceeds a preset range in combination with a set activity area and activity time; by integrating a plurality of module structures, the system has multiple functions of environment recognition, gait analysis, emergency call and the like, and comprehensive protection from prevention to emergency treatment is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent shoes, and in particular relates to an intelligent anti-fall shoe. Background Art

[0002] As the global population ages, falls have become a major health threat to the elderly and patients with acute and chronic diseases. Falls can not only cause serious physical injuries, such as fractures and craniocerebral injuries, but also often lead to a series of complications, greatly increasing the medical burden and social costs. In addition, falls also have a profound impact on the psychology and quality of life of the elderly. In response to this situation, there are already some anti-fall auxiliary devices on the market, such as smart bracelets and emergency call devices, but these devices often have single functions and lack comprehensive fall prevention and immediate response mechanisms.

[0003] The Chinese patent "Anti-fall shoes, fall warning methods, devices and computer-readable storage media" with publication number CN111011974A proposes that anti-fall shoes are equipped with a V-shaped fixed support at the back cover of the shoe upper, which can support the foot and prevent falls, and can effectively correct the walking posture of people with abnormal ankles; obtain the sole pressure information and the shoe body inclination angle information, and send a warning when it is determined that the pressure information meets the preset pressure condition and the shoe body inclination angle information meets the preset inclination condition.

[0004] However, the above technical solutions fail to effectively combine multi-dimensional protection such as environmental recognition, gait analysis, and automatic call for help in emergency situations, and are difficult to meet the needs of comprehensive monitoring of people at high risk of falling in scenarios such as homes and hospitals. Therefore, the present invention provides a smart anti-fall shoe that can achieve comprehensive protection for people at high risk of falling. Summary of the invention

[0005] The present invention provides an intelligent anti-fall shoe, which is used to solve the technical problems that the existing anti-fall equipment lacks a comprehensive fall prevention and immediate response mechanism, and cannot meet the comprehensive monitoring of people at high risk of falling in home and hospital scenarios.

[0006] In order to achieve the above-mentioned purpose, the technical scheme adopted by the present invention is as follows: an intelligent anti-fall shoe, composed of a setting module, an identification module, a reminder and judgment module, an upload and save module, and a positioning and calling for help module. The setting module inputs the basic information of the wearer and personalizes the activity area, activity time and emergency contact; the identification module integrates high-precision sensors to accurately identify road surface water, obstacles and uneven road conditions, and timely feedback to the wearer and the monitoring end. At the same time, combined with the set activity area and activity time, when the wearer exceeds the preset range, it automatically sends a reminder to medical staff; the reminder and judgment module monitors the load-bearing force distribution and gait characteristics of both feet, uses a neural network algorithm to intelligently evaluate the fatigue of both legs and the stability of gait, distinguishes between temporary load-bearing force changes caused by normal physiological activities and the real risk of falling, and effectively reduces false alarms. The upload and save module stores and uploads the wearer's activity video and walking data. The positioning and calling for help module can immediately flash a red light to attract attention and automatically dial a preset emergency contact number when the wearer falls.

[0007] Furthermore, the basic information set by the setting module establishes a connection with the family members' mobile phones and hospital monitoring equipment to achieve real-time monitoring.

[0008] Furthermore, the recognition module uses sensors to collect images of road waterlogging, obstacles, and uneven road surfaces, and inputs them into the YOLO-v8n-CSC network for recognition to obtain road waterlogging, obstacles, and uneven road surfaces.

[0009] Furthermore, the reminder and judgment module distinguishes between temporary changes in load-bearing capacity caused by normal physiological activities and real fall risks by inputting feature data output by the YOLO-v8n-CSC network into a logistic regression judgment model.

[0010] Compared with the prior art, the present invention has the following beneficial effects:

[0011] (1) The anti-fall shoes of the present invention integrate multiple module structures and have multiple functions such as environmental recognition, gait analysis, and emergency calls, thereby realizing all-round protection from prevention to emergency treatment; and can realize instant response, through real-time monitoring and automatic calls for help, greatly shortening the rescue response time and effectively reducing the damage caused by falls.

[0012] (2) The present invention utilizes the YOLO-v8n-CSC network model and the logistic regression model to improve the accuracy of fall recognition and control the false alarm rate, reduce unnecessary interference, and the real-time recording and storage function provides data support for subsequent analysis of fall events. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1This is a structural diagram of an intelligent anti-fall shoe of the present invention. DETAILED DESCRIPTION

[0014] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0015] like Figure 1 As shown, a smart anti-fall shoe consists of a setting module, an identification module, a reminder and judgment module, an upload and save module, and a positioning and calling module. The setting module inputs the basic information of the wearer and personalizes the activity area, activity time, and emergency contact. The identification module integrates high-precision sensors to accurately identify road water, obstacles, and uneven road conditions, and promptly feeds back to the wearer and the monitoring end. At the same time, combined with the set activity area and activity time, when the wearer exceeds the preset range, it automatically sends a reminder to medical staff; the reminder and judgment module monitors the load-bearing force distribution and gait characteristics of both feet, uses a neural network algorithm to intelligently evaluate leg fatigue and gait stability, distinguishes temporary load-bearing force changes caused by normal physiological activities from real fall risks, and effectively reduces false alarms. The upload and save module stores and uploads the wearer's activity video and walking data. The positioning and calling module can immediately flash a red light to attract attention and automatically dial a preset emergency contact number when the wearer falls.

[0016] Setting module: Input basic information through simple operation, and establish connection with family members' mobile phones and hospital monitoring equipment to achieve real-time monitoring. It allows personalized setting of activity areas, activity times and emergency contacts to ensure the accuracy and timeliness of information.

[0017] Recognition module: Integrates high-precision sensors to collect images of road waterlogging, obstacles, and uneven road surfaces, and promptly feeds back to the wearer and the monitoring end. At the same time, combined with the set activity area and time, when the wearer exceeds the preset range, an alert is automatically sent to medical staff to ensure safety. The collected images are trained using the YOLO-v8n-CSC network. During the training phase, the YOLO-v8n-CSC network is used for training and the model weights are iteratively optimized. In the recognition phase, the collected images (including road waterlogging, obstacles, and uneven road surfaces) are first cropped according to fixed areas to reduce the amount of model inference data; then, according to the model downsampling depth, the input image is adjusted to the target size; finally, it is input into the YOLO v8n-CSC network loaded with the optimal training weights for recognition and detection to obtain road waterlogging, obstacles, and uneven road surfaces.

[0018] The YOLO-v8n-CSC network is improved based on the YOLO v8n model. The overall network adopts the feature pyramid network (FPN) and path aggregation network (PAN) architecture. In the above architecture, the function block Startblock is set to use 7*1 deep convolution to complete the input image feature extraction, and the output is efficiently standardized after normalization BN (Batchnormalization); ReLU6 is used for activation; then element multiplication is performed on the two outputs to achieve global interaction. At the same time, feature dimensionality reduction and extraction are achieved through 1x1 convolution and 7x1 deep convolution.

[0019] The feature extraction data is F, the feature dimension is D, and the weights of the i-th and j-th channels in the first star operation are w. i (1,1), w j (1,2), after the bias is introduced, the overall weights are W(1,1) and W(1,2), then the output of the mth star operation is O l ,

[0020]

[0021] The output results are divided into areas set under different conditions (including road surface water, obstacles, and uneven road surface).

[0022] The reminder and judgment module inputs the feature data output by the YOLO-v8n-CSC network into the logistic regression judgment model to distinguish between temporary load-bearing changes caused by normal physiological activities and real fall risks, and can evaluate leg fatigue and gait stability. Once uneven load-bearing or abnormal gait is detected, a rest reminder is immediately issued to prevent falls. l , the bipedal load-bearing force distribution and gait characteristics are input into the logistic regression judgment model to intelligently distinguish the temporary load-bearing force change B caused by normal physiological activities from the real fall risk M.

[0023] The specific process of the logistic regression judgment model is as follows:

[0024] (1) Output region O of the YOLO-v8n-CSC network l 、Basic foot weight distribution a i , gait characteristics f i , Leg fatigue P i The four feature data are processed by outliers removal, normalization and missing value filling. After the above data processing, a new data feature quantity S,x is formed. i ∈S;

[0025] For each feature x i (The above output area O through the YOLO-v8n-CSC network l 、Basic foot weight distribution a i , gait characteristics f i , Leg fatigue P i , x i =λ Q Q i +λ a ·a i +λ f ·f i +λ P ·P i , i∈{1,2,3...,N},), where λ xx is the normalized weight value. Calculate its average value The difference χ i =x i -u, and with standard deviation If the difference χ i If the value of -s exceeds the threshold m=3s (3 times the standard deviation), the data point is considered an outlier and is removed or replaced.

[0026] Through the normalization algorithm, the data is converted into dimensionless values ​​between 0 and 1, which is convenient for subsequent algorithm processing and analysis. Minimum-maximum normalization formula: Among them, x is the normalized data, x min and x max are the minimum and maximum values ​​in the data respectively. After normalization, the value range of the data is between 0 and 1.

[0027] Missing value filling: Using linear interpolation, for missing values ​​x m , whose two non-missing values ​​are x a and x b , and a<m<b, then After the above data processing, a new data feature quantity S,x is formed. i ∈S.

[0028] (2) When extracting the feature quantity S,x i ∈S, use it as a training sample to build a model of logical judgment regression, with n feature quantities [x1, x2, ..., x n ], that is, input feature quantity, 1 bias term b and corresponding weights [ω1,ω2,...,ω n ], the model of logistic regression is expressed as: Wherein, logSigmoid is a logarithmic domain activation function used to convert the output into a probability value; σ is the dot product of the weight vector and the feature vector;

[0029] (3) Set a threshold K. The Sigmoid function maps the output feature quantity of the temporary bearing force change B and the real fall risk M to the interval [0, 1]. According to different values, it can intelligently distinguish the temporary bearing force change caused by normal physiological activities from the real fall risk. If H > K, it indicates a real fall risk. At this time, a reminder needs to be automatically sent to medical staff to ensure the safety of the patient. If H < K, it is the temporary bearing force change caused by normal physiological activities.

[0030] Upload and save module: Similar to the principle of a car driving recorder, the intelligent shoe has built-in storage and upload functions, which can record and save the activity videos and key data of the shoe wearer in real time. Once a fall event occurs, it can quickly restore the scene situation, providing a basis for medical treatment and liability determination.

[0031] Positioning and calling for help module: Integrated with a GPS positioning system and an emergency calling mechanism. When a fall behavior is detected, the intelligent shoe immediately flashes a red light to attract the attention of the surrounding people, automatically dials the preset emergency contact number, and uploads the current location information at the same time to ensure quick response and rescue.

[0032] The anti-fall shoe of the present invention integrates multiple module structures, has multiple functions such as environment recognition, gait analysis, and emergency calling for help, realizing all-round protection from prevention to emergency handling; and can achieve instant response. Through real-time monitoring and automatic calling for help, the rescue response time is greatly shortened, and the injury caused by falls is effectively reduced. Using a neural network model, the accuracy of fall recognition and the control of false alarm rate are improved, unnecessary interference is reduced, and the real-time recording and saving function provides data support for the subsequent analysis of fall events.

[0033] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. Without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A smart anti-fall shoe, characterized in that: It consists of a setting module, an identification module, a reminder and judgment module, an upload and save module, and a positioning and emergency call module. The setting module inputs the basic information of the wearer and personalizes the activity area, activity time, and emergency contact. The identification module integrates high-precision sensors to accurately identify road surface water, obstacles, and uneven road conditions, and promptly feeds back to the wearer and the monitoring end. At the same time, combined with the set activity area and activity time, when the wearer exceeds the preset range, it automatically sends a reminder to medical staff. The reminder and judgment module monitors the load-bearing force distribution and gait characteristics of both feet, uses a neural network algorithm to intelligently evaluate leg fatigue and gait stability, distinguishes temporary load-bearing force changes caused by normal physiological activities from real fall risks, and effectively reduces false alarms. The upload and storage module stores and uploads the wearer's activity video and walking data. The positioning and emergency call module can immediately flash a red light to attract attention and automatically call a preset emergency contact number when the wearer falls.

2. The intelligent anti-fall shoe according to claim 1, characterized in that: The basic information set by the setting module establishes a connection with the family member's mobile phone and the hospital monitoring equipment to achieve real-time monitoring.

3. The intelligent anti-fall shoe according to claim 1, characterized in that: The recognition module uses sensors to collect images of road waterlogging, obstacles, and uneven road surfaces, and inputs them into the YOLO-v8n-CSC network for recognition to obtain road waterlogging, obstacles, and uneven road surfaces.

4. The intelligent anti-fall shoe according to claim 1, characterized in that: The reminder and judgment module inputs the feature data output by the YOLO-v8n-CSC network into a logistic regression judgment model to distinguish between temporary changes in load-bearing capacity caused by normal physiological activities and real risks of falling.

5. The intelligent anti-fall shoe according to claim 4, characterized in that: The YOLO-v8n-CSC network model adopts a feature pyramid network and a path aggregation network architecture, and sets a functional block to use a 7*1 deep convolution to complete feature extraction of input collected images, wherein the images include images of three conditions: road surface water, obstacles, and uneven road surface.

6. The intelligent anti-fall shoe according to claim 5, characterized in that: Then the output is standardized through normalized BN; ReLU6 is used for activation; then element-wise multiplication is performed on the two outputs to achieve global interaction. At the same time, feature dimensionality reduction and feature extraction are achieved through 1x1 convolution and 7x1 depth convolution.

7. The intelligent anti-fall shoe according to claim 6, characterized in that: The output results are divided into areas set under different conditions including road waterlogging, obstacles, and uneven road surfaces.

8. The intelligent anti-fall shoe according to claim 7, characterized in that: The upload and save module has storage and upload functions, and can record and save the activity video of the wearer in real time. Once a fall occurs, the scene can be quickly restored.

9. The intelligent anti-fall shoe according to claim 1, characterized in that: The positioning and rescue module integrates the GPS positioning system and the emergency rescue mechanism. When a fall is detected, the smart shoe immediately flashes a red light to attract attention from the surroundings, and automatically dials the preset emergency contact number, while uploading the current location information to ensure rapid response and rescue.

Images