Search

CN-121795889-B - Fall-preventing monitoring method and system based on intelligent bed

CN121795889BCN 121795889 BCN121795889 BCN 121795889BCN-121795889-B

Abstract

The invention discloses an intelligent bed-based fall-prevention monitoring method and system, which relate to the technical field of intelligent beds for behavior feature recognition, and accurately divide the coordinate ranges of a bed body and surrounding space by constructing a three-dimensional airspace monitoring frame, acquire human motion data in real time by fusing a multi-sensor technology, dynamically tracking the change of the centroid track, calculating the centroid horizontal offset rate and centroid offset direction vector, deeply analyzing the limb suspension proportion and the gesture inclination angle, determining the trend of increasing the limb suspension duration, and further evaluating the falling risk probability value. According to the invention, through the combination of centroid dynamic track analysis and multi-level monitoring grids, an accurate three-dimensional airspace monitoring range is established around the bed body, and the dynamic relation change rate of the human centroid and the bed body coordinates is calculated in real time, so that the early positioning of the falling occurrence airspace is realized.

Inventors

  • LIU YIXIANG

Assignees

  • 凯莱思(厦门)智能家居有限责任公司

Dates

Publication Date
20260508
Application Date
20260312

Claims (6)

  1. 1. A smart bed-based fall prevention monitoring method, performed by a computer, comprising: Dividing the space around the bed body based on a pre-established three-dimensional model of the bed body to obtain coordinate ranges of a bed surface, a bed side and the ground, and constructing a multi-level monitoring grid for the divided space region based on the coordinate ranges of the bed surface, the bed side and the ground to obtain an initial three-dimensional airspace monitoring frame; Acquiring human body motion data in real time by adopting a multi-sensor fusion technology based on the initial three-dimensional airspace monitoring frame, and determining the spatial coordinate distribution of each part of a human body based on the human body motion data, wherein the spatial coordinate distribution comprises the suspended state of limbs of the human body and the initial three-dimensional spatial coordinate of the mass center; The method comprises the steps of continuously tracking the spatial coordinate distribution of each part of a human body to obtain a centroid horizontal offset rate and a centroid offset direction vector, calculating the centroid dynamic trajectory of the human body centroid in the three-dimensional space by continuously tracking the spatial coordinate distribution of each part of the human body to obtain a real-time coordinate sequence, calculating the human body centroid position change of a current frame based on the real-time coordinate sequence to obtain a centroid dynamic trajectory of the human body centroid in the three-dimensional space, smoothing the centroid dynamic trajectory by adopting Kalman filtering to determine a smoothed centroid trajectory, calculating the centroid horizontal displacement and the centroid vertical displacement between adjacent frames based on the smoothed centroid trajectory to obtain the centroid horizontal offset rate and the centroid vertical sinking rate, and synthesizing the centroid offset direction vector based on the centroid horizontal offset rate and the centroid vertical sinking rate; If the centroid horizontal offset rate exceeds a preset threshold range and the centroid offset direction vector points to a bedside coordinate range, carrying out depth analysis on the dynamic change track, and calculating the proportion and the posture inclination angle of the suspended limb at the bedside to determine the trend of increasing the suspended duration of the limb; based on the trend of the increase of the duration of the suspended limb, analyzing the analysis result of the human body posture change in real time, and based on the analysis result of the human body posture change, calculating the curvature change of the centroid movement track so as to determine the association data between the centroid acceleration component and the track curvature change; The method comprises the steps of acquiring real-time centroid acceleration data based on the correlation data to obtain a centroid acceleration sequence, calculating trajectory curvature change of a centroid based on the centroid acceleration sequence to obtain a trajectory curvature change sequence, extracting centroid offset direction based on the trajectory curvature change sequence to obtain a centroid offset direction vector, judging the consistency degree of the offset direction by adopting a sliding window statistical method based on the centroid offset direction vector to obtain a direction consistency index, calculating the coordinate offset of a bed body based on the direction consistency index and the coordinate ranges of the bed surface, the bed side and the ground, and determining the coordinate offset value, and inputting the coordinate offset value into a support vector machine model to obtain the pre-judging probability of the falling risk output by the support vector machine model.
  2. 2. The intelligent bed-based fall prevention monitoring method according to claim 1, wherein the acquiring human motion data in real time by using a multi-sensor fusion technology based on the initial three-dimensional airspace monitoring frame, and determining the spatial coordinate distribution of each part of the human based on the human motion data, comprises: Acquiring human body motion data in real time through a multi-sensor fusion technology based on the initial three-dimensional airspace monitoring frame, and denoising the human body motion data to obtain cleaned motion signal data; based on the cleaned motion signal data, adopting a preset filtering rule to perform preliminary screening on the position information in the motion signal data so as to remove abnormal values and determine preliminary space coordinates of each part of the human body; based on the preliminary space coordinates, calculating the relative position relation of each part of the human body by using a geometric transformation method so as to determine the suspended state of limbs of the human body based on the relative position relation; Based on the suspended state of the human body limbs, calculating initial three-dimensional space coordinates of the mass center by a weighted average method to obtain a preliminary estimated value of the position of the mass center; Based on the preliminary estimated value of the centroid position, dynamically correcting the centroid position by adopting a Kalman filtering algorithm to obtain a centroid three-dimensional coordinate; based on the centroid three-dimensional coordinates, combining the human limb suspension state to construct a human body posture space model so as to determine the space coordinate distribution based on the human body posture space model.
  3. 3. The intelligent bed-based fall prevention monitoring method according to claim 1, wherein if the centroid horizontal offset rate exceeds a preset threshold range and the centroid offset direction vector points to a bedside coordinate range, performing depth analysis on the dynamic change track, and calculating a bedside suspended limb proportion and a posture inclination angle to determine a trend of increasing a limb suspended duration, comprising: If the centroid horizontal offset rate exceeds a preset threshold range, determining that the centroid offset direction vector points to a bedside coordinate range based on the included angle relation between the centroid offset direction vector and the bedside coordinate range; when the centroid offset direction vector points to the bedside coordinate range, carrying out limb segmentation processing on the dynamic change track, identifying limb parts positioned outside the bedside coordinate range, and calculating the suspended limb proportion at the bedside; Calculating to obtain an attitude inclination angle based on the shortest distance between a centroid projection point and the edge of the bed surface in the dynamic change track; Based on the bedside suspended limb proportion and the posture inclination angle, fitting a linear regression model to the change slope of the bedside suspended limb proportion along with time, and determining the trend of increasing the suspended duration of the limb; And if the change slope is a positive value, determining that the duration of the suspended limb is in an increasing trend.
  4. 4. The intelligent bed-based fall prevention monitoring method according to claim 1, wherein the real-time analysis of the human body posture change analysis result based on the trend of the increase of the suspended duration of the limb, and the calculation of the curvature change of the centroid movement track based on the human body posture change analysis result, to determine the correlation data between the centroid acceleration component and the track curvature change, comprises: Based on the trend of the increase of the duration of the suspended limbs, calculating three-dimensional coordinate changes of key points of the human body in real time, and determining a human body posture change analysis result; Extracting a limb inclination angle according to the analysis result of the human body posture change, and dynamically adjusting a mapping parameter between an inclination angle threshold and a bed coordinate based on the limb inclination angle; calculating the depth of the limb part entering the high risk area based on the adjusted mapping parameters to obtain a depth change sequence; Determining a gesture adjustment frequency based on the depth change sequence, and determining a centroid position sequence of a human centroid based on the gesture adjustment frequency and combining the human gesture change analysis result; and calculating curvature change of the centroid movement track based on the centroid position sequence, and determining association data between a centroid acceleration component and track curvature change by adopting a random forest regression model based on the centroid position sequence and the curvature change.
  5. 5. The intelligent bed-based fall prevention monitoring method of claim 1, further comprising: if the fall risk pre-judging probability value reaches a preset alarm threshold value, generating a protection trigger signal, wherein the protection trigger signal is used for determining a protection measure starting time; acquiring attitude and position data of a bedside based on the protection trigger signal, classifying attitude features by adopting a random forest classifier based on the attitude and position data, and determining classification results of the attitude features; if the classification result has a trigger condition, extracting signal type information and severity information from the protection trigger signal; determining signal priority based on the signal type information and the severity information, and sorting a plurality of concurrent signals of bedside protection action based on the signal priority to obtain a sorted signal priority sequence; Receiving the ordered signal priority sequence through an intelligent bed control module, and generating a target execution instruction corresponding to the bedside protection action; Based on the target execution instruction, the intelligent bed is driven to execute the bedside lifting column or the bed body tilting action so as to intercept the bedside falling risk in real time.
  6. 6. A smart-bed based fall prevention monitoring system employing a smart-bed based fall prevention monitoring method as claimed in any one of claims 1-5, comprising: the monitoring frame generation module is used for dividing the space around the bed body based on a pre-established three-dimensional model of the bed body to obtain coordinate ranges of the bed surface, the bed edge and the ground, and constructing a multi-level monitoring grid for the divided space region based on the coordinate ranges of the bed surface, the bed edge and the ground to obtain an initial three-dimensional airspace monitoring frame; The system comprises a space coordinate distribution generation module, a control module and a control module, wherein the space coordinate distribution generation module is used for acquiring human motion data in real time by adopting a multi-sensor fusion technology based on the initial three-dimensional airspace monitoring frame and determining the space coordinate distribution of each part of a human body based on the human motion data, wherein the space coordinate distribution comprises a human limb suspension state and a centroid initial three-dimensional space coordinate; The centroid track determining module is used for continuously tracking the space coordinate distribution of each part of the human body, calculating the dynamic change track of the human body centroid in the three-dimensional space and obtaining the centroid horizontal offset rate and the centroid offset direction vector; The trend determining module is used for carrying out depth analysis on the dynamic change track if the centroid horizontal offset rate exceeds a preset threshold range and the centroid offset direction vector points to a bedside coordinate range, and calculating the proportion and the posture inclination angle of the suspended limb at the bedside so as to determine the trend of the increase of the suspended duration of the limb; the related data generation module is used for analyzing a human body posture change analysis result in real time based on the trend of the increase of the suspended duration of the limb, and calculating the curvature change of the centroid movement track based on the human body posture change analysis result so as to determine related data between a centroid acceleration component and the track curvature change; and the falling risk generation module is used for calculating a centroid deviation direction vector and a bed coordinate deviation based on the associated data so as to determine a falling risk pre-judging probability value based on the centroid deviation direction vector and the bed coordinate deviation.

Description

Fall-preventing monitoring method and system based on intelligent bed Technical Field The invention relates to the technical field of intelligent beds for behavior feature recognition, in particular to an anti-falling monitoring method and system based on an intelligent bed. Background With the aging of population, the risk of falling down of the aged and patients with mobility impairment on the bed is increasingly prominent, and the problem has become a potential safety hazard which needs to be solved in the medical care field, and is directly related to life health and family burden. The current fall-prevention monitoring method relies on a wearable device or a fixed camera in a room, but the methods are easily affected by light change, privacy limitation and poor wearing compliance of patients, so that the monitoring continuity is insufficient, and particularly, the full-period reliable coverage is difficult to realize at night or when the patients are in a bedridden state alone. In the field of intelligent bed fall prevention monitoring, accurate division of the space around the bed body and realization of effective management of risk areas face significant challenges. The space around the bed is complex and changeable, and comprises a bed surface, a bedside, a ground and other various areas, the areas are required to be classified by the system, otherwise, the moment that the human body part enters the high-risk position is difficult to accurately capture. However, simply dividing the space is not enough to cope with the practical problem, because the posture and the gravity center of the human body are continuously changed in the process of leaving the bed, the relative position relationship between the body part and the boundary of the bed body is adjusted at any time, so that the monitoring system is difficult to grasp the dynamic space occupation state in real time. The combination of the space division and the dynamic position relation further causes the difficult problem of predicting the position of falling. When the two legs of a patient are suspended at the bedside or the gravity center of the upper body slowly deviates beyond the bedside, if the system cannot quickly calculate the coordinate change and the gravity center deviation rate of the body part in the three-dimensional space, the optimal intervention time is lost before the protective measures are started, so that the transient process from the intention of getting out of the bed to the actual falling cannot be effectively intercepted. The real-time capturing of the space dynamic change is correlated with the advanced judgment of the falling airspace position, and the accurate triggering of the protective measures is directly restricted. Therefore, how to build an accurate three-dimensional airspace monitoring range around the bed body and calculate the dynamic relation change rate of the mass center of the human body and the coordinates of the bed body in real time, so that the early positioning of the falling occurrence airspace is realized, and the intelligent bed-based falling prevention monitoring method and system research are key problems. Disclosure of Invention The invention provides an intelligent bed-based fall-prevention monitoring method and system, which are used for establishing an accurate three-dimensional airspace monitoring range around a bed body, calculating the dynamic relation change rate of the mass center of a human body and the coordinates of the bed body in real time and realizing the advanced positioning of a falling occurrence airspace. The invention provides an intelligent bed-based fall prevention monitoring method, which is executed by a computer and comprises the following steps: Dividing the space around the bed body based on a pre-established three-dimensional model of the bed body to obtain coordinate ranges of a bed surface, a bed side and the ground, and constructing a multi-level monitoring grid for the divided space region based on the coordinate ranges of the bed surface, the bed side and the ground to obtain an initial three-dimensional airspace monitoring frame; Acquiring human body motion data in real time by adopting a multi-sensor fusion technology based on the initial three-dimensional airspace monitoring frame, and determining the spatial coordinate distribution of each part of a human body based on the human body motion data, wherein the spatial coordinate distribution comprises the suspended state of limbs of the human body and the initial three-dimensional spatial coordinate of the mass center; Continuously tracking the space coordinate distribution of each part of the human body, and calculating the dynamic change track of the mass center of the human body in a three-dimensional space to obtain the horizontal offset rate and the offset direction vector of the mass center; If the centroid horizontal offset rate exceeds a preset threshold range and the centroid offset direction vector