Transfusion state monitoring method and system for intelligent ward

Abstract

The invention belongs to the technical field of intelligent medical treatment, and particularly relates to an infusion state monitoring method and system for an intelligent ward, and the method comprises the steps: fixing the relative position between an infusion bag and a collection device through a relative position regulator, and connecting the relative position regulator with an infusion support; in response to the monitoring starting instruction, each acquisition device starts to acquire infusion bag information corresponding to the acquisition device according to a preset detection rule; in response to the monitoring stopping instruction, each collecting device stops collecting the information of the infusion bag; inputting all the infusion bag information into a liquid medicine residue comprehensive judgment model for comprehensive judgment to obtain an infusion bag residue state; and sending early warning information according to the remaining amount state of the infusion bag. The defect that cost, precision and environmental adaptability are difficult to consider at the same time in the prior art is overcome, and the comprehensive requirements of medical institutions for economical efficiency, reliability and universality are met.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent healthcare, and particularly relates to a method and system for monitoring the infusion status in an intelligent ward. Background Art

[0002] In the prior art, clinical infusion monitoring mainly relies on the following technical means: Firstly, a weighing sensor is used to estimate the remaining liquid volume through weight changes. However, factors such as the shaking of the infusion stand and uneven liquid dripping speed can cause significant fluctuations in the weight data, especially with large errors when the patient moves or adjusts the body position. Secondly, for the image recognition solution based on a visible light camera, although the liquid level position can be directly observed, limited by the characteristics of the semi-transparent infusion bag material, the recognition accuracy significantly decreases in backlight, strong light, or night environments, and problems such as liquid surface reflection and bubble interference have not been effectively solved. Thirdly, although high-precision three-dimensional sensing devices such as depth cameras can detect the liquid level height, the device cost is too high, and a high-performance processor is required to support point cloud data analysis, making it difficult to be deployed on a large scale in ordinary wards.

[0003] Therefore, there is an urgent need for a method and system for monitoring the infusion status in an intelligent ward. By using a relative position adjuster to rigidly fix the relative position between the infusion bag and the acquisition device, the perspective shift caused by patient movement is eliminated, and the liquid level detection error rate is significantly reduced. By combining RGB image preprocessing and near-infrared structured light three-dimensional point cloud dynamic compensation technology, the interference of reflection, bubbles, and complex lighting is overcome, the liquid surface recognition accuracy is effectively improved, and a low-cost alternative solution is adopted. Through the FLSI model to fuse multi-parameter weights such as liquid level and deformation for analysis, a four-level remaining volume status dynamic grading warning is realized, the false alarm rate is significantly reduced, and the modular design is adapted to the existing hospital system, supporting low-cost large-scale deployment. Summary of the Invention

[0004] The object of the present invention is to provide a method for monitoring the infusion status in an intelligent ward, including the following steps:

[0005] Use a relative position adjuster to fix the relative position between the infusion bag and the acquisition device, and connect the relative position adjuster to the infusion stand; in response to the start monitoring instruction, each acquisition device starts to obtain the infusion bag information corresponding to the acquisition device according to the preset detection rules; in response to the stop monitoring instruction, each acquisition device stops collecting the infusion bag information; input all the infusion bag information into the comprehensive judgment model of the liquid medicine remaining volume for comprehensive judgment to obtain the remaining volume status of the infusion bag; issue a warning message according to the shown remaining volume status of the infusion bag;

[0006] The obtaining of the infusion bag information corresponding to the acquisition device includes: using a front RGB image acquisition device to collect the RGB image of the front of the infusion bag, and obtaining the liquid level occupancy information based on the RGB image of the front of the infusion bag; using a side profile information acquisition device to collect the set of infusion bag widths on the side of the infusion bag, and obtaining the average profile information based on the set of infusion bag widths;

[0007] The infusion bag information includes: liquid level ratio information and average profile information;

[0008] The relative position adjuster includes a compatible connection part, a collection device fixing part, and an infusion bag fixing part that are connected in sequence; the compatible connection part is used to connect to the existing infusion stand instead of the infusion bag; the collection device fixing part includes several telescopic fixing arms, and the collection device is fixed on the fixing arms; the infusion bag fixing part is fixedly connected to the infusion bag.

[0009] Further, obtaining the liquid level ratio information based on the RGB image of the front of the infusion bag includes:

[0010] Real-time obtain the front RGB image of the infusion bag through the RGB image collection device fixed on the front of the infusion bag; preprocess the front RGB image to generate standardized image data; extract the liquid level line position information of the infusion bag from the standardized image data based on an image algorithm; calculate the percentage of the liquid level height in the total height of the infusion bag according to the liquid level line position information to obtain the liquid level ratio information.

[0011] Further, preprocessing the front RGB image to generate standardized image data includes:

[0012] Image denoising: Use Gaussian filtering or median filtering to remove image noise;

[0013] Image enhancement: Adjust the image brightness and contrast to highlight the liquid level line and the edge of the infusion bag;

[0014] Image cropping: Crop the region of interest according to the position of the infusion bag in the image;

[0015] Extracting the liquid level line position information of the infusion bag from the standardized image data based on an image algorithm includes:

[0016] Use the Canny edge detection algorithm to extract the liquid level line of the infusion bag;

[0017] Detect the straight line feature of the liquid level line through the Hough transform to determine the position of the liquid level line in the image;

[0018] Calculate the pixel distance between the liquid level line and the bottom of the infusion bag as the liquid level line position information;

[0019] Obtaining the liquid level ratio information includes:

[0020] Liquid level ratio information = liquid level line position information / infusion bag height.

[0021] Further, collecting the infusion bag width set on the side of the infusion bag using the side profile information collection device includes:

[0022] Use a side profile information acquisition device to obtain a depth image of the infusion bag; the side profile information acquisition device includes: a near-infrared structured light projection device and a band-pass filtered CMOS camera;

[0023] Convert the depth image into a three-dimensional point cloud using a sub-pixel interpolation algorithm;

[0024] Perform motion compensation processing on the three-dimensional point cloud to obtain optimized point cloud data; the motion compensation processing includes: Kalman filtering based on an inertial sensor and quaternion attitude solution;

[0025] Obtain the profile information of the side of the infusion bag based on the optimized point cloud data, and based on the profile information, use a number of equally spaced parallel lines to obtain a set of infusion bag widths on the side of the infusion bag.

[0026] Further, the obtaining of the average profile information based on the set of infusion bag widths includes:

[0027] Traverse the set of infusion bag widths in sequence from the top to the bottom of the infusion bag. If the total width from the top to the middle of the infusion bag is greater than a preset first threshold, use all the infusion bag width data from the top to the bottom of the infusion bag to calculate the average value to obtain the average profile information; otherwise, use all the infusion bag width data from the middle to the bottom of the infusion bag to calculate the average value to obtain the average profile information.

[0028] Further, the inputting of all the infusion bag information into the liquid medicine remaining amount comprehensive judgment model for comprehensive judgment includes:

[0029] The liquid medicine remaining amount comprehensive judgment model is:

[0030] FLSI = (W_avg × α) + (H_per × β) + (T_inv × γ) + (ΔW_rate × δ)

[0031] It is necessary to satisfy the weight coefficients: α + β + γ + δ = 1

[0032] In the formula, FLSI is the liquid level state index, W_avg is the average infusion bag width, H_per is the percentage of the liquid level height, T_inv is the reciprocal of the detection duration, and ΔW_rate is the width change rate;

[0033] The process of obtaining the remaining amount state of the infusion bag includes:

[0034] If FLSI≥80%, then judge that the remaining amount state of the infusion bag is: high remaining amount;

[0035] If 60%≤FLST≤80%, then judge that the remaining amount state of the infusion bag is: normal remaining amount;

[0036] If 10% ≤ FLST ≤ 60%, it is determined that the remaining amount status of the infusion bag is: low remaining amount;

[0037] If FLSI ≤ 10%, it is determined that the remaining amount status of the infusion bag is: need to be replaced.

[0038] Another object of the present invention is to provide an infusion status monitoring device according to the infusion status monitoring method of the intelligent ward described in the present invention, including:

[0039] A relative position regulator, a collection device, a calculation module, a power supply module, an alarm module, and a user control module;

[0040] The relative position regulator is used to fix the relative position between the infusion bag and the collection device, and connect the relative position regulator to the infusion stand;

[0041] The collection device includes: a front RGB image collection device and a side profile information collection device;

[0042] The calculation module, the power supply module, the alarm module, and the user control module are installed inside the relative position regulator;

[0043] The calculation module is used to perform comprehensive judgment by executing the comprehensive judgment model of the remaining amount of the liquid medicine, input the information of the infusion bag, and output the remaining amount status of the infusion bag; issue a warning message according to the remaining amount status of the infusion bag shown;

[0044] The power supply module is powered by a detachable battery;

[0045] The alarm module includes a buzzer and a color display device;

[0046] The user control module includes: a start detection button, a pause detection button, and a reset button.

[0047] Furthermore, the front RGB image collection device is fixedly connected to the relative position regulator through a telescopic fixed arm; the connection line between the front RGB image collection device and the center of the infusion bag is perpendicular to the plane where the front of the infusion bag is located; the side profile information collection device is fixedly connected to the relative position regulator through a telescopic fixed arm; the connection line between the side profile information collection device and the center of the infusion bag is parallel to the plane where the front of the infusion bag is located;

[0048] The collection device is: two front RGB image collection devices respectively located on both sides of the front of the infusion bag and two side profile information collection devices respectively located on both sides of the side of the infusion bag;

[0049] Or

[0050] The acquisition device is: a front RGB image acquisition device and a side profile information acquisition device;

[0051] Or

[0052] The acquisition device is: a front RGB image acquisition device;

[0053] Or

[0054] The acquisition device is: a side profile information acquisition device.

[0055] Another object of the present invention is to provide a computer device, comprising: a memory for storing instructions; a processor for executing the instructions such that the device executes the infusion state monitoring method of the intelligent ward as described in the present invention.

[0056] Another object of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, which when executed, implements the infusion state monitoring method as described in the present invention.

[0057] The beneficial effects of the present invention are as follows:

[0058] 1. Stable monitoring perspective construction: By designing a relative position regulator (compatible connection part, acquisition device fixing part, infusion bag fixing part), rigid fixation of the infusion bag and the acquisition device is achieved, eliminating perspective shift caused by patient movement or body position adjustment, and effectively reducing the liquid level detection error rate.

[0059] 2. Multi-modal high-precision liquid surface recognition: Based on front RGB image preprocessing (denoising, enhancement, cropping) and Canny edge detection and Hough transform algorithms, combined with side near-infrared structured light three-dimensional point cloud compensation (sub-pixel interpolation, motion compensation), it effectively overcomes the influence of specular reflection of semi-transparent materials, bubble interference and complex illumination, and effectively improves the liquid surface recognition accuracy.

[0060] 3. Low-cost dynamic contour modeling: Using low-cost near-infrared structured light to replace the depth camera, real-time monitoring of the deformation state of the infusion bag is achieved through the dynamic contour width set and adaptive mean calculation.

[0061] 4. Intelligent margin comprehensive judgment: Design a liquid level state index (FLSI) model that fuses multiple parameters such as liquid level height, contour width, detection duration, and deformation rate (weight coefficients α, β, γ, δ) to accurately divide high / normal / low margin and the state requiring replacement, effectively reducing the false alarm rate.

[0062] 5. Scalable Deployment Architecture: Modular devices (scalable fixed arms, multi-acquisition device layouts) are used to adapt to various infusion stands and ward scenarios. The lightweight computing module and edge storage solution support low-cost large-scale deployment and are compatible with existing hospital information systems. Description of the Drawings

[0063] Figure 1 It is a schematic flowchart of a method for monitoring the infusion status in an intelligent ward according to the present invention;

[0064] Figure 2 It is a schematic structural diagram of an infusion status monitoring device in an embodiment of the present invention;

[0065] Figure 3 It is a schematic diagram of four layout modes of the acquisition device in an embodiment of the present invention;

[0066] Among them, (a) is a front RGB image acquisition device, (b) is a side profile information acquisition device, (c) is a front RGB image acquisition device and a side profile information acquisition device, and (d) is two front RGB image acquisition devices and two side profile information acquisition devices;

[0067] In the figure, 100: infusion bag, 210: compatible connection part, 220: acquisition device fixing part, 230: infusion bag fixing part, 240: fixed arm, 310: front RGB image acquisition device, 320: side profile information acquisition device. Detailed Embodiment

[0068] The present invention provides a method and system for monitoring the infusion status in an intelligent ward. The following further describes the present invention in detail with reference to the accompanying drawings.

[0069] As Figure 1 shown, an embodiment of the present invention discloses a method for monitoring the infusion status in an intelligent ward, including the following steps:

[0070] Use a relative position regulator to fix the relative position between the infusion bag and the acquisition device, and connect the relative position regulator to the infusion stand; in response to the start monitoring instruction, each acquisition device starts to obtain the infusion bag information corresponding to the acquisition device according to the preset detection rules; in response to the stop monitoring instruction, each acquisition device stops collecting the infusion bag information; input all the infusion bag information into the liquid medicine remaining amount comprehensive judgment model for comprehensive judgment to obtain the remaining amount status of the infusion bag; issue a warning message according to the shown remaining amount status of the infusion bag;

[0071] The obtaining of the infusion bag information corresponding to the acquisition device includes: using the front RGB image acquisition device 310 to acquire the RGB image of the front of the infusion bag, and obtaining the liquid level ratio information based on the RGB image of the front of the infusion bag; using the side profile information acquisition device 320 to acquire the set of infusion bag widths on the side of the infusion bag, and obtaining the average profile information based on the set of infusion bag widths;

[0072] The infusion bag information includes: liquid level ratio information and average profile information;

[0073] The relative position regulator includes a compatible connection part 210, an acquisition device fixing part 220, and an infusion bag fixing part 230 that are connected in sequence; the compatible connection part 210 is used to connect to the existing infusion stand instead of the infusion bag; the acquisition device fixing part 220 includes a plurality of telescopic fixing arms 240, and the acquisition device is fixed on the fixing arms 240; the infusion bag fixing part 230 is fixedly connected to the infusion bag.

[0074] In this embodiment, the infusion bag is fixedly connected to the relative position regulator through the infusion bag fixing part 230; the acquisition device is fixedly connected to the relative position regulator through the acquisition device fixing part 220; the relative position between the infusion bag and the acquisition device is fixed through the relative position regulator, so as to acquire the infusion bag information by the acquisition device in the subsequent process.

[0075] Connecting to the existing infusion stand through the compatible connection part 210 realizes compatibility with various existing infusion stands without further modification. Adjusting the relative position of the acquisition device through the telescopic fixing arms 240 realizes compatibility with acquisition devices with different performance parameters.

[0076] The specific implementation processes of each step are explained separately below.

[0077] The obtaining of the liquid level ratio information based on the RGB image of the front of the infusion bag includes:

[0078] Through the RGB image acquisition device fixed on the front of the infusion bag, the front RGB image of the infusion bag is acquired in real time; the front RGB image is preprocessed to generate standardized image data; the liquid level line position information of the infusion bag is extracted from the standardized image data based on an image algorithm; according to the liquid level line position information, the percentage of the liquid level height in the total height of the infusion bag is calculated to obtain the liquid level ratio information.

[0079] The preprocessing of the front RGB image to generate standardized image data includes:

[0080] Image denoising: Using Gaussian filtering or median filtering to remove image noise;

[0081] Image enhancement: Adjusting the image brightness and contrast to highlight the liquid level line and the edge of the infusion bag;

[0082] Image cropping: Crop the region of interest according to the position of the infusion bag in the image;

[0083] The extracting of the liquid level line position information of the infusion bag from the standardized image data based on the image algorithm includes:

[0084] Use the Canny edge detection algorithm to extract the liquid level line of the infusion bag;

[0085] Detect the straight line feature of the liquid level line through the Hough transform to determine the position of the liquid level line in the image;

[0086] Calculate the pixel distance between the liquid level line and the bottom of the infusion bag as the liquid level line position information;

[0087] The obtaining of the liquid level ratio information includes:

[0088] Liquid level ratio information = Liquid level line position information / Infusion bag height.

[0089] In this embodiment, the front RGB image acquisition device 310 acquires the liquid level height information and judges the remaining amount state of the infusion bag, analyzes the liquid level height through the image processing algorithm, and combines the remaining amount comprehensive judgment model of the infusion bag to achieve precise monitoring and early warning of the remaining amount state of the infusion bag.

[0090] The specific implementation process is as follows:

[0091] Step 1: Installation and initialization

[0092] Fix the front RGB image acquisition device 310 in front of the infusion bag through the acquisition device fixing part of the relative position regulator to ensure that the acquisition device maintains a preset distance and angle from the infusion bag.

[0093] Initialize the front RGB image acquisition device 310, set parameters such as image acquisition resolution and frame rate to ensure that the image clarity meets the liquid level detection requirements.

[0094] Step 2: Image acquisition and preprocessing

[0095] The front RGB image acquisition device 310 continuously acquires the RGB image of the front of the infusion bag.

[0096] Preprocess the acquired image, including:

[0097] Image denoising: Use Gaussian filtering or median filtering to remove image noise.

[0098] Image enhancement: Adjust the image brightness and contrast to highlight the liquid level line and the edge of the infusion bag.

[0099] Image cropping: According to the position of the infusion bag in the image, crop the region of interest (ROI) to reduce the amount of calculation.

[0100] Step 3: Liquid level line detection

[0101] Use an edge detection algorithm (such as Canny edge detection) to extract the liquid level line of the infusion bag.

[0102] Detect the linear feature of the liquid level line through the Hough Transform to determine the position of the liquid level line in the image.

[0103] Calculate the pixel distance between the liquid level line and the bottom of the infusion bag as a preliminary measurement value of the liquid level height.

[0104] In an optional embodiment, it further includes:

[0105] Step 4: Liquid level height calibration;

[0106] According to the known dimensions of the infusion bag (such as height, width) and the pixel ratio in the image, convert the pixel value of the liquid level height into the actual physical height. Consider the tilt angle of the infusion bag and the perspective deviation of the acquisition device during the calibration process, and use geometric transformation to correct the measurement error.

[0107] Step 5: Calculate the pixel distance between the liquid level line and the bottom of the infusion bag as the liquid level line position information;

[0108] The obtaining of the liquid level ratio information includes:

[0109] Liquid level ratio information = Liquid level line position information / Infusion bag height.

[0110] In an optional embodiment of this, when the remaining amount is lower than the preset threshold, the system issues a warning message, including a sound prompt, a light flash, or a remote notification; the warning message is displayed in real time through the intelligent ward monitoring platform to remind medical staff to handle it in time. No specific limitation is made in this embodiment.

[0111] In a preferred embodiment, the alarm module includes a buzzer and a color display device; omitting modules such as the Internet of Things and data wireless transmission is more conducive to reducing power consumption and more meets the actual application scenario requirements of the intelligent ward. It has the following advantages:

[0112] High precision: Ensure the accuracy of the liquid level height measurement through image processing algorithms and geometric calibration.

[0113] Real-time performance: The front RGB image acquisition device continuously monitors and updates the remaining amount status in real time.

[0114] Adaptability: Compatible with infusion bags of different specifications, and adapt to various scenarios through calibration and model adjustment.

[0115] Intelligent: Combining with the remaining liquid medicine comprehensive judgment model to achieve automatic early warning and reduce manual intervention.

[0116] The collection of the infusion bag width set on the side of the infusion bag by using the side profile information collection device 320 includes:

[0117] Using the side profile information collection device 320 to obtain the depth image of the infusion bag; the side profile information collection device 320 includes: a near-infrared structured light projection device and a band-pass filtered CMOS camera;

[0118] Adopting a sub-pixel interpolation algorithm to convert the depth image into a three-dimensional point cloud;

[0119] Performing motion compensation processing on the three-dimensional point cloud to obtain optimized point cloud data; the motion compensation processing includes: Kalman filtering based on an inertial sensor and quaternion attitude solution;

[0120] Obtaining the profile information of the side of the infusion bag based on the optimized point cloud data, and based on the profile information, using a number of equally spaced parallel lines to obtain the infusion bag width set on the side of the infusion bag.

[0121] In this embodiment, the specific steps for obtaining the depth image of the infusion bag include:

[0122] a) Perform laser safety self-check, set the power density ≤ 5 mW / cm², and adjust the camera focal length through a voice coil motor to make the imaging clarity function:

[0123]

[0124] where Gx and Gy are the Sobel gradient operators in the horizontal and vertical directions respectively; the Sobel operator is used to detect edges and measure the image clarity;

[0125] b) Drive the near-infrared laser to project a speckle pattern with a 120 Hz square wave, and the speckle size σ = 1.22λ / NA,

[0126]

[0127] where λ = 850 nm and the numerical aperture NA = 0.05;

[0128] c) Synchronously collect two frames of images and calculate the effective image:

[0129]

[0130] where I on is the frame when the structured light is activated, I off is the frame with only ambient light, and ε = 0.01;

[0131] d) Perform the Census transform within a 9×9 window to generate the bit string C(u, v), and obtain the sub-pixel disparity d through Hamming distance matching sub :

[0132]

[0133] where d is the disparity;

[0134] e) Fuse the inertial sensor data to compensate for pixel displacement:

[0135]

[0136] where f is the focal length of the camera, Z is the depth of the scene, V x is the velocity obtained by integrating the gyroscope data, a x is the acceleration provided by the accelerometer, and Δt is the time interval;

[0137] f) Generate the depth matrix , and calculate the confidence Q:

[0138]

[0139] where N valid is the number of valid disparity measurements, N total is the total number of measurements, is the variance of the disparity;

[0140] g) Trigger the re-sampling mechanism when Q < 0.85, otherwise complete the generation of the infusion bag depth image.

[0141] The specific sub-pixel interpolation algorithm includes:

[0142] Adopt bicubic interpolation in the disparity calculation stage to improve the spatial resolution;

[0143] Introduce the disparity compensation amount:

[0144]

[0145] where a is the instantaneous acceleration of the infusion bag and t is the sampling interval.

[0146] The specific steps of the motion compensation are as follows:

[0147] Establish the state vector:

[0148]

[0149] where x, y, z are the position coordinates of the object in three-dimensional space, and θx, θy, θz are the Euler angles of the object around the x, y, z axes;

[0150] The Mahony complementary filter is used to fuse the gyroscope angular velocity ω and the accelerometer data a;

[0151] The point cloud coordinates are corrected in real time through the coordinate transformation matrix:

[0152]

[0153] Wherein, R is the rotation matrix and T is the translation vector.

[0154] The obtaining of the average profile information based on the infusion bag width set includes:

[0155] Traverse the infusion bag width set in sequence from the top to the bottom of the infusion bag. If the total width from the top to the middle of the infusion bag is greater than a preset first threshold, use all the infusion bag width data from the top to the bottom of the infusion bag to calculate the average value to obtain the average profile information; otherwise, use all the infusion bag width data from the middle to the bottom of the infusion bag to calculate the average value to obtain the average profile information.

[0156] In this embodiment, considering the effect of gravity, the side of the infusion bag presents a spindle shape, narrow at the top and wide at the bottom. In the later stage of infusion, the change in the side width of the infusion bag is mainly concentrated at the bottom of the infusion bag. Therefore, when the total width from the top to the middle of the infusion bag is greater than a preset first threshold, use all the infusion bag width data from the top to the bottom of the infusion bag to calculate the average value to obtain the average profile information; otherwise, use all the infusion bag width data from the middle to the bottom of the infusion bag to calculate the average value to obtain the average profile information. In this embodiment, according to the actual change characteristics of the infusion bag, for the later stage of infusion with higher importance, only the bottom of the infusion bag is processed. Obtaining the average profile information based on the infusion bag width set can be more accurate and safer.

[0157] The inputting of all the infusion bag information into the liquid medicine remaining amount comprehensive judgment model for comprehensive judgment includes:

[0158] Use the liquid medicine remaining amount comprehensive judgment model to calculate the liquid level status index (Fluid Level Status Index, FLSI).

[0159] The liquid medicine remaining amount comprehensive judgment model is:

[0160] FLSI = (W_avg × α) + (H_per × β) + (T_inv × γ) + (ΔW_rate × δ)

[0161] The weight coefficients need to satisfy: α + β + γ + δ = 1

[0162] In the formula, FLSI is the liquid level status index, W_avg is the average width of the infusion bag, H_per is the percentage of the liquid level height, T_inv is the reciprocal of the detection duration, and ΔW_rate is the width change rate;

[0163] The process of obtaining the remaining amount status of the infusion bag includes:

[0164] If FLSI ≥ 80%, it is determined that the remaining amount status of the infusion bag is: high remaining amount;

[0165] If 60% ≤ FLST ≤ 80%, it is determined that the remaining amount status of the infusion bag is: normal remaining amount;

[0166] If 10% ≤ FLST ≤ 60%, it is determined that the remaining amount status of the infusion bag is: low remaining amount;

[0167] If FLSI ≤ 10%, it is determined that the remaining amount status of the infusion bag is: need to be replaced.

[0168] In this embodiment, the specific values of the parameters of the comprehensive judgment model for the remaining amount of the liquid medicine are as follows:

[0169] 1. Average width of the infusion bag (W_avg)

[0170] · Calculation method: Arithmetic average of the width data collected within a single detection period

[0171] · Weight suggestion: α = 0.2

[0172] · Basis: Reflects the overall expansion / shrinkage state of the infusion bag, but there is a lag

[0173] 2. Percentage of the liquid level height (H_per)

[0174] · Calculation method: H_per = (current liquid level height / total height of the container) × 100%

[0175] · Weight suggestion: β = 0.5

[0176] · Basis: Directly reflects the remaining amount of the liquid, and is the core monitoring index

[0177] 3. Reciprocal of the detection duration (T_inv)

[0178] · Calculation method: T_inv = 1 / detection duration (unit: s⁻¹)

[0179] · Weight suggestion: γ = 0.1

[0180] · Basis: The shorter the detection interval, the higher the real-time performance of the data. Refer to the requirements for detection timeliness

[0181] 4. Width change rate (ΔW_rate)

[0182] · Calculation method: ΔW_rate = (W_now - W_prev) / Δt × 100%

[0183] · Weight suggestion: δ = 0.2

[0184] · Basis: Reflects abnormal liquid flow rate and can warn of blockage / leakage

[0185] In this embodiment, the comprehensive judgment model of the remaining liquid medicine comprehensively processes the percentage of the liquid level height after processing the RGB image on the front of the infusion bag and the average width of the infusion bag obtained by the near-infrared structured light projection device and the band-pass filter CMOS camera on the side of the infusion bag. Combining the respective advantages of image processing technology and three-dimensional reconstruction technology. At the same time, a relative position regulator is used to fix the relative position between the infusion bag and the acquisition device to ensure better recognition efficiency and compatibility.

[0186] As Figure 2 shown, another embodiment of the present invention discloses an infusion state monitoring device according to the infusion state monitoring method of the intelligent ward described in the present invention, including:

[0187] A relative position regulator, an acquisition device, a calculation module, a power supply module, an alarm module, and a user control module;

[0188] The relative position regulator is used to fix the relative position between the infusion bag and the acquisition device and connect the relative position regulator to the infusion stand;

[0189] The acquisition device includes: a front RGB image acquisition device 310 and a side profile information acquisition device 320;

[0190] The calculation module, the power supply module, the alarm module, and the user control module are installed inside the relative position regulator;

[0191] The calculation module is used to execute the comprehensive judgment of the comprehensive judgment model of the remaining liquid medicine, input the infusion bag information, and output the remaining liquid state of the infusion bag; issue a warning message according to the shown remaining liquid state of the infusion bag;

[0192] The power supply module is powered by a detachable battery;

[0193] The alarm module includes a buzzer and a color display device;

[0194] The user control module includes: a start detection button, a pause detection button, and a reset button.

[0195] In this embodiment, the actual situation of the intelligent ward is fully considered. The existing infusion racks can be fully compatible through the compatible connection part 210 without physical transformation. The relative position regulator is used to fix the relative position between the infusion bag and the acquisition device, so that both the collected RGB images and depth images maintain a relatively fixed perspective and are as close as possible to the front view perspective, effectively reducing the complexity of the algorithm. Without relying on a complex dynamic compensation process, the detection efficiency is effectively improved and the power consumption of the device operation is reduced. The comprehensive judgment model of the liquid medicine remaining amount is used to comprehensively judge the recognition results of the RGB image and the depth image, which can adapt to the complex light environment in the ward and improve the robustness of the system.

[0196] The front RGB image acquisition device 310 is fixedly connected to the relative position regulator through the telescopic fixing arm 240; the connection line between the front RGB image acquisition device 310 and the center of the infusion bag is perpendicular to the plane where the front of the infusion bag is located; the side profile information acquisition device 320 is fixedly connected to the relative position regulator through the telescopic fixing arm 240; the connection line between the side profile information acquisition device 320 and the center of the infusion bag is parallel to the plane where the front of the infusion bag is located.

[0197] In this embodiment, the four layout modes of the acquisition device of the embodiment of the present invention are as Figure 3 shown;

[0198] Among them, (a) is one front RGB image acquisition device, (b) is one side profile information acquisition device, (c) is one front RGB image acquisition device and one side profile information acquisition device, and (d) is two front RGB image acquisition devices and two side profile information acquisition devices.

[0199] The acquisition device is: two front RGB image acquisition devices 310 respectively located on both sides of the front of the infusion bag and two side profile information acquisition devices 320 respectively located on both sides of the side of the infusion bag 100.

[0200] Or

[0201] The acquisition device is: one front RGB image acquisition device 310 and one side profile information acquisition device 320.

[0202] Or

[0203] The acquisition device is: one front RGB image acquisition device 310.

[0204] Or

[0205] The acquisition device is: one side profile information acquisition device 320.

[0206] In this embodiment, according to the specific application scenario, the user can switch different working modes among the above four modes and can perform special processing according to special situations such as light avoidance requirements and rigid infusion bottles.

[0207] During the process of this embodiment, the relevant implementation environment information is as follows:

[0208] Hardware configuration:

[0209] Front RGB image acquisition device 310: High-definition camera (resolution ≥ 1080p, frame rate ≥ 30fps).

[0210] Relative position regulator: Telescopic fixed arm 240 to ensure the distance between the camera and the infusion bag 100 is 20 - 30 cm.

[0211] Software configuration:

[0212] Image processing algorithm: Implement edge detection, Hough transform, etc. based on the OpenCV library.

[0213] Comprehensive judgment model for liquid medicine remaining amount: Remaining amount calculation module developed based on Python or MATLAB.

[0214] Early warning system:

[0215] Local early warning method: LED lamp and buzzer.

[0216] Another embodiment of the present invention discloses a computer device, including: a memory for storing instructions; a processor for executing the instructions, so that the device executes the infusion state monitoring method of the intelligent ward as described in the present invention.

[0217] Another embodiment of the present invention discloses a computer-readable storage medium, on which a computer program is stored. When the computer program is executed, it implements the infusion state monitoring method of the intelligent ward as described in the present invention.

Claims

1. A method for monitoring the infusion status of an intelligent ward, characterized in that, Including the following steps: Fix the relative position between the infusion bag and the collection device using a relative position adjuster, and connect the relative position adjuster to the infusion stand; in response to the start monitoring instruction, each collection device starts to obtain the infusion bag information corresponding to the collection device according to the preset detection rules; In response to the stop monitoring instruction, each collection device stops collecting the infusion bag information; input all the infusion bag information into the liquid medicine remaining amount comprehensive judgment model for comprehensive judgment to obtain the remaining amount state of the infusion bag; Send out a warning message according to the shown remaining amount state of the infusion bag; The obtaining the infusion bag information corresponding to the collection device includes: using a front RGB image collection device (310) to collect the RGB image of the front of the infusion bag, and obtaining the liquid level proportion information based on the RGB image of the front of the infusion bag; using a side profile information collection device (320) to collect the set of infusion bag widths on the side of the infusion bag, and obtaining the average profile information based on the set of infusion bag widths; The infusion bag information includes: liquid level proportion information and average profile information; The relative position adjuster includes a compatible connection part (210), a collection device fixing part (220), and an infusion bag fixing part (230) connected in sequence; the compatible connection part (210) is used to connect to the existing infusion stand instead of the infusion bag; the collection device fixing part (220) includes a number of telescopic fixing arms (240), and the collection device is fixed on the fixing arms (240); the infusion bag fixing part (230) is fixedly connected to the infusion bag.

2. The infusion status monitoring method of the intelligent ward according to claim 1, wherein The obtaining the liquid level proportion information based on the RGB image of the front of the infusion bag includes: Real-time obtain the front RGB image of the infusion bag through the RGB image collection device fixed on the front of the infusion bag; preprocess the front RGB image to generate standardized image data; extract the liquid level line position information of the infusion bag from the standardized image data based on an image algorithm; calculate the percentage of the liquid level height in the total height of the infusion bag according to the liquid level line position information to obtain the liquid level proportion information.

3. The infusion status monitoring method for the intelligent ward according to claim 2, characterized in that, The preprocessing the front RGB image to generate standardized image data includes: Image denoising: Use Gaussian filtering or median filtering to remove image noise; Image enhancement: Adjust the image brightness and contrast to highlight the liquid level line and the edge of the infusion bag; Image cropping: Crop the region of interest according to the position of the infusion bag in the image; The extracting the liquid level line position information of the infusion bag from the standardized image data based on an image algorithm includes: Use the Canny edge detection algorithm to extract the liquid level line of the infusion bag; Detect the straight line feature of the liquid level line through the Hough transform to determine the position of the liquid level line in the image; Calculate the pixel distance between the liquid level line and the bottom of the infusion bag as the liquid level line position information; The obtaining the liquid level proportion information includes: Liquid level proportion information = Liquid level line position information / Infusion bag height.

4. The infusion status monitoring method of the intelligent ward according to claim 1, characterized in that, The using the side profile information collection device (320) to collect the set of infusion bag widths on the side of the infusion bag includes: Use the side profile information collection device (320) to obtain the depth image of the infusion bag; the side profile information collection device (320) includes: a near-infrared structured light projection device and a band-pass filtered CMOS camera; Convert the depth image into a 3D point cloud using a sub-pixel interpolation algorithm; Perform motion compensation processing on the 3D point cloud to obtain optimized point cloud data; the motion compensation processing includes: Kalman filtering based on an inertial sensor and quaternion attitude resolution; Obtain the contour information of the side of the infusion bag based on the optimized point cloud data, and based on the contour information, use a number of equally spaced parallel lines to obtain a set of infusion bag widths on the side of the infusion bag.

5. The infusion status monitoring method of the intelligent ward according to claim 1, wherein, The obtaining of the average contour information based on the set of infusion bag widths includes: Traverse the set of infusion bag widths in sequence from the top to the bottom of the infusion bag. If the total width from the top to the middle of the infusion bag is greater than a preset first threshold, calculate the average value using all the infusion bag width data from the top to the bottom of the infusion bag to obtain the average contour information; otherwise, calculate the average value using all the infusion bag width data from the middle to the bottom of the infusion bag to obtain the average contour information.

6. The infusion status monitoring method for the intelligent ward according to claim 1, wherein The inputting of all the infusion bag information into the comprehensive judgment model of the liquid medicine remaining amount for comprehensive judgment includes: The comprehensive judgment model of the liquid medicine remaining amount is: FLSI = (W_avg × α) + (H_per × β) + (T_inv × γ) + (ΔW_rate × δ) It is necessary to satisfy the weight coefficients: α + β + γ + δ = 1 In the formula, FLSI is the liquid level state index, W_avg is the average value of the infusion bag width, H_per is the percentage of the liquid level height, T_inv is the reciprocal of the detection duration, and ΔW_rate is the width change rate; The process of obtaining the remaining amount state of the infusion bag includes: If FLSI≥80%, then judge that the remaining amount state of the infusion bag is: high remaining amount; If 60%≤FLST≤80%, then judge that the remaining amount state of the infusion bag is: normal remaining amount; If 10%≤FLST≤60%, then judge that the remaining amount state of the infusion bag is: low remaining amount; If FLSI≤10%, then judge that the remaining amount state of the infusion bag is: need to be replaced.

7. An infusion status monitoring device for the infusion status monitoring method of the intelligent ward according to any one of claims 1-6, characterized in that, Includes: A relative position regulator, a collection device, a calculation module, a power supply module, an alarm module, and a user control module; The relative position regulator is used to fix the relative position between the infusion bag and the collection device, and connect the relative position regulator to the infusion stand; The collection device includes: a front RGB image collection device (310) and a side contour information collection device (320); The calculation module, the power supply module, the alarm module, and the user control module are installed inside the relative position regulator; The calculation module is used to perform comprehensive judgment by executing the comprehensive judgment model of the liquid medicine remaining amount, input the infusion bag information, and output the remaining amount state of the infusion bag; send a warning message according to the remaining amount state of the infusion bag shown; The power supply module is powered by a detachable battery; The alarm module includes a buzzer and a color display device; The user control module includes: a start detection button, a pause detection button, and a reset button.

8. The infusion state monitoring device according to claim 7, wherein, The front RGB image acquisition device (310) is fixedly connected to the relative position adjuster through a telescopic fixed arm (240); the line connecting the front RGB image acquisition device (310) and the center of the infusion bag is perpendicular to the plane where the front of the infusion bag is located; the side profile information acquisition device (320) is fixedly connected to the relative position adjuster through a telescopic fixed arm (240); the line connecting the side profile information acquisition device (320) and the center of the infusion bag is parallel to the plane where the front of the infusion bag is located; The acquisition device is: two front RGB image acquisition devices (310) respectively located on both sides of the front of the infusion bag and two side profile information acquisition devices (320) respectively located on both sides of the side of the infusion bag; Or The acquisition device is: one front RGB image acquisition device (310) and one side profile information acquisition device (320); Or The acquisition device is: one front RGB image acquisition device (310); Or The acquisition device is: one side profile information acquisition device (320).

9. A computer device, characterized in that: Including: A memory for storing instructions; A processor for executing the instructions, so that the device executes the infusion state monitoring method of the intelligent ward described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, it implements the infusion state monitoring method of the intelligent ward described in any one of claims 1-6.

Images