Patient falling monitoring system based on behavior feature recognition

Abstract

The invention discloses a patient falling monitoring system based on behavior feature recognition, and belongs to the technical field of falling monitoring, and the system comprises a data collection module which is used for collecting patient behavior real-time data; the data processing module is used for processing the patient behavior real-time data and determining patient behavior characteristic data; the analysis and identification module is used for carrying out prediction analysis and risk identification on the falling behavior of the patient and determining a falling risk identification result of the patient; and the early warning management and control module is used for formulating a patient falling early warning management and control scheme and performing early warning management and control on the falling behavior of the patient. According to the invention, the problems of low patient safety and increased medical cost caused by the fact that the falling behavior of the patient cannot be effectively monitored and identified based on behavior characteristics in the prior art are solved. According to the invention, identification and early warning of the falling behavior of the patient can be realized, and when the falling risk of the patient is identified, an alarm is automatically given and measures are taken to prevent falling, so that the safety of the patient can be improved, and the medical cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of fall monitoring, and in particular to a patient fall monitoring system based on behavior feature recognition. Background Art

[0002] When patients fall, it not only causes physical harm to the patients, but may also lead to waste of medical resources and medical safety issues.

[0003] The Chinese patent with publication number CN113456059A discloses the use of radar to monitor the patient's body, including: one or more radar sensors can be used to monitor the patient in a variety of different environments, the radar sensor can be used to monitor the patient's movement, including movement on the bed and movement around the room, and the patient's position in the bed can be monitored; however, the patent has the following defects: The existing system cannot effectively monitor and identify patients' falls based on behavioral characteristics, nor can it provide timely early warning and control of patients' falls. It cannot effectively prevent patients from falling, resulting in low patient safety and increased medical costs. Summary of the invention

[0004] The purpose of the present invention is to provide a patient fall monitoring system based on behavioral feature recognition, which can realize the recognition and early warning of patient fall behavior. When it is identified that the patient is at risk of falling, an alarm is automatically issued and measures are taken to prevent the fall from occurring, which can improve the safety of the patient and reduce medical costs, solving the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: The patient fall monitoring system based on behavioral feature recognition includes: A data acquisition module is used to collect the patient's movements, postures and gait using sensors to determine the patient's real-time behavior data; The data processing module is used to clean, convert and extract features of the real-time data of patient behavior and determine the patient behavior feature data; The analysis and identification module is used to predict and analyze the patient's falling behavior and identify risks, determine whether the patient has a falling risk, and determine the patient's falling risk identification results; The early warning and control module is used to formulate a patient fall early warning and control plan, to conduct early warning and control of patient fall behaviors, and to prevent falls from occurring.

[0006] Preferably, the data acquisition module comprises: A motion monitoring unit, used to collect patient motion data; Monitor and collect the patient's movements in real time based on sensors to obtain the patient's movement data; A posture monitoring unit, used to collect patient posture data; Monitor and collect the patient's posture in real time based on sensors to obtain the patient's posture data; A gait monitoring unit, used to collect gait data of patients; Monitor and collect the patient's gait in real time based on sensors to obtain the patient's gait data; Among them, the real-time data of patient behavior is determined based on the patient motion data, patient posture data and patient gait data.

[0007] Preferably, the data processing module includes: Data cleaning unit, used to clean the real-time data of patient behavior; Obtain real-time data on patient behavior; Conduct consistency checks on real-time data on patient behavior; According to the data consistency requirements, each parameter in the real-time data of patient behavior is checked one by one to check whether there is inconsistent data in the real-time data of patient behavior that is useless for patient fall monitoring, and the inconsistent data in the real-time data of patient behavior is removed; Check invalid and missing values ​​in real-time patient behavior data; According to the requirements of data validity and completeness, each parameter in the real-time patient behavior data is checked one by one to check whether there are invalid values ​​and missing values ​​in the real-time patient behavior data that are useless for patient fall monitoring. The invalid values ​​and missing values ​​in the real-time patient behavior data are removed to determine the real-time patient behavior data that is useful for patient fall monitoring.

[0008] Preferably, the data processing module further includes: A data conversion unit, used to convert real-time data of patient behavior; Obtain real-time data on patient behavior useful for patient fall monitoring; Converting the real-time patient behavior data useful for patient fall monitoring to unify the formats of the real-time patient behavior data useful for patient fall monitoring; Eliminate the dimensional differences between real-time patient behavior data that are useful for patient fall monitoring and determine standardized real-time patient behavior data; A feature extraction unit, used for extracting features from real-time patient behavior data; Obtain standardized real-time data on patient behavior; Extract features from standardized real-time patient behavior data; Features that can reflect patient falls and are related to patient falls are extracted from standardized real-time patient behavior data to determine patient behavior feature data.

[0009] Preferably, the analysis and identification module includes: A model training unit, used to train a patient fall risk identification model; Collect patient behavior history data based on patient fall monitoring needs based on behavioral feature identification; Define the patient's behavioral history data, define the behavioral characteristics related to the patient's fall, classify the behavioral characteristics, and determine the training set and test set; Select a machine learning framework suitable for patient fall risk identification based on behavioral signature recognition; Based on the training set, the selected machine learning framework suitable for patient fall risk identification based on behavioral feature recognition is trained to determine a patient fall risk identification model based on behavioral feature recognition.

[0010] Preferably, the analysis and identification module further includes: A test optimization unit, used to test and optimize the patient fall risk identification model; Obtain a patient fall risk identification model based on behavioral feature recognition; Based on the test set, the performance of the patient fall risk identification model based on behavioral feature recognition is tested to determine whether the patient fall risk identification model based on behavioral feature recognition can achieve the expected effect; Determine the performance test results of the patient fall risk identification model; According to the performance test results of the patient fall risk identification model, the patient fall risk identification model based on behavioral feature identification was analyzed, and the parameters of the patient fall risk identification model based on behavioral feature identification were adjusted and the structure was optimized. After repeated iterations, the optimal patient fall risk identification model was determined.

[0011] Preferably, the analysis and identification module further includes: Analysis and identification unit, used for risk analysis and identification of patient falling behavior; Obtain the optimal patient fall risk identification model; Deploy the optimal patient fall risk identification model in the actual patient fall monitoring environment; Input patient behavioral characteristic data into the optimal patient fall risk identification model; Based on the optimal patient fall risk identification model, predict and analyze the patient's behavioral characteristic data and identify risks, determine whether the patient is at risk of falling, and determine the patient's fall risk identification results; The patient fall risk identification result is that the patient is at risk of falling or the patient is not at risk of falling.

[0012] Preferably, the early warning control module includes: The plan formulation unit is used to formulate patient fall warning and control plans; Obtain patient fall risk identification results; When a patient is at risk of falling, the patient's behavioral characteristic data is analyzed to develop a personalized patient fall warning and control plan for the patient; Early warning control unit, used for early warning control of patients' falling behavior; Among them, based on the personalized patient fall warning and control plan, the patient's fall behavior is warned and controlled. For the patient's fall behavior, an alarm is automatically issued and measures are taken to prevent the fall from occurring, reminding people around the patient to protect the patient.

[0013] Preferably, the patient fall monitoring system based on behavioral feature recognition further includes: An environmental assessment module is used to obtain environmental data of the monitoring space and evaluate the environmental data to obtain environmental assessment data; A patient assessment module is used to obtain the patient's physical condition data and evaluate the physical condition data to obtain patient assessment data; A model configuration module, used for configuring a model used by the analysis and identification module to perform predictive analysis and risk identification of patient fall behaviors based on environmental assessment data and patient assessment data; Among them, the environmental assessment module performs the following operations: When receiving an environmental assessment request, output a preset environmental simulation construction interface; Receiving a call to a three-dimensional model corresponding to each object on the environment simulation construction interface and constructing an environment simulation model in a construction area of ​​the environment simulation construction interface; Performing risk assessment on each three-dimensional model involved in building the environmental simulation model to obtain first risk assessment data; Based on the preset association library, each three-dimensional model is associated and evaluated to obtain second risk assessment data; Extract features from the environmental simulation model and conduct a comprehensive evaluation based on the preset comprehensive analysis library to obtain the third risk assessment data; The first risk assessment data, the second risk assessment data and the third risk assessment data are integrated to obtain environmental assessment data; Among them, the patient assessment module performs the following operations: When receiving a patient assessment application, output a preset data input interface; Receive basic condition data input in the basic condition input area of ​​the data input interface, case data input in the historical case data input area, and current test data input in the current test data input area; Comprehensively analyze basic condition data, case data and current test data to obtain patient assessment data.

[0014] Preferably, the model configuration module performs the following operations: Performing feature extraction on the first risk assessment data, the second risk assessment data, and the third risk assessment data respectively to obtain a plurality of first feature parameters; Performing feature extraction on the patient assessment data to obtain a plurality of second feature parameters; The plurality of first characteristic parameters and the plurality of second characteristic parameters are sequentially filled into a preset array template to form an analysis data set and the data in the analysis data set are normalized. The normalization processing formula is as follows: In the formula, The first Line The value of the column; is the first Line The value of the column; is the first Line Column values; analyze the data in the data set OK List; The analysis data set is matched with the call data sets corresponding to the models used for predictive analysis and risk identification of patient fall behavior in the pre-configured model call library. The matching is performed by calculating the similarity between the two. The similarity calculation formula is as follows: In the formula, Indicates similarity; The first Line The value of the column; To call the first Line The value of the column; The model associated with the calling data set with the largest similarity and greater than the preset similarity threshold is retrieved.

[0015] Preferably, the patient fall monitoring system based on behavioral feature recognition further includes: Visual monitoring module, used to realize visual monitoring of patients; Among them, the visual monitoring module includes: A scene model building unit, used to build a scene model according to scene data input by a user; A personnel model building module is used to retrieve a corresponding personnel model from a pre-configured personnel model library according to the patient's motion data, posture data and gait data; A positioning module is used to obtain the patient's position in the scene and place the person model into the scene model according to the obtained position to form a visual monitoring model; The real-time risk analysis module is used to analyze the visual monitoring model, identify risk items, and issue corresponding reminders; The real-time risk analysis module analyzes the visual monitoring model and determines risk items, including: Extract various risk objects within the preset area around the personnel model in the visual monitoring model; Retrieve the risk assessment items corresponding to each risk object as items to be screened; record each item to be screened as ; Indicates The first risk risk assessment items; Analyze the patient's motion data, posture data, and gait data to determine the behavior set; Based on the pre-configured association library, determine the association between each behavior in the behavior set and the risk object and retrieve the association coefficient, which is recorded as and ; Indicates Behavior and The first risk The correlation coefficient between the risk assessment items for distance; Indicates Behavior and The first risk The correlation coefficient between the risk assessment items for the angle; Based on the correlation coefficient, the evaluation value between each behavior and each risk assessment item is determined. The calculation formula is as follows: ;in, Indicates The distance between the risk object and the patient; Indicates The angle between the risk object and the patient; Extract risk assessment items whose assessment values ​​are greater than a preset assessment threshold as risk items; The patient's motion data, posture data and gait data are analyzed to determine the behavior set, including: based on a preset behavior analysis library, calling the corresponding behavior set; The behavior set in the behavior analysis library is matched with the first retrieved parameter of the corresponding action data, the second retrieved parameter of the corresponding posture, and the third retrieved parameter of the corresponding gait data in detail by the following formula: In the formula, TD is the matching degree; is the dth action feature parameter after feature extraction of action data; The first call parameter for the dth one; is the jth posture feature parameter after feature extraction of posture data; The second call parameter for the jth one; is the u-th posture feature parameter after feature extraction of gait data; is the uth third call parameter; D is the total number of first call parameters; J is the total number of second call parameters; U is the total number of third call parameters; , , is a pre-configured weight coefficient; A behavior set corresponding to a set of the first calling parameter, the second calling parameter, and the third calling parameter with the greatest matching degree is called.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention utilizes sensors to collect the patient's movements, postures and gait conditions, determines the patient's real-time behavior data, determines the patient's behavior feature data by cleaning, converting and extracting features from the patient's real-time behavior data, performs predictive analysis and risk identification on the patient's behavior feature data based on the optimal patient fall risk identification model, determines whether the patient has a fall risk, determines the patient's fall risk identification result, and formulates a patient fall early warning and control plan to perform early warning and control on the patient's fall behavior, which can realize the identification and early warning of the patient's fall behavior. When it is identified that the patient is at risk of falling, an alarm is automatically issued and measures are taken to prevent the fall from occurring, which can improve the patient's safety and reduce medical costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a module framework diagram of the patient fall monitoring system based on behavior feature recognition of the present invention; Figure 2 The figure is an algorithm flow chart of the patient fall monitoring system based on behavioral feature recognition of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] In order to solve the existing problems of not being able to effectively monitor and identify patient falls based on behavioral characteristics, not being able to conduct timely early warning and control of patient falls, and not being able to effectively prevent patient falls, resulting in low patient safety and increased medical costs, please refer to Figure 1-Figure 2 , this embodiment provides the following technical solutions: The patient fall monitoring system based on behavioral feature recognition includes: a data acquisition module, a data processing module, an analysis and recognition module and an early warning control module.

[0020] Specifically, the patient's movements, postures and gait are collected through the data acquisition module to determine the patient's real-time behavior data; the patient's real-time behavior data is cleaned, converted and feature extracted through the data processing module to determine the patient's behavior feature data; the patient's fall behavior is predictively analyzed and risk identified through the analysis and identification module to determine whether the patient is at risk of falling and determine the patient's fall risk identification result; the patient's fall early warning and control plan is formulated through the early warning and control module to carry out early warning and control of the patient's fall behavior to prevent falls.

[0021] In this embodiment, the data acquisition module includes: A motion monitoring unit, used to collect patient motion data; Monitor and collect the patient's movements in real time based on sensors to obtain the patient's movement data; A posture monitoring unit, used to collect patient posture data; Monitor and collect the patient's posture in real time based on sensors to obtain the patient's posture data; A gait monitoring unit, used to collect gait data of patients; Monitor and collect the patient's gait in real time based on sensors to obtain the patient's gait data; Among them, the real-time data of patient behavior is determined based on the patient motion data, patient posture data and patient gait data.

[0022] It should be noted that by setting infrared sensors, accelerometers, gyroscopes and magnetometers on and around the patient, the patient's movements, postures and gait are collected through the infrared sensors, accelerometers, gyroscopes and magnetometers, thereby determining the patient's real-time behavior data.

[0023] Among them, the infrared sensor is a sensor that uses infrared rays to process data and has the advantages of high sensitivity. The infrared sensor can be used to collect the patient's movements, postures and gait.

[0024] Infrared sensors work on the principle of infrared radiation. They can sense infrared radiation emitted by the human body and convert it into electrical signals for processing and analysis. This type of sensor can detect information such as human movement, position, and posture, thereby monitoring human movements, postures, and gait. Specifically: ‌Motion data‌: Infrared sensors can detect the movement state of the human body, such as movement speed and direction, which is very useful for monitoring the patient's activity; ‌Position data‌: By analyzing the intensity and direction of infrared radiation, the position of the human body in space can be determined, which is very helpful for tracking the patient's position changes; ‌Posture data‌: It can detect the human body's posture, such as standing, sitting, lying, etc., which is very useful for evaluating the patient's physical condition; ‌Gait data‌: By monitoring the changes in the human body's gait, the patient's walking pattern can be analyzed, which is of great significance for medical monitoring and rehabilitation training.

[0025] Among them, infrared sensors have a variety of application scenarios in the medical field: they can also be used in medical monitoring: used to monitor patients' vital signs such as body temperature and respiration, and detect abnormal conditions in a timely manner; used in rehabilitation training: by monitoring the patient's gait and posture, it helps with rehabilitation training and effect evaluation; used in smart homes: used in conjunction with household appliances to achieve intelligent control and improve the quality of life; in short, infrared sensors are widely used in the medical field, and can provide rich motion, posture and gait data, providing strong support for medical monitoring and rehabilitation training.

[0026] Among them, the accelerometer is a sensor that measures the acceleration of an object. It works by measuring the acceleration caused by gravity or other forces. It can be used to detect the patient's movement status, posture, etc. In gait analysis, accelerometers and gyroscopes are used to capture motion information by measuring the position and status information of multiple parts of the legs at each moment. After data processing, the simulation and analysis of biomechanics can be completed with the assistance of computer software.

[0027] Among them, the gyroscope is an angular velocity detection device that uses the momentum of a high-speed rotating body to sense the shell's relative inertial space around one or two axes orthogonal to the axis of rotation. The gyroscope can play a key role in gait detection by measuring and maintaining direction and is not affected by environmental factors such as atmosphere and light. It can accurately measure key data such as acceleration and angular velocity during walking. These data are crucial for evaluating key indicators such as walking stability, cadence, and step length.

[0028] Specifically, the applications of gyroscopes in gait detection are as follows: 1) Accurately measure walking data: The gyroscope can accurately measure key data such as acceleration and angular velocity during walking, which are essential for evaluating key indicators such as walking stability, cadence, and step length; 2) Identify walking patterns and abnormalities: By analyzing the data collected by the gyroscope, different walking patterns can be effectively identified, such as normal gait, abnormal gait, etc., which is of great significance for timely detection and correction of bad gait and prevention of sports injuries; 3) Evaluate rehabilitation effect: In rehabilitation medicine, gyroscopes are also widely used to evaluate the rehabilitation effect of patients. By comparing the gait data before and after treatment, doctors can intuitively understand the patient's recovery and adjust the treatment plan.

[0029] Among them, the magnetometer can be used to test the strength and direction of the magnetic field and locate the patient's position. The principle of the magnetometer is similar to that of the compass, and can measure the angle between the patient and the four directions of east, south, west and north.

[0030] The magnetometer measures changes in the Earth's magnetic field and, combined with data from the accelerometer and gyroscope, can accurately analyze a patient's movements, posture, and gait.

[0031] In this embodiment, the data processing module includes: Data cleaning unit, used to clean the real-time data of patient behavior; Obtain real-time data on patient behavior; Conduct consistency checks on real-time data on patient behavior; According to the data consistency requirements, each parameter in the real-time data of patient behavior is checked one by one to check whether there is inconsistent data in the real-time data of patient behavior that is useless for patient fall monitoring, and the inconsistent data in the real-time data of patient behavior is removed; Check invalid and missing values ​​in real-time patient behavior data; According to the requirements of data validity and completeness, each parameter in the real-time patient behavior data is checked one by one to check whether there are invalid values ​​and missing values ​​in the real-time patient behavior data that are useless for patient fall monitoring. The invalid values ​​and missing values ​​in the real-time patient behavior data are removed to determine the real-time patient behavior data that is useful for patient fall monitoring.

[0032] It should be noted that data cleaning refers to the process of processing and organizing raw data during data analysis to improve the quality and usability of the data.

[0033] Therefore, by cleaning the real-time patient behavior data, inconsistent data, invalid values ​​and missing values ​​in the real-time patient behavior data that are useless for patient fall monitoring can be removed, and then the real-time patient behavior data that is useful for patient fall monitoring can be determined, which can improve the subsequent processing accuracy and efficiency of real-time patient behavior data.

[0034] In this embodiment, the data processing module further includes: A data conversion unit, used to convert real-time data of patient behavior; Obtain real-time data on patient behavior useful for patient fall monitoring; Converting the real-time patient behavior data useful for patient fall monitoring to unify the formats of the real-time patient behavior data useful for patient fall monitoring; Eliminate the dimensional differences between real-time patient behavior data that are useful for patient fall monitoring and determine standardized real-time patient behavior data; A feature extraction unit, used for extracting features from real-time patient behavior data; Obtain standardized real-time data on patient behavior; Extract features from standardized real-time patient behavior data; Features that can reflect patient falls and are related to patient falls are extracted from standardized real-time patient behavior data to determine patient behavior feature data.

[0035] It should be noted that by converting and extracting features from real-time patient behavior data, the patient behavior feature data can be determined, which facilitates subsequent risk analysis and identification of the patient's fall behavior.

[0036] In this embodiment, the analysis and identification module includes: A model training unit, used to train a patient fall risk identification model; Collect patient behavior history data based on patient fall monitoring needs based on behavioral feature identification; Define the patient's behavioral history data, define the behavioral characteristics related to the patient's fall, classify the behavioral characteristics, and determine the training set and test set; Select a machine learning framework suitable for patient fall risk identification based on behavioral signature recognition; Based on the training set, the selected machine learning framework suitable for patient fall risk identification based on behavioral feature recognition is trained to determine a patient fall risk identification model based on behavioral feature recognition.

[0037] In this embodiment, the analysis and identification module further includes: A test optimization unit, used to test and optimize the patient fall risk identification model; Obtain a patient fall risk identification model based on behavioral feature recognition; Based on the test set, the performance of the patient fall risk identification model based on behavioral feature recognition is tested to determine whether the patient fall risk identification model based on behavioral feature recognition can achieve the expected effect; Determine the performance test results of the patient fall risk identification model; According to the performance test results of the patient fall risk identification model, the patient fall risk identification model based on behavioral feature identification was analyzed, and the parameters of the patient fall risk identification model based on behavioral feature identification were adjusted and the structure was optimized. After repeated iterations, the optimal patient fall risk identification model was determined.

[0038] In this embodiment, the analysis and identification module further includes: Analysis and identification unit, used for risk analysis and identification of patient falling behavior; Obtain the optimal patient fall risk identification model; Deploy the optimal patient fall risk identification model in the actual patient fall monitoring environment; Input patient behavioral characteristic data into the optimal patient fall risk identification model; Based on the optimal patient fall risk identification model, predict and analyze the patient's behavioral characteristic data and identify risks, determine whether the patient is at risk of falling, and determine the patient's fall risk identification results; The patient fall risk identification result is that the patient is at risk of falling or the patient is not at risk of falling.

[0039] It should be noted that the patient behavior characteristic data is predicted and analyzed based on the optimal patient fall risk identification model to determine whether the patient is at risk of falling and determine the patient fall risk identification result. Among them, when the patient is at risk of falling, an alarm is automatically issued for the patient's fall behavior and measures are taken to prevent the fall from occurring, reminding people around the patient to protect the patient.

[0040] In this embodiment, the early warning control module includes: The plan formulation unit is used to formulate patient fall warning and control plans; Obtain patient fall risk identification results; When a patient is at risk of falling, the patient's behavioral characteristic data is analyzed to develop a personalized patient fall warning and control plan for the patient; Early warning control unit, used for early warning control of patients' falling behavior; Among them, based on the personalized patient fall warning and control plan, the patient's fall behavior is warned and controlled. For the patient's fall behavior, an alarm is automatically issued and measures are taken to prevent the fall from occurring, reminding people around the patient to protect the patient.

[0041] When the model is actually applied to monitor patient falls, due to the influence of objective environmental factors and subjective factors of the patient, the same model is not applicable to all situations. For the specific application scenario of this model, in one embodiment, the patient fall monitoring system based on behavioral feature recognition also includes: An environmental assessment module is used to obtain environmental data of the monitoring space and evaluate the environmental data to obtain environmental assessment data; A patient assessment module is used to obtain the patient's physical condition data and evaluate the physical condition data to obtain patient assessment data; A model configuration module, used for configuring a model used by the analysis and identification module to perform predictive analysis and risk identification of patient fall behaviors based on environmental assessment data and patient assessment data; This embodiment comprehensively analyzes the environment and patients, and calls the model based on the evaluation to ensure the accuracy and effectiveness of the detection; Among them, the environmental assessment module performs the following operations: When receiving an environmental assessment request, output a preset environmental simulation construction interface; Receiving a call to a three-dimensional model corresponding to each object on the environment simulation construction interface and constructing an environment simulation model in a construction area of ​​the environment simulation construction interface; the environment simulation model may include a three-dimensional modeling of a real environment; Perform risk assessment on each three-dimensional model involved in the construction of the environmental simulation model to obtain first risk assessment data; that is, perform an assessment on each individual object in the environment to assess whether it will cause the patient's fall risk. Specifically, this can be performed through a pre-configured object risk analysis library, in which the first risk assessment data corresponding to each object is configured. For example, a stool has the risk of inducing a patient to fall. Based on the preset association library, each 3D model is evaluated for association to obtain the second risk assessment data; there is association (linkage) between each object, and the combination of two or more objects may be more dangerous or safer; for example: a stool and a table, when the stool is placed under the table, it eliminates the risk of inducing patient falls when the stool is placed alone; a stool and an aisle, this association increases the risk of inducing patient falls; etc. Extract features from the environmental simulation model, and perform a comprehensive evaluation based on a preset comprehensive analysis library to obtain third risk assessment data; extract features from the environmental simulation model including: segmenting the environmental simulation model based on a segmentation grid, determining the three-dimensional model number corresponding to each segmentation unit, arranging the three-dimensional model number according to the position of the segmentation unit, and forming a feature data set corresponding to the environmental simulation model; matching the feature data set with the analysis data set corresponding to each third risk assessment data in the comprehensive analysis library, and retrieving the matched third risk assessment data; the third risk assessment data includes: a comprehensive risk assessment value, the location and area of ​​the risk area, etc.; The first risk assessment data, the second risk assessment data and the third risk assessment data are integrated to obtain environmental assessment data; Among them, the patient assessment module performs the following operations: When receiving a patient assessment application, output a preset data input interface; Receive basic condition data input in the basic condition input area of ​​the data input interface, case data input in the historical case data input area, and current test data input in the current test data input area; Comprehensively analyze basic condition data, case data and current test data to obtain patient assessment data.

[0042] Among them, the model configuration module performs the following operations: Performing feature extraction on the first risk assessment data, the second risk assessment data, and the third risk assessment data respectively to obtain a plurality of first feature parameters; quantifying the first risk assessment data, the second risk assessment data, and the third risk assessment data according to a pre-configured quantization model to obtain quantization parameter values ​​corresponding to each risk assessment item, wherein the first feature parameters include: an average value, a maximum value, and a minimum value of the quantization parameter value; Extracting features from the patient assessment data to obtain a plurality of second feature parameters; the second feature parameters include a parameter indicating age, a parameter indicating weight, a parameter indicating gender, a parameter indicating the presence of a disease and the type of the disease, a parameter indicating whether the disease has ever existed and the type of the disease, etc.; The plurality of first characteristic parameters and the plurality of second characteristic parameters are sequentially filled into a preset array template to form an analysis data set and the data in the analysis data set are normalized. The normalization processing formula is as follows: In the formula, The first Line The value of the column; is the first Line The value of the column; is the first Line Column values; analyze the data in the data set OK List; The analysis data set is matched with the call data sets corresponding to the models used for predictive analysis and risk identification of patient fall behavior in the pre-configured model call library. The matching is performed by calculating the similarity between the two. The similarity calculation formula is as follows: In the formula, Indicates similarity; The first Line The value of the column; To call the first Line The value of the column; The model associated with the calling data set with the largest similarity and greater than the preset similarity threshold is retrieved.

[0043] This embodiment uses a pre-configured model call library to perform analysis calls. The principle to be followed is that the greater the risk of the patient being assessed or the greater the risk of the environment, the greater the sensitivity requirement of the model to be used.

[0044] In one embodiment, the patient fall monitoring system based on behavioral feature recognition further includes: Visual monitoring module, used to realize visual monitoring of patients; Among them, the visual monitoring module includes: The scene model building unit is used to build a scene model according to the scene data input by the user. The user can upload an image of the scene by shooting, and the platform will analyze the image and build a corresponding scene model framework. The customer adjusts the parameters of the scene model accordingly to form the final scene model. A personnel model building module is used to retrieve a corresponding personnel model from a pre-configured personnel model library according to the patient's motion data, posture data and gait data; The positioning module is used to obtain the patient's position in the scene and place the person model into the scene model according to the obtained position to form a visual monitoring model; the positioning module can be positioned by an image analysis positioning method or a positioning method of a positioning module on a terminal worn by the patient; the image positioning method is to analyze the image in the scene and locate the patient based on the relative position relationship between each marker in the scene and the patient; the positioning method of the positioning module is an existing mature technology and will not be explained in detail here; The real-time risk analysis module is used to analyze the visual monitoring model, identify risk items, and issue corresponding reminders; The real-time risk analysis module analyzes the visual monitoring model and determines risk items, including: Extract various risk objects within the preset area around the personnel model in the visual monitoring model; Retrieve the risk assessment items corresponding to each risk object as items to be screened; record each item to be screened as ; Indicates The first risk risk assessment items; Analyze the patient's motion data, posture data, and gait data to determine the behavior set; Based on the pre-configured association library, determine the association between each behavior in the behavior set and the risk object and retrieve the association coefficient; denoted as and ; Indicates Behavior and The first risk The correlation coefficient between the risk assessment items for distance; Indicates Behavior and The first risk The correlation coefficient between the risk assessment items for the angle; Based on the correlation coefficient, the evaluation value between each behavior and each risk assessment item is determined. The calculation formula is as follows: ;in, Indicates The distance between the risk object and the patient; Indicates The angle between the risk object and the patient; Extract risk assessment items whose assessment values ​​are greater than a preset assessment threshold as risk items; The patient's motion data, posture data and gait data are analyzed to determine the behavior set, including: based on a preset behavior analysis library, calling the corresponding behavior set; The behavior set in the behavior analysis library is matched with the first retrieved parameter of the corresponding action data, the second retrieved parameter of the corresponding posture, and the third retrieved parameter of the corresponding gait data in detail by the following formula: In the formula, TD is the matching degree; is the dth action feature parameter after feature extraction of action data; The first call parameter for the dth one; is the jth posture feature parameter after feature extraction of posture data; The second call parameter for the jth one; is the u-th posture feature parameter after feature extraction of gait data; is the uth third call parameter; D is the total number of first call parameters; J is the total number of second call parameters; U is the total number of third call parameters; , , is a pre-configured weight coefficient; A behavior set corresponding to a set of the first calling parameter, the second calling parameter, and the third calling parameter with the greatest matching degree is called.

[0045] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0046] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A patient fall monitoring system based on behavioral feature recognition, characterized in that: include: A data acquisition module is used to collect the patient's movements, postures and gait using sensors to determine the patient's real-time behavior data; The data processing module is used to clean, convert and extract features of the real-time data of patient behavior and determine the patient behavior feature data; The analysis and identification module is used to predict and analyze the patient's falling behavior and identify risks, determine whether the patient has a falling risk, and determine the patient's falling risk identification results; The early warning and control module is used to formulate a patient fall early warning and control plan, to conduct early warning and control of patient fall behaviors, and to prevent falls from occurring.

2. The patient fall monitoring system based on behavioral feature recognition according to claim 1 is characterized in that: The data acquisition module comprises: A motion monitoring unit, used to collect patient motion data; Monitor and collect the patient's movements in real time based on sensors to obtain the patient's movement data; A posture monitoring unit, used to collect patient posture data; Monitor and collect the patient's posture in real time based on sensors to obtain the patient's posture data; A gait monitoring unit, used to collect gait data of patients; Monitor and collect the patient's gait in real time based on sensors to obtain the patient's gait data; Among them, the real-time data of patient behavior is determined based on the patient motion data, patient posture data and patient gait data.

3. The patient fall monitoring system based on behavioral feature recognition according to claim 2 is characterized in that: The data processing module comprises: Data cleaning unit, used to clean the real-time data of patient behavior; Obtain real-time data on patient behavior; Conduct consistency checks on real-time data on patient behavior; According to the data consistency requirements, each parameter in the real-time data of patient behavior is checked one by one to check whether there is inconsistent data in the real-time data of patient behavior that is useless for patient fall monitoring, and the inconsistent data in the real-time data of patient behavior is removed; Check invalid and missing values ​​in real-time patient behavior data; According to the requirements of data validity and completeness, each parameter in the real-time patient behavior data is checked one by one to check whether there are invalid values ​​and missing values ​​in the real-time patient behavior data that are useless for patient fall monitoring. The invalid values ​​and missing values ​​in the real-time patient behavior data are removed to determine the real-time patient behavior data that is useful for patient fall monitoring.

4. The patient fall monitoring system based on behavioral feature recognition according to claim 3 is characterized in that: The data processing module further includes: A data conversion unit, used to convert real-time data of patient behavior; Obtain real-time data on patient behavior useful for patient fall monitoring; Converting the real-time patient behavior data useful for patient fall monitoring to unify the formats of the real-time patient behavior data useful for patient fall monitoring; Eliminate the dimensional differences between real-time patient behavior data that are useful for patient fall monitoring and determine standardized real-time patient behavior data; A feature extraction unit, used for extracting features from real-time patient behavior data; Obtain standardized real-time data on patient behavior; Extract features from standardized real-time patient behavior data; Features that can reflect patient falls and are related to patient falls are extracted from standardized real-time patient behavior data to determine patient behavior feature data.

5. The patient fall monitoring system based on behavioral feature recognition according to claim 4 is characterized in that: The analysis and identification module comprises: A model training unit, used to train a patient fall risk identification model; Collect patient behavior history data based on patient fall monitoring needs based on behavioral feature identification; Define the patient's behavioral history data, define the behavioral characteristics related to the patient's fall, classify the behavioral characteristics, and determine the training set and test set; Select a machine learning framework suitable for patient fall risk identification based on behavioral signature recognition; Based on the training set, the selected machine learning framework suitable for patient fall risk identification based on behavioral feature recognition is trained to determine a patient fall risk identification model based on behavioral feature recognition.

6. The patient fall monitoring system based on behavioral feature recognition according to claim 5, characterized in that: The analysis and identification module further includes: A test optimization unit, used to test and optimize the patient fall risk identification model; Obtain a patient fall risk identification model based on behavioral feature recognition; Based on the test set, the performance of the patient fall risk identification model based on behavioral feature recognition is tested to determine whether the patient fall risk identification model based on behavioral feature recognition can achieve the expected effect; Determine the performance test results of the patient fall risk identification model; According to the performance test results of the patient fall risk identification model, the patient fall risk identification model based on behavioral feature identification was analyzed, and the parameters of the patient fall risk identification model based on behavioral feature identification were adjusted and the structure was optimized. After repeated iterations, the optimal patient fall risk identification model was determined.

7. The patient fall monitoring system based on behavioral feature recognition according to claim 6, characterized in that: The analysis and identification module further includes: Analysis and identification unit, used for risk analysis and identification of patient falling behavior; Obtain the optimal patient fall risk identification model; Deploy the optimal patient fall risk identification model in the actual patient fall monitoring environment; Input patient behavioral characteristic data into the optimal patient fall risk identification model; Based on the optimal patient fall risk identification model, predict and analyze the patient's behavioral characteristic data and identify risks, determine whether the patient is at risk of falling, and determine the patient's fall risk identification results; The patient fall risk identification result is that the patient is at risk of falling or the patient is not at risk of falling.

8. The patient fall monitoring system based on behavioral feature recognition according to claim 7, characterized in that: The early warning control module includes: The plan formulation unit is used to formulate patient fall warning and control plans; Obtain patient fall risk identification results; When a patient is at risk of falling, the patient's behavioral characteristic data is analyzed to develop a personalized patient fall warning and control plan for the patient; Early warning control unit, used for early warning control of patients' falling behavior; Among them, based on the personalized patient fall warning and control plan, the patient's fall behavior is warned and controlled. For the patient's fall behavior, an alarm is automatically issued and measures are taken to prevent the fall from occurring, reminding people around the patient to protect the patient.

9. The patient fall monitoring system based on behavioral feature recognition according to claim 1, characterized in that: Also includes: An environmental assessment module is used to obtain environmental data of the monitoring space and evaluate the environmental data to obtain environmental assessment data; A patient assessment module is used to obtain the patient's physical condition data and evaluate the physical condition data to obtain patient assessment data; A model configuration module, used for configuring a model used by the analysis and identification module to perform predictive analysis and risk identification of patient fall behaviors based on environmental assessment data and patient assessment data; Among them, the environmental assessment module performs the following operations: When receiving an environmental assessment request, output a preset environmental simulation construction interface; Receiving a call to a three-dimensional model corresponding to each object on the environment simulation construction interface and constructing an environment simulation model in a construction area of ​​the environment simulation construction interface; Performing risk assessment on each three-dimensional model involved in building the environmental simulation model to obtain first risk assessment data; Based on the preset association library, each three-dimensional model is associated and evaluated to obtain second risk assessment data; Extract features from the environmental simulation model and conduct a comprehensive evaluation based on the preset comprehensive analysis library to obtain the third risk assessment data; The first risk assessment data, the second risk assessment data and the third risk assessment data are integrated to obtain environmental assessment data; Among them, the patient assessment module performs the following operations: When receiving a patient assessment application, output a preset data input interface; Receive basic condition data input in the basic condition input area of ​​the data input interface, case data input in the historical case data input area, and current test data input in the current test data input area; Comprehensively analyze basic condition data, case data and current test data to obtain patient assessment data; The model configuration module performs the following operations: Performing feature extraction on the first risk assessment data, the second risk assessment data, and the third risk assessment data respectively to obtain a plurality of first feature parameters; Performing feature extraction on the patient assessment data to obtain a plurality of second feature parameters; The plurality of first characteristic parameters and the plurality of second characteristic parameters are sequentially filled into a preset array template to form an analysis data set and the data in the analysis data set are normalized. The normalization processing formula is as follows: In the formula, The first Line The value of the column; is the first Line The value of the column; is the first Line Column values; analyze the data in the data set OK List; The analysis data set is matched with the call data sets corresponding to the models used for predictive analysis and risk identification of patient fall behavior in the pre-configured model call library. The matching is performed by calculating the similarity between the two. The similarity calculation formula is as follows: In the formula, Indicates similarity; The first Line The value of the column; To call the first Line The value of the column; The model associated with the calling data set with the largest similarity and greater than the preset similarity threshold is retrieved.

10. The patient fall monitoring system based on behavioral feature recognition according to claim 1, characterized in that: The patient fall monitoring system based on behavioral feature recognition also includes: Visual monitoring module, used to realize visual monitoring of patients; Among them, the visual monitoring module includes: A scene model building unit, used to build a scene model according to scene data input by a user; A personnel model building module is used to retrieve a corresponding personnel model from a pre-configured personnel model library according to the patient's motion data, posture data and gait data; A positioning module is used to obtain the patient's position in the scene and place the person model into the scene model according to the obtained position to form a visual monitoring model; The real-time risk analysis module is used to analyze the visual monitoring model, identify risk items, and issue corresponding reminders; The real-time risk analysis module analyzes the visual monitoring model and determines risk items, including: Extract various risk objects within the preset area around the personnel model in the visual monitoring model; Retrieve the risk assessment items corresponding to each risk object as items to be screened; record each item to be screened as ; Indicates The first risk risk assessment items; Analyze the patient's motion data, posture data, and gait data to determine the behavior set; Based on the pre-configured association library, determine the association between each behavior in the behavior set and the risk object and retrieve the association coefficient, which is recorded as and ; Indicates Behavior and The first risk The correlation coefficient between the risk assessment items for distance; Indicates Behavior and The first risk The correlation coefficient between the risk assessment items for the angle; Based on the correlation coefficient, the evaluation value between each behavior and each risk assessment item is determined. The calculation formula is as follows: ;in, Indicates The distance between the risk object and the patient; Indicates The angle between the risk object and the patient; Extract risk assessment items whose assessment values ​​are greater than a preset assessment threshold as risk items; The patient's motion data, posture data and gait data are analyzed to determine the behavior set, including: based on a preset behavior analysis library, calling the corresponding behavior set; The behavior set in the behavior analysis library is matched with the first retrieved parameter of the corresponding action data, the second retrieved parameter of the corresponding posture, and the third retrieved parameter of the corresponding gait data in detail by the following formula: In the formula, TD is the matching degree; is the dth action feature parameter after feature extraction of action data; The first call parameter for the dth one; is the jth posture feature parameter after feature extraction of posture data; The second call parameter for the jth one; is the u-th posture feature parameter after feature extraction of gait data; is the uth third call parameter; D is the total number of first call parameters; J is the total number of second call parameters; U is the total number of third call parameters; , , is a pre-configured weight coefficient; A behavior set corresponding to a set of the first calling parameter, the second calling parameter, and the third calling parameter with the greatest matching degree is called.

Images