Nursing information visual management system and method

Abstract

The invention relates to the technical field of nursing data management, in particular to a visual management system and method for nursing information. The method comprises the following steps: carrying out infusion multi-parameter sensing signal acquisition on an outlet of an infusion bag to generate real-time liquid medicine flow rate characteristic spectrum data; performing liquid medicine remaining volume calculation on the infusion bag to obtain liquid medicine remaining volume data; according to the real-time liquid medicine flow rate characteristic spectrum data and the liquid medicine remaining volume data, infusion state comprehensive evaluation is carried out, and infusion state evaluation data are generated; performing intelligent regulation and control action decision processing according to the infusion state evaluation data to generate intelligent infusion regulation and control suggestion data; and according to the intelligent infusion regulation and control suggestion data, carrying out manual regulation and execution recording to obtain infusion nursing state record data. By continuously and accurately monitoring the infusion speed of the patient, automatic and intelligent management of infusion nursing is realized, and the infusion nursing efficiency and safety are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of nursing data management, and in particular to a nursing information visualization management system and method. Background Art

[0002] Traditional infusion nursing management relies on nursing staff to regularly patrol and observe each infusion patient. Nursing staff need to manually record key parameters such as infusion speed and remaining amount of medicine, and promptly handle abnormal infusion conditions (such as infusion speed is too fast or too slow, medicine is about to be infused, infusion tube is blocked or leaking, etc.). However, with a large number of patients in the hospital and insufficient nursing staff, it is difficult for nursing staff to monitor and manage the infusion status of each patient in real time. This nursing model is not only inefficient, but also prone to infusion accidents due to omissions or delayed treatment. Summary of the invention

[0003] Based on this, the present invention provides a nursing information visualization management system and method to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for visual management of nursing information includes the following steps: Step S1: integrating the microfluidic chip at the outlet of the infusion bag to obtain a nursing infusion monitoring network; using the nursing infusion monitoring network to collect multi-parameter sensor signals of the drug solution to generate real-time drug solution fluid dynamics parameters; performing drug solution flow rate spectrum characterization processing according to the real-time drug solution fluid dynamics parameters to generate real-time drug solution flow rate characteristic spectrum data; Step S2: continuously photographing the liquid surface of the infusion bag at a periodic frame rate to generate corrected infusion bag image sequence data; calculating the remaining volume of the liquid medicine according to the corrected infusion bag image sequence data to obtain remaining volume data of the liquid medicine; analyzing the flow rate state of the liquid medicine according to the real-time liquid medicine flow rate characteristic spectrum data to obtain adaptive liquid medicine flow rate state data; using the adaptive liquid medicine flow rate state data to perform a comprehensive infusion state evaluation on the remaining volume data of the liquid medicine to generate infusion state evaluation data; Step S3: Perform differentiated infusion risk processing according to the infusion status assessment data, and perform abnormal infusion key state parameter analysis on the real-time drug liquid flow rate characteristic map data to obtain abnormal infusion key state parameters; use the electronic medical record system to query infusion doctor order information and generate patient doctor order structured data; construct a control suggestion action decision model for the infusion bag; use the control suggestion action decision model to perform intelligent control action decision processing on abnormal infusion key state parameters and patient doctor order structured data to generate intelligent infusion control suggestion data; Step S4: Encode manual adjustment action instructions according to the intelligent infusion control suggestion data, and perform visual prompts to obtain visual nursing suggestion data; record manual adjustment execution according to the visual nursing suggestion data, and evaluate the manual adjustment results to obtain infusion nursing status record data.

[0005] The present invention integrates a microfluidic chip at the outlet of the infusion bag to construct a nursing infusion monitoring network, which can collect multi-parameter sensor signals of the drug solution in real time and generate real-time drug solution fluid dynamic parameters and flow rate characteristic maps. This enables nursing staff to continuously and accurately monitor the patient's infusion speed, avoiding the time lag and data inaccuracy caused by traditional manual regular inspections and observations, and effectively preventing medical accidents caused by too fast or too slow infusion. At the same time, the system continuously shoots the infusion bag at a periodic frame rate, and combined with real-time flow rate data, it can accurately calculate the remaining volume of the drug solution and conduct a comprehensive evaluation of the infusion status. This can not only remind nursing staff to replace the drug solution in time, avoid interruption of treatment due to exhaustion of the drug solution and failure to replace it in time, but also perform risk differentiation processing based on the infusion status evaluation data, concentrate limited nursing resources on high-risk patients, and improve nursing efficiency. The system can automatically analyze the real-time drug solution flow rate characteristic map data, identify abnormal key state parameters of infusion, and combine the patient's electronic medical record information to make intelligent control action decisions and generate personalized infusion control suggestions. This intelligent control method not only reduces the workload of nursing staff and the risk of human error, but more importantly, it can be precisely controlled according to the specific conditions of each patient to ensure that the drug has the best therapeutic effect, further improving the patient's treatment effect and safety. For example, when the system detects that the patient's infusion speed is too fast, it will automatically sound an alarm and, based on the patient's doctor's orders and condition, recommend adjusting the infusion speed or suspending the infusion to avoid adverse reactions caused by too fast infusion. For situations where the infusion speed is too slow, the system will also remind the nursing staff in time and give corresponding adjustment suggestions based on the specific situation. The intelligent infusion control suggestions are presented to the nursing staff in an intuitive and easy-to-understand way, which facilitates their quick understanding and execution. The system will encode the manual adjustment action instructions and provide prompts in a visual form, such as a graphical interface, color coding, etc., to avoid operational errors caused by unclear information transmission or misunderstanding. At the same time, the system will also record the execution of manual adjustments and the result evaluation to form a complete infusion care status record data for subsequent quality control and data analysis. Therefore, a method for visual management of nursing information of the present invention collects multi-parameter sensor signals of the liquid medicine at the outlet of the infusion bag in real time through the integration of microfluidic chips and flexible transparent visual modules. At the same time, the liquid medicine level of the infusion bag is continuously photographed at a periodic frame rate, the remaining volume data of the liquid medicine is calculated, and a comprehensive evaluation of the infusion status is performed. The system performs differentiated processing of infusion risks based on the evaluation data, and visually prompts nursing staff to make manual adjustments, records the execution of adjustments, and forms complete infusion nursing status record data. Real-time monitoring, intelligent management and personalized regulation of the infusion process are realized, improving nursing efficiency and patient safety.

[0006] Preferably, step S1 comprises the following steps: Step S11: initializing microfluidic parameters of the infusion bag outlet and integrating the microfluidic chip to obtain a nursing infusion monitoring network; wherein the microfluidic chip includes a piezoresistive sensor, a capacitive sensor and an optical sensor; Step S12: when the infusion starts, the original drug solution multi-parameter sensor signal is collected by using the nursing infusion monitoring network to obtain the original drug solution sensor signal data; Step S13: preprocessing the liquid medicine fluid signal according to the original liquid medicine sensor signal data to generate real-time liquid medicine fluid dynamics parameters; Step S14: Obtaining liquid medicine type coding data; Step S15: calculating the flow rate of the medicine liquid according to the real-time medicine liquid fluid dynamics parameters by using the medicine liquid type coding data to generate real-time medicine liquid flow rate data; Step S16: Performing drug liquid flow rate graph characterization processing according to the real-time drug liquid flow rate data to generate real-time drug liquid flow rate characteristic graph data.

[0007] The present invention introduces the drug liquid type coding data and uses it for the calculation of the drug liquid flow rate. Different types of drug liquids have different physical properties such as viscosity and density, so these factors need to be considered when calculating the flow rate. Through the drug liquid type coding data, the system can perform personalized flow rate calculation according to the characteristics of different drug liquids, thereby improving the accuracy of the calculation results. For example, for drug liquids with higher viscosity, the system will adopt a corresponding correction algorithm to avoid the flow rate calculation deviation caused by viscosity differences. The system performs drug liquid flow rate map characterization processing based on real-time drug liquid flow rate data to generate real-time drug liquid flow rate characteristic map data. This visualized map can intuitively reflect the changing trend of the drug liquid flow rate, which is convenient for nursing staff to monitor the infusion situation in real time. At the same time, the flow rate characteristic map data also provides important data support for subsequent infusion state evaluation and intelligent regulation. For example, by analyzing the flow rate map, abnormal situations such as too fast, too slow or fluctuating infusion speed can be identified, and corresponding intervention measures can be taken in time.

[0008] Preferably, step S16 comprises the following steps: Step S161: performing time axis mapping according to the real-time liquid medicine flow rate data to obtain signal acquisition time axis mapping data; Step S162: performing flow rate value axis mapping according to the real-time liquid medicine flow rate data to obtain flow rate value mapping coordinate data; Step S163: Draw a drug flow rate scatter plot according to the signal acquisition time axis mapping data and the flow rate value mapping coordinate data to obtain a preliminary drug solution flow rate characteristic map; Step S164: performing flow rate pattern matching on the real-time fluid dynamics parameters of the drug liquid through a preset normal flow rate pattern library based on the drug liquid type coding data to obtain drug liquid flow rate pattern matching data; Step S165: performing a drug liquid flow rate deviation analysis on the preliminary drug liquid flow rate characteristic map through drug liquid flow rate pattern matching data, and performing a drug liquid flow rate fluctuation abnormality detection to generate drug liquid flow rate abnormality marking data; Step S166: using the abnormal marking data of the drug liquid flow rate to mark abnormal areas of the preliminary drug liquid flow rate characteristic map, thereby obtaining real-time drug liquid flow rate characteristic map data.

[0009] The present invention converts real-time drug flow rate data into a visual scatter plot by mapping the time axis and the flow rate value axis, forming a preliminary drug flow rate characteristic map. This allows nursing staff to intuitively observe the changes in the drug flow rate over time and preliminarily determine whether there are abnormalities. A preset normal flow rate pattern library is introduced, and flow rate pattern matching is performed based on the drug type coding data. This means that the system can determine whether the current flow rate conforms to the normal pattern based on the characteristics of different drug liquids. For example, some drugs require uniform infusion, while some drugs require rapid infusion within a specific time period. By comparing with the normal flow rate pattern library, the system can quickly identify flow rate changes that are inconsistent with expectations, such as infusion speeds that are too fast, too slow, or too volatile.

[0010] Preferably, step S2 comprises the following steps: Step S21: attaching a flexible transparent visual module to the outside of the infusion bag, and continuously photographing the liquid surface of the drug at a periodic frame rate to obtain original infusion bag image sequence data; wherein the flexible transparent visual module has a built-in micro camera; Step S22: performing image distortion correction on the original infusion bag image sequence data to generate corrected infusion bag image sequence data; Step S23: identifying the liquid surface of the drug solution according to the corrected infusion bag image sequence data to obtain liquid surface segmentation mask data; Step S24: calculating the remaining volume of the liquid medicine according to the liquid surface segmentation mask data to obtain the remaining volume data of the liquid medicine; Step S25: performing a drug liquid flow rate state analysis according to the real-time drug liquid flow rate characteristic spectrum data to obtain adaptive drug liquid flow rate state data; Step S26: Calculating the remaining infusion time of the residual volume data of the drug solution by using the adaptive drug solution flow rate state data to obtain the remaining infusion time data; Step S27: Perform a comprehensive evaluation of the infusion status according to the remaining infusion time data and the adaptive drug liquid flow rate status data to generate infusion status evaluation data.

[0011] The present invention can accurately segment the liquid surface area and calculate the remaining volume of the liquid medicine by identifying the liquid surface of the corrected image sequence. Compared with the traditional manual visual estimation, this volume calculation method based on image analysis is more accurate, avoids human errors, and improves the accuracy of infusion management. At the same time, the system combines the real-time liquid medicine flow rate characteristic spectrum data to analyze the liquid medicine flow rate state and obtain adaptive liquid medicine flow rate state data. This means that the system can dynamically predict the remaining infusion time according to the current infusion speed. Using the adaptive liquid medicine flow rate state data to calculate the remaining infusion time of the liquid medicine remaining volume data can more accurately predict the completion time of the infusion. This not only makes it convenient for nursing staff to arrange their work, but more importantly, it can warn in advance that the liquid medicine is about to be infused, avoid interrupting treatment due to depletion of liquid medicine, and improve the patient's medication safety.

[0012] Preferably, step S24 comprises the following steps: Step S241: extracting liquid surface edge pixel points according to the liquid surface segmentation mask data to obtain liquid surface edge pixel point data; Step S242: performing sub-pixel edge positioning on the liquid surface segmentation mask data by using the liquid surface edge pixel point data to obtain floating point coordinate data of the liquid surface edge point; Step S243: Obtain the geometric model of the inner wall of the infusion bag, Step S244: constructing an image pose space coordinate system according to the geometric model of the inner wall of the infusion bag to generate infusion bag space coordinate system data; Step S245: setting camera shooting parameters according to the flexible transparent vision module, and performing coordinate transformation on the floating point coordinate data of the liquid surface edge point through the spatial coordinate system data of the infusion bag, thereby obtaining the three-dimensional coordinate data of the liquid surface edge point; Step S246: Use the three-dimensional coordinate data of the liquid surface edge point to perform liquid surface curve fitting on the geometric model of the inner wall of the infusion bag, and calculate the remaining volume of the drug liquid to obtain the remaining volume data of the drug liquid.

[0013] The present invention extracts liquid surface edge pixel points from liquid surface segmentation mask data, and further performs sub-pixel edge positioning to obtain floating-point coordinate data of liquid surface edge points. Compared with traditional pixel-level positioning, sub-pixel-level positioning can more accurately determine the position of the liquid surface edge, thereby improving the accuracy of volume calculation. The geometric model of the inner wall of the infusion bag is obtained, and the image posture space coordinate system is constructed according to the model. This provides a reference system for subsequent three-dimensional coordinate transformation, so that the liquid surface edge points in the two-dimensional image can be converted into three-dimensional space coordinates. Through the camera shooting parameters set by the flexible transparent visual module and the spatial coordinate system data, the system converts the floating-point coordinate data of the liquid surface edge points into three-dimensional coordinate data of the liquid surface edge points. The surface fitting method based on the three-dimensional model can more accurately describe the shape of the liquid surface, especially when the infusion bag is deformed, it can still maintain a high calculation accuracy. Compared with the traditional volume estimation method based on two-dimensional images, this method can effectively avoid errors caused by factors such as infusion bag deformation and illumination, thereby more accurately calculating the remaining amount of liquid medicine.

[0014] Preferably, step S246 includes the following steps: The liquid surface edge point 3D coordinate data is used to fit the liquid surface curve of the inner wall geometric model of the infusion bag, and the liquid surface of the infusion bag is 3D reconstructed to obtain real-time liquid 3D model data; Perform model point cloud processing according to the real-time liquid medicine three-dimensional model data to obtain liquid medicine point cloud data; Perform bounding box processing on the liquid medicine point cloud data to obtain liquid medicine bounding box data; Perform spatial grid division according to the liquid medicine bounding box data to obtain model spatial grid data; Mark the liquid occupancy according to the model space grid data and the liquid point cloud data, and perform grid connected domain analysis to obtain the liquid connected domain data; The maximum connected domain of the model space grid data is extracted through the drug liquid connected domain data, and the residual volume of the drug liquid is accurately calculated to obtain the residual volume data of the drug liquid.

[0015] After the present invention converts the three-dimensional model data into liquid medicine point cloud data, the system performs bounding box processing on the point cloud data. The bounding box can quickly define the spatial range where the liquid medicine is located, effectively reduce the complexity of subsequent calculations, and improve calculation efficiency. Then, the system performs spatial grid division according to the liquid medicine bounding box data, and discretizes the three-dimensional space where the liquid medicine is located into small grids. This discretization processing method can simplify complex geometric calculations and improve calculation efficiency. The system performs liquid medicine occupancy marking on the liquid medicine point cloud data and performs grid connected domain analysis. This can effectively distinguish between liquid medicine and other objects, such as the wall of an infusion bag, thereby avoiding counting non-liquid medicine parts into volume calculations. By extracting the maximum connected domain, the system can exclude some small, isolated point cloud areas, such as bubbles or impurities, and further improve the accuracy of calculation. The maximum connected domain is extracted from the model space grid data through the liquid medicine connected domain data, and the liquid medicine residual volume is accurately calculated to obtain the liquid medicine residual volume data. This volume calculation method based on three-dimensional point clouds and spatial grids can more accurately handle complex situations such as irregular liquid surface shapes and infusion bag deformations.

[0016] Preferably, step S3 comprises the following steps: Step S31: judging the infusion risk level according to the infusion status assessment data; when the remaining infusion time is less than the first nursing time threshold and the drug flow rate fluctuates abnormally, marking the infusion status assessment data as a high-risk infusion status, and automatically implementing a high-priority alarm push; Step S32: when the remaining infusion time is less than the second nursing time threshold and the drug flow rate fluctuates abnormally, the infusion status assessment data is marked as a medium-risk infusion status, and the electronic medical record system is used to query the patient's infusion order information to obtain the patient's infusion order data; otherwise, the infusion status assessment data is marked as a low-risk infusion status, and no related operations are performed; wherein the second nursing time threshold is greater than the first nursing time threshold; Step S33: formatting the patient's infusion order data to generate structured patient order data; Step S34: performing abnormal infusion key state parameter analysis on the real-time drug liquid flow rate characteristic spectrum data through the medium-risk infusion state to obtain abnormal infusion key state parameters; Step S35: performing transfer learning on the convolutional neural network model through the preset infusion doctor's order-infusion action mapping rule, and performing spatial probability distribution processing of the control suggestion action on the infusion bag to train the control suggestion action decision model; Step S36: Use the control suggestion action decision model to perform intelligent control action decision processing on the patient's medical order structured data and abnormal infusion key state parameters, and perform action safety verification to generate intelligent infusion control suggestion data.

[0017] The present invention determines the infusion risk level based on the infusion status assessment data, and adopts different treatment strategies according to the risk level. For high-risk infusion status, the system will automatically implement high-priority alarm push, and promptly remind nursing staff to intervene to avoid serious consequences due to negligence or delayed treatment. According to the specific doctor's orders of each patient, personalized risk assessment and regulation suggestions are carried out to avoid a "one-size-fits-all" approach, thereby improving the accuracy and effectiveness of regulation. For low-risk infusion status, the system does not perform additional operations to avoid excessive intervention and reduce the workload of nursing staff. The convolutional neural network model is transferred and learned through the preset infusion doctor's order-infusion action mapping rule, and is trained in combination with the probability distribution of the regulation suggestion action space of the infusion bag. The system can intelligently give regulation suggestions based on the patient's doctor's order and the current infusion status. For example, when the system determines that the patient's infusion speed is too fast, it will recommend adjusting the infusion speed or suspending the infusion according to the patient's doctor's order and condition.

[0018] Preferably, step S4 comprises the following steps: Step S41: encoding manual adjustment action instructions according to the intelligent infusion control suggestion data to obtain manual adjustment action instruction data; Step S42: The recipient of the nursing staff is determined by the intelligent infusion control suggestion data, and the manual adjustment action instruction data is visually prompted to obtain visual nursing suggestion data; Step S43: Based on the visualized nursing suggestion data, the nursing staff manually adjusts the dripping rate of the liquid medicine using the infusion regulator to obtain the manually adjusted liquid medicine, and at the same time collects the nursing staff's confirmation / adjustment operation instructions to obtain the manual operation instruction data; Step S44: binding the patient's identity according to the manual operation instruction data, and performing manual adjustment execution record to obtain manual adjustment execution record data; Step S45: monitoring the state of the adjusted drug solution based on the nursing infusion monitoring network to obtain the adjusted drug solution fluid dynamics parameters; performing manual adjustment deviation calculation based on the adjusted drug solution fluid dynamics parameters to obtain manual adjustment deviation data; Step S46: Evaluate the manual adjustment result of the manual adjustment deviation data based on the preset deviation threshold, and bind the manual adjustment execution record data to the nursing record to obtain the infusion nursing status record data.

[0019] The present invention converts the intelligent infusion control suggestion data into manual adjustment action instruction data, and encodes it to facilitate the understanding and execution of nursing staff. At the same time, the system will determine the recipient according to the intelligent infusion control suggestion data, and prompt the corresponding nursing staff with the manual adjustment action instruction in a visual form, avoiding information transmission errors or delays, and improving operation efficiency. After the nursing staff performs manual adjustment operations according to the visual nursing suggestions, the system will collect the nursing staff's confirmation / adjustment operation instructions, bind the patient's identity, and generate manual adjustment execution record data. This step realizes the full recording of manual operations, improves the transparency and traceability of operations, monitors the adjusted liquid state, and calculates the manual adjustment deviation. By comparing with the preset deviation threshold, the system can evaluate whether the manual adjustment has achieved the expected effect. For example, if the system recommends reducing the infusion rate to 20 drops per minute, but the actual adjusted rate is 25 drops per minute, the system will record the deviation and prompt the nursing staff to make adjustments. By analyzing manual adjustment deviation data, we can identify which types of infusion controls are prone to deviations, and improve control strategies or provide more detailed operational guidance in a targeted manner, thereby continuously improving the accuracy and safety of infusion management.

[0020] Preferably, step S45 includes the following steps: Step S451: estimating the theoretical flow rate change of the manual adjustment action instruction data to obtain theoretical flow rate change data; Step S452: monitoring the state of the adjusted drug solution based on the nursing infusion monitoring network to obtain the fluid dynamics parameters of the adjusted drug solution; Step S453: analyzing the actual change of the flow rate of the drug solution according to the adjusted fluid dynamics parameters of the drug solution to generate actual change data of the flow rate of the drug solution; Step S454: Calculate the manual flow adjustment deviation of the actual liquid medicine flow change data using the theoretical flow change data to obtain manual adjustment deviation data.

[0021] The present invention calculates the deviation of manual flow regulation based on the actual liquid medicine flow change data through theoretical flow change data. That is, the expected regulation effect is compared with the actual regulation effect, so as to quantify the deviation of manual regulation. For example, if the flow should be reduced by 10ml / min after adjustment in theory, but the actual monitored flow is only reduced by 5ml / min, the system will calculate a deviation value of 5ml / min. This deviation calculation method based on the comparison between theory and practice can more accurately reflect the accuracy of manual regulation and avoid the errors caused by relying solely on experience or subjective judgment. By quantifying the deviation value, the system can more objectively evaluate the effect of manual regulation and promptly discover potential risks. For example, if the deviation value exceeds the preset threshold, the system will sound an alarm to remind the nursing staff to readjust, thereby ensuring the accuracy and safety of the infusion.

[0022] The present invention also provides a nursing information visualization management system, which executes the above-mentioned nursing information visualization management method, and the nursing information visualization management system comprises: The drug liquid flow rate monitoring module is used to integrate the microfluidic chip at the outlet of the infusion bag to obtain a nursing infusion monitoring network; the nursing infusion monitoring network is used to collect multi-parameter sensor signals of the drug liquid to generate real-time drug liquid fluid dynamics parameters; the drug liquid flow rate spectrum is characterized and processed according to the real-time drug liquid fluid dynamics parameters to generate real-time drug liquid flow rate characteristic spectrum data; The infusion state evaluation module is used to continuously shoot the liquid surface of the infusion bag at a periodic frame rate to generate corrected infusion bag image sequence data; calculate the remaining volume of the liquid medicine according to the corrected infusion bag image sequence data to obtain the remaining volume data of the liquid medicine; analyze the liquid medicine flow rate state according to the real-time liquid medicine flow rate characteristic spectrum data to obtain adaptive liquid medicine flow rate state data; use the adaptive liquid medicine flow rate state data to perform a comprehensive infusion state evaluation on the remaining volume data of the liquid medicine to generate infusion state evaluation data; The intelligent control decision module is used to perform differentiated processing of infusion risks based on infusion status assessment data, and to analyze the key state parameters of abnormal infusion on the real-time drug flow rate characteristic map data to obtain the key state parameters of abnormal infusion; to use the electronic medical record system to query the infusion doctor's order information and generate patient doctor's order structured data; to build a control suggestion action decision model for the infusion bag; to use the control suggestion action decision model to perform intelligent control action decision processing on the key state parameters of abnormal infusion and the patient doctor's order structured data to generate intelligent infusion control suggestion data; The nursing advice visualization module is used to encode manual adjustment action instructions according to the intelligent infusion control advice data, and provide visual prompts to obtain visualized nursing advice data; it records manual adjustment execution according to the visualized nursing advice data, and evaluates the manual adjustment results, thereby obtaining infusion nursing status record data. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic diagram of the steps of the method for visual management of nursing information according to the present invention; Figure 2 for Figure 1 Detailed implementation steps of step S3 in FIG. Figure 3 for Figure 1 Detailed implementation steps of step S4 in FIG. The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0024] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are 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 technicians in this field without creative work are within the scope of protection of the present invention.

[0025] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0026] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0027] To achieve this, please refer to Figures 1 to 3 The present invention provides a method for visual management of nursing information, comprising the following steps: Step S1: integrating the microfluidic chip at the outlet of the infusion bag to obtain a nursing infusion monitoring network; using the nursing infusion monitoring network to collect multi-parameter sensor signals of the drug solution to generate real-time drug solution fluid dynamics parameters; performing drug solution flow rate spectrum characterization processing according to the real-time drug solution fluid dynamics parameters to generate real-time drug solution flow rate characteristic spectrum data; Step S2: continuously photographing the liquid surface of the infusion bag at a periodic frame rate to generate corrected infusion bag image sequence data; calculating the remaining volume of the liquid medicine according to the corrected infusion bag image sequence data to obtain remaining volume data of the liquid medicine; analyzing the flow rate state of the liquid medicine according to the real-time liquid medicine flow rate characteristic spectrum data to obtain adaptive liquid medicine flow rate state data; using the adaptive liquid medicine flow rate state data to perform a comprehensive infusion state evaluation on the remaining volume data of the liquid medicine to generate infusion state evaluation data; Step S3: Perform differentiated infusion risk processing according to the infusion status assessment data, and perform abnormal infusion key state parameter analysis on the real-time drug liquid flow rate characteristic map data to obtain abnormal infusion key state parameters; use the electronic medical record system to query infusion doctor order information and generate patient doctor order structured data; construct a control suggestion action decision model for the infusion bag; use the control suggestion action decision model to perform intelligent control action decision processing on abnormal infusion key state parameters and patient doctor order structured data to generate intelligent infusion control suggestion data; Step S4: Encode manual adjustment action instructions according to the intelligent infusion control suggestion data, and perform visual prompts to obtain visual nursing suggestion data; record manual adjustment execution according to the visual nursing suggestion data, and evaluate the manual adjustment results to obtain infusion nursing status record data.

[0028] In an embodiment of the present invention, the method for visual management of nursing information includes the following steps: The method for visual management of nursing information comprises the following steps: Step S1: integrating the microfluidic chip at the outlet of the infusion bag to obtain a nursing infusion monitoring network; using the nursing infusion monitoring network to collect multi-parameter sensor signals of the drug solution to generate real-time drug solution fluid dynamics parameters; performing drug solution flow rate spectrum characterization processing according to the real-time drug solution fluid dynamics parameters to generate real-time drug solution flow rate characteristic spectrum data; In an embodiment of the present invention, a microfluidic chip is integrated into the outlet of the infusion bag. The microfluidic chip contains multiple micro sensors, such as a pressure sensor, a flow sensor, and a temperature sensor. When the liquid medicine flows through the microfluidic chip, the pressure sensor can sense the pressure change when the liquid medicine flows through, the flow sensor can measure the flow rate of the liquid medicine, and the temperature sensor can measure the temperature of the liquid medicine. These sensors transmit the collected data in real time to the nursing infusion monitoring network, which is composed of multiple wireless transmission nodes and is responsible for data collection, integration and transmission. The nursing infusion monitoring network preprocesses the received raw liquid medicine fluid dynamic parameters, including data cleaning, filtering and unit conversion. For example, invalid data points are removed, high-frequency noise is filtered out, and pressure data is converted from Pascal to kilopascal, etc. Then, based on the preprocessed data, a specific algorithm is used to characterize the liquid medicine flow rate map. For example, the average flow rate per minute can be calculated, and parameters such as flow resistance can be calculated based on pressure and flow data to form real-time liquid medicine flow rate characteristic map data containing parameters such as time, average flow rate, pressure, temperature and flow resistance.

[0029] Step S2: continuously photographing the liquid surface of the infusion bag at a periodic frame rate to generate corrected infusion bag image sequence data; calculating the remaining volume of the liquid medicine according to the corrected infusion bag image sequence data to obtain remaining volume data of the liquid medicine; analyzing the flow rate state of the liquid medicine according to the real-time liquid medicine flow rate characteristic spectrum data to obtain adaptive liquid medicine flow rate state data; using the adaptive liquid medicine flow rate state data to perform a comprehensive infusion state evaluation on the remaining volume data of the liquid medicine to generate infusion state evaluation data; In an embodiment of the present invention, a pre-trained image segmentation model is used to process each image to identify the boundary between the liquid level of the medicine and the infusion bag. For example, a U-Net network model based on deep learning is used for image segmentation to accurately extract the medicine area. Then, the actual volume of the medicine is calculated according to the calibration parameters and geometric model of the camera. For example, a cylindrical approximation model can be used to calculate the remaining volume of the medicine according to the liquid level height and the radius of the infusion bag, and generate the remaining volume data of the medicine. The flow rate state of the medicine is analyzed according to the real-time liquid flow rate characteristic spectrum data generated in step S1. For example, by analyzing the trend, fluctuation amplitude and other characteristics of the flow rate curve, it is determined whether the current flow rate is within the preset normal range. For example, if the current flow rate is 50 ml / h, and the flow rate range set according to the doctor's order is 40-60 ml / h, the current flow rate state is determined to be "normal". Combined with the flow rate historical data, the flow rate change trend for a period of time in the future is predicted. Using these analysis results, adaptive liquid flow rate status data is generated, such as "normal", "too fast", "too slow" or "blocked".

[0030] Step S3: Perform differentiated infusion risk processing according to the infusion status assessment data, and perform abnormal infusion key state parameter analysis on the real-time drug liquid flow rate characteristic map data to obtain abnormal infusion key state parameters; use the electronic medical record system to query infusion doctor order information and generate patient doctor order structured data; construct a control suggestion action decision model for the infusion bag; use the control suggestion action decision model to perform intelligent control action decision processing on abnormal infusion key state parameters and patient doctor order structured data to generate intelligent infusion control suggestion data; In an embodiment of the present invention, for example, if the evaluation result shows that the medicine is about to be infused, the system will issue a reminder of "changing the medicine" 5 minutes in advance; if the evaluation result shows that the flow rate is too fast or too slow, the system will issue an alarm of the corresponding level and give preliminary suggestions, such as "please check whether the infusion line is unobstructed" or "please adjust the drip rate". At the same time, the system will deeply analyze the real-time medicine flow rate characteristic map data to identify the key state parameters of abnormal infusion. For example, by detecting the mutation point, fluctuation frequency and other characteristics of the flow rate curve, it is determined whether there are abnormal conditions such as bubbles, blockages or leaks. When it is detected that the flow rate suddenly drops to 0 and lasts for more than 1 minute, it is judged as "blockage" and the time and related parameters of the event are recorded. The system is connected to the hospital's electronic medical record system through an interface to query the current patient's infusion order information. For example, if the doctor's order for patient Zhang San is "0.9% sodium chloride injection 500 ml, intravenous drip, 60 drops / minute", the system will structure the order information, extract key information such as drug name, dosage, route of administration and drip rate, and generate patient order structured data. For example, drug name: 0.9% sodium chloride injection; dosage: 500 ml; route of administration: intravenous drip; drip rate: 60 drops / minute. It can be built based on expert experience and clinical guidelines, or it can be learned from historical data through machine learning algorithms. For example, if an infusion line is detected to be blocked and the patient is receiving an irritant drug, the model will recommend "stop the infusion immediately and replace the infusion line"; if the flow rate is detected to be too fast and the patient is receiving an antihypertensive drug, the model will recommend "reducing the drip rate to 40 drops / minute".

[0031] Step S4: Encode manual adjustment action instructions according to the intelligent infusion control suggestion data, and perform visual prompts to obtain visual nursing suggestion data; record manual adjustment execution according to the visual nursing suggestion data, and evaluate the manual adjustment results to obtain infusion nursing status record data.

[0032] In the embodiment of the present invention, according to the intelligent infusion control suggestion data generated in step S3, the system converts it into an adjustment action instruction that can be understood by humans, and prompts it through a visual interface. For example, if the control suggestion is "reduce the drip rate to 40 drops / minute", the system will display the corresponding text prompt on the monitoring screen of the nurse station, and highlight "40 drops / minute" in red. At the same time, the system can also remind the nurse to operate by voice broadcast. This information constitutes the visual nursing suggestion data. According to the intelligent infusion control suggestion data generated in step S3, the system converts it into an adjustment action instruction that can be understood by humans, and prompts it through a visual interface. For example, if the control suggestion is "reduce the drip rate to 40 drops / minute", the system will display the corresponding text prompt on the monitoring screen of the nurse station, and highlight "40 drops / minute" in red. At the same time, the system can also remind the nurse to operate by voice broadcast. This information constitutes the visual nursing suggestion data. The system integrates the key data of the entire infusion process, including the characteristic map of the liquid flow rate, the infusion state evaluation result, the control suggestion, the manual adjustment record and the adjustment result evaluation, into the infusion nursing state record data, and manages and stores it in the database.

[0033] Preferably, step S1 comprises the following steps: Step S11: initializing microfluidic parameters of the infusion bag outlet and integrating the microfluidic chip to obtain a nursing infusion monitoring network; wherein the microfluidic chip includes a piezoresistive sensor, a capacitive sensor and an optical sensor; Step S12: when the infusion starts, the original drug solution multi-parameter sensor signal is collected by using the nursing infusion monitoring network to obtain the original drug solution sensor signal data; Step S13: preprocessing the liquid medicine fluid signal according to the original liquid medicine sensor signal data to generate real-time liquid medicine fluid dynamics parameters; Step S14: Obtaining liquid medicine type coding data; Step S15: calculating the flow rate of the medicine liquid according to the real-time medicine liquid fluid dynamics parameters by using the medicine liquid type coding data to generate real-time medicine liquid flow rate data; Step S16: Performing drug liquid flow rate graph characterization processing according to the real-time drug liquid flow rate data to generate real-time drug liquid flow rate characteristic graph data.

[0034] In an embodiment of the present invention, the microfluidic parameters of the infusion bag outlet are initialized. The initialization content includes setting the sampling frequency, range and calibration parameters of each sensor in the microfluidic chip. For example, the sampling frequency of the piezoresistive sensor is set to 10 Hz, the range is 0-50 kPa, and it is calibrated according to a standard pressure source; the sampling frequency of the capacitive sensor is set to 5 Hz, and it is calibrated according to a liquid with a known dielectric constant; the sampling frequency of the optical sensor is set to 1 Hz, and it is calibrated according to a standard turbidity solution. After the initialization is completed, the microfluidic chip is integrated into the outlet of the infusion bag. The microfluidic chip integrates a piezoresistive sensor, a capacitive sensor and an optical sensor. The microfluidic chip is connected to a micro data processing unit through a wire, which is responsible for analog-to-digital conversion and preliminary processing of the sensor signal. The micro data processing units of multiple infusion bags are connected to the same wireless access point through a wireless communication module (such as a Wi-Fi module) to form a nursing infusion monitoring network. For example, there are 10 infusion positions in a ward, and the microfluidic chip at each infusion position is connected to the wireless router in the ward through its own Wi-Fi module, so as to realize centralized data collection and monitoring. When the infusion starts, the liquid medicine flows through the outlet of the infusion bag integrated with the microfluidic chip. The piezoresistive sensor, capacitive sensor and optical sensor inside the microfluidic chip start working at the same time, collecting parameter signals such as pressure, dielectric constant and transmittance of the liquid medicine in real time. For example, at a certain moment, the voltage signal collected by the piezoresistive sensor is 2.5 volts, corresponding to a pressure of 10 kilopascals; the capacitance value collected by the capacitive sensor is 100 pF, corresponding to the dielectric constant of a specific liquid medicine; the light intensity signal collected by the optical sensor is 80%, corresponding to the transmittance of the liquid medicine. These original analog signals are pre-processed by the signal conditioning circuit inside the microfluidic chip, such as amplification and filtering, and then converted into digital signals through an analog-to-digital converter. For example, the 2.5 volt voltage signal output by the piezoresistive sensor is converted into a digital quantity of 2048 after passing through a 12-bit analog-to-digital converter. These digital signals are packaged into data frames and transmitted to the central server in real time through the nursing infusion monitoring network. The data frame contains information such as timestamp, sensor type, sensor number, and collected data value. According to the sensor type and number in the data frame, the data is classified and stored. Then, the raw data is calibrated and the unit is converted. For example, according to the pre-set calibration curve of the piezoresistive sensor, the digital quantity 2048 is converted to the actual pressure value 10 kPa; the capacitance value of the capacitive sensor is converted to the dielectric constant; the light intensity signal of the optical sensor is converted to the transmittance. For example, the sliding average filtering algorithm is used to smooth the pressure data and remove high-frequency noise. Then, based on the processed data, the fluid dynamics parameters of the liquid medicine, such as flow rate and flow resistance, are calculated. For example, the Hagen-Poiseuille formula can be used to estimate the flow rate based on the relationship between pressure and flow.If the inner diameter and length of the infusion line and the viscosity of the liquid medicine are known, the flow resistance can be calculated based on the pressure gradient. For example, each infusion bag is affixed with a barcode that contains the unique code of the liquid medicine. Before the infusion begins, the nurse uses a barcode scanner to scan the barcode on the infusion bag. The system automatically decodes the barcode information and queries the name, specification, density, and viscosity of the liquid medicine corresponding to the code from the pre-established liquid medicine database. For liquid medicines with lower viscosity, the Hagen-Poiseuille formula can be used for calculation; for liquid medicines with higher viscosity, the modified Hagen-Poiseuille formula or other empirical formulas can be used for calculation. According to the selected model and the fluid dynamics parameters of the liquid medicine, the corrected real-time liquid medicine flow rate data is calculated. The real-time flow rate data is arranged in chronological order, and some statistical features are calculated, such as the average flow rate, the standard deviation of the flow rate, and the rate of change of the flow rate. For example, the average flow rate in the past minute is calculated to be 60.5 ml / hour, the standard deviation is 1.2 ml / hour, the maximum flow rate is 62.5 ml / hour, and the minimum flow rate is 58.5 ml / hour. For example, by plotting the flow rate data over a period of time into a curve, the changing trend of the liquid medicine flow rate can be intuitively displayed, forming a real-time liquid medicine flow rate characteristic map.

[0035] Preferably, step S16 comprises the following steps: Step S161: performing time axis mapping according to the real-time liquid medicine flow rate data to obtain signal acquisition time axis mapping data; Step S162: performing flow rate value axis mapping according to the real-time liquid medicine flow rate data to obtain flow rate value mapping coordinate data; Step S163: Draw a drug flow rate scatter plot according to the signal acquisition time axis mapping data and the flow rate value mapping coordinate data to obtain a preliminary drug solution flow rate characteristic map; Step S164: performing flow rate pattern matching on the real-time fluid dynamics parameters of the drug liquid through a preset normal flow rate pattern library based on the drug liquid type coding data to obtain drug liquid flow rate pattern matching data; Step S165: performing a drug liquid flow rate deviation analysis on the preliminary drug liquid flow rate characteristic map through drug liquid flow rate pattern matching data, and performing a drug liquid flow rate fluctuation abnormality detection to generate drug liquid flow rate abnormality marking data; Step S166: using the abnormal marking data of the drug liquid flow rate to mark abnormal areas of the preliminary drug liquid flow rate characteristic map, thereby obtaining real-time drug liquid flow rate characteristic map data.

[0036] In an embodiment of the present invention, the real-time liquid medicine flow rate data is mapped to the time axis according to the order of the acquisition time. For example, assuming that the flow rate data is collected once per second, the time corresponding to the first data point is 0 seconds, the time corresponding to the second data point is 1 second, and so on. The real-time liquid medicine flow rate data is mapped to the flow rate value axis according to the numerical value. For example, assuming that the unit of the flow rate is milliliters per second, the flow rate value is converted into a coordinate value on the flow rate value axis. The scale of the flow rate value axis can be set according to the range of the flow rate. For example, if the range of the flow rate is between 0 and 10 milliliters per second, the scale of the flow rate value axis can be set to 1 milliliter per second per grid. In this way, the flow rate value mapping coordinate data is obtained, and each flow rate data point is corresponding to a specific position on the flow rate value axis. With the signal acquisition time axis mapping data as the horizontal coordinate and the flow rate value mapping coordinate data as the vertical coordinate, each data point is plotted in a two-dimensional coordinate system to form a scatter plot. For example, if the time axis coordinate of the first data point is 0 and the flow rate value coordinate corresponds to 60 ml / h, a point is drawn at the position (0,60) in the coordinate system; if the time axis coordinate of the second data point is 6 and the flow rate value coordinate corresponds to 62 ml / h, a point is drawn at the position (6,62) in the coordinate system. Similarly, all data points are drawn in the coordinate system to form a drug flow rate scatter plot. According to the drug type coding data, the corresponding flow rate mode is selected from the normal flow rate mode library. For example, if the drug type coding data is "NS001", corresponding to 0.9% sodium chloride injection, the normal flow rate mode of this type of drug is selected. Then, the real-time drug fluid dynamics parameters, including flow rate data and other sensor data, such as pressure, are compared with the selected normal flow rate mode to determine whether the current flow rate conforms to the normal mode. For example, the average flow rate and fluctuation range within the current 1 minute are compared with the normal flow rate mode to calculate the deviation value. If the deviation value is within the allowable range, it is considered that the current flow rate conforms to the normal mode, and the drug flow rate mode matching data is generated. If the liquid medicine flow rate pattern matching data is "mismatched", it is necessary to further analyze the specific situation of the deviation. For example, calculate the difference between the current flow rate and the normal flow rate pattern, and determine the direction and magnitude of the deviation. For example, if the current average flow rate is 15 ml / h lower than the normal pattern, it is marked as "low flow rate". If the standard deviation or coefficient of variation exceeds the threshold, the flow rate fluctuation is considered abnormal, such as marked as "excessive fluctuation". It is also possible to detect whether there are mutation points or abnormal trend changes in the flow rate curve. For example, if the flow rate is detected to suddenly drop by more than 20% in a short period of time, it is marked as "sudden drop in flow rate". Based on the results of deviation analysis and anomaly detection, the liquid medicine flow rate abnormality marking data is generated. For example, "low flow rate, excessive fluctuation", "sudden drop in flow rate". Using the liquid medicine flow rate abnormality marking data, the preliminary liquid medicine flow rate characteristic map is marked with abnormal areas. For example, the area where the abnormal flow rate data point is located is filled with different colors, or marked with eye-catching symbols.In this way, it is possible to clearly show which time periods have abnormal flow rates. The final real-time liquid flow rate characteristic graph data contains information about flow rate changes over time and marks of abnormal flow rates.

[0037] Preferably, step S2 comprises the following steps: Step S21: attaching a flexible transparent visual module to the outside of the infusion bag, and continuously photographing the liquid surface of the drug at a periodic frame rate to obtain original infusion bag image sequence data; wherein the flexible transparent visual module has a built-in micro camera; Step S22: performing image distortion correction on the original infusion bag image sequence data to generate corrected infusion bag image sequence data; Step S23: identifying the liquid surface of the drug solution according to the corrected infusion bag image sequence data to obtain liquid surface segmentation mask data; Step S24: calculating the remaining volume of the liquid medicine according to the liquid surface segmentation mask data to obtain the remaining volume data of the liquid medicine; Step S25: performing a drug liquid flow rate state analysis according to the real-time drug liquid flow rate characteristic spectrum data to obtain adaptive drug liquid flow rate state data; Step S26: Calculating the remaining infusion time of the residual volume data of the drug solution by using the adaptive drug solution flow rate state data to obtain the remaining infusion time data; Step S27: Perform a comprehensive evaluation of the infusion status according to the remaining infusion time data and the adaptive drug liquid flow rate status data to generate infusion status evaluation data.

[0038] In an embodiment of the present invention, a flat area is selected on the outside of the infusion bag, and the flexible transparent visual module is closely attached. The module is made of transparent material, such as medical-grade polyurethane film, to ensure that it does not block the nurse's observation of the infusion bag. A micro camera is embedded in the module, for example, the camera has a resolution of 1280x720 pixels and a viewing angle of 90 degrees, and can clearly capture the liquid level in the infusion bag. The camera performs periodic continuous shooting at a fixed frame rate, for example, it is set to shoot an image every 5 seconds. When the infusion starts, the camera starts working and transmits the captured image data to the image processing unit in real time. The camera is calibrated to obtain the internal parameters and distortion coefficients of the camera. For example, a checkerboard calibration plate can be used for calibration, and the focal length, principal point coordinates, radial distortion coefficient and tangential distortion coefficient of the camera are calculated by shooting multiple checkerboard images at different angles using Zhang Zhengyou's calibration method. According to the parameters obtained by calibration, each image in the original infusion bag image sequence data is subjected to distortion correction. For example, you can use the undistort function in the OpenCV library, input the original image and calibration parameters, and you can get the corrected image. The corrected image eliminates the distortion phenomenon, and the straight lines in the image remain straight lines in the actual space. The purpose of liquid surface recognition is to separate the liquid area from the background area in the image. You can use a deep learning-based image segmentation method, such as using a U-Net network model. Prepare a labeled data set that contains a large number of infusion bag images, and each image is labeled with a mask of the liquid area. Then, use this data set to train the U-Net network model. After the training is completed, the corrected infusion bag image sequence data is input into the trained model, and the model will output the liquid surface segmentation mask data corresponding to each image. The mask data is a binary image, in which the white area represents the liquid area and the black area represents the background area. For example, for an infusion bag image, in the mask data output by the model, the liquid part inside the infusion bag is white, and the air part outside the infusion bag and on the top of the infusion bag is black. The liquid surface segmentation mask data accurately marks the boundary of the liquid area. If the shape of the infusion bag is approximately cylindrical, the height of the liquid surface can be calculated based on the highest and lowest points of the liquid surface contour, and the volume of the liquid medicine can be calculated based on the radius of the infusion bag. If the shape of the infusion bag is irregular, a more complex geometric model can be used for calculation, such as dividing the infusion bag into multiple small cylinders, calculating the volume of each small cylinder separately, and then adding up the total volume of the liquid medicine. Based on the calculated volume of the liquid medicine, generate the residual volume data of the liquid medicine. Based on the flow rate data of the past 10 minutes, calculate the rate of change of the flow rate, and judge whether the flow rate is in an "acceleration", "deceleration" or "stable" state based on the size and direction of the rate of change. Combine the current flow rate state with the predicted flow rate change trend to generate adaptive liquid medicine flow rate state data.The remaining infusion time is calculated using the adaptive liquid flow rate status data and the liquid remaining volume data. For example, if the flow rate status is normal, the remaining infusion time can be estimated by dividing the liquid remaining volume by the average flow rate. If the flow rate status is abnormal, it is necessary to make adjustments according to the specific abnormal situation. For example, if the flow rate is slow, the estimated value of the remaining infusion time needs to be appropriately extended. The remaining infusion time data and the adaptive liquid flow rate status data are comprehensively considered to comprehensively evaluate the infusion status. For example, if the remaining infusion time is short and the flow rate is normal, the evaluation result is "the infusion is about to end"; if the remaining infusion time is long but the flow rate is abnormally slow, the evaluation result is "the infusion speed is too slow and there is a blockage"; if the remaining infusion time decreases rapidly and the flow rate is too fast, the evaluation result is "the infusion speed is too fast and there is leakage". The evaluation results are used to generate infusion status evaluation data.

[0039] Preferably, step S24 comprises the following steps: Step S241: extracting liquid surface edge pixel points according to the liquid surface segmentation mask data to obtain liquid surface edge pixel point data; Step S242: performing sub-pixel edge positioning on the liquid surface segmentation mask data by using the liquid surface edge pixel point data to obtain floating point coordinate data of the liquid surface edge point; Step S243: Obtain the geometric model of the inner wall of the infusion bag, Step S244: constructing an image pose space coordinate system according to the geometric model of the inner wall of the infusion bag to generate infusion bag space coordinate system data; Step S245: setting camera shooting parameters according to the flexible transparent vision module, and performing coordinate transformation on the floating point coordinate data of the liquid surface edge point through the spatial coordinate system data of the infusion bag, thereby obtaining the three-dimensional coordinate data of the liquid surface edge point; Step S246: Use the three-dimensional coordinate data of the liquid surface edge point to perform liquid surface curve fitting on the geometric model of the inner wall of the infusion bag, and calculate the remaining volume of the drug liquid to obtain the remaining volume data of the drug liquid.

[0040] In an embodiment of the present invention, the liquid surface segmentation mask data is a binary image, in which the white area (pixel value is 255) represents the liquid area, and the black area (pixel value is 0) represents the background area. The goal of extracting the edge pixels of the liquid surface is to find the boundary pixels between these white areas and black areas. Gaussian filtering is performed on the liquid surface segmentation mask data to remove noise. Then, the gradient amplitude and direction of the image are calculated using the Sobel operator. Next, non-maximum suppression is performed to refine the edge to a single pixel width. Finally, the double threshold method is used to detect and connect the edges to obtain the final edge pixel points. A gradient-based method is used for sub-pixel edge positioning. For example, for each edge pixel point, its gradient direction is calculated, and the position with the maximum gradient amplitude is found along the gradient direction, which is the sub-pixel edge position. The geometric model of the inner wall of the infusion bag is obtained, and the geometric model of the inner wall of the infusion bag describes the shape and size of the internal space of the infusion bag. The center of the bottom of the infusion bag is selected as the origin of the coordinate system, the central axis of the infusion bag is used as the Z axis, and the X axis and Y axis are determined according to the right-hand rule. According to the geometric model of the inner wall of the infusion bag, the three-dimensional coordinates of any point on the inner wall of the infusion bag in the coordinate system can be calculated. For example, if the shape of the infusion bag is a cylinder and its radius and height are known, the three-dimensional coordinates of any point on the inner wall can be calculated by a simple geometric relationship. If the shape of the infusion bag is irregular, the three-dimensional coordinates of any point on the inner wall can be calculated according to its surface equation. According to the internal parameters of the camera, the floating point coordinate data of the edge point of the liquid surface is converted from the image coordinate system to the camera coordinate system. Then, according to the external parameters of the camera, the points in the camera coordinate system are converted to the world coordinate system, that is, the spatial coordinate system of the infusion bag defined in step S244. Using the thin plate spline interpolation method, a smooth surface is fitted according to the three-dimensional coordinate data of the edge point of the liquid surface, which accurately passes through all edge points and has the smallest curvature. Then, according to the fitted liquid surface and the geometric model of the inner wall of the infusion bag, the volume of the liquid is calculated. For example, the volume of the liquid can be regarded as the space between the inner wall of the infusion bag and the liquid surface, and the volume of the space is calculated using the integral method.

[0041] Preferably, step S246 includes the following steps: The liquid surface edge point 3D coordinate data is used to fit the liquid surface curve of the inner wall geometric model of the infusion bag, and the liquid surface of the infusion bag is 3D reconstructed to obtain real-time liquid 3D model data; Perform model point cloud processing according to the real-time liquid medicine three-dimensional model data to obtain liquid medicine point cloud data; Perform bounding box processing on the liquid medicine point cloud data to obtain liquid medicine bounding box data; Perform spatial grid division according to the liquid medicine bounding box data to obtain model spatial grid data; Mark the liquid occupancy according to the model space grid data and the liquid point cloud data, and perform grid connected domain analysis to obtain the liquid connected domain data; The maximum connected domain of the model space grid data is extracted through the drug liquid connected domain data, and the residual volume of the drug liquid is accurately calculated to obtain the residual volume data of the drug liquid.

[0042] In an embodiment of the present invention, a surface fitting algorithm is used to generate a smooth surface based on the three-dimensional coordinate data of the edge points of the liquid surface, and the surface approximates all edge points. For example, a radial basis function (RBF) interpolation method can be used to construct a surface function passing through all edge points. The function takes the spatial coordinates (x, y, z) as input and outputs a scalar value representing the distance from the point to the liquid surface. The geometric model of the inner wall of the infusion bag is represented as a closed surface, and the fitted liquid surface is used as an opening of the closed surface. In this way, the two surfaces together constitute a three-dimensional model, which describes the shape and position of the liquid in the infusion bag at the current moment. For example, if the geometric model of the inner wall of the infusion bag is a cylinder and the liquid surface is a smooth surface passing through all edge points, the result of the three-dimensional reconstruction is a three-dimensional model with a cylinder at the bottom and the smooth surface at the top. The real-time liquid three-dimensional model is sampled, and a series of three-dimensional points are generated on its surface and inside, and these points constitute point cloud data. For example, a uniform sampling method can be used to uniformly generate 10,000 points on the surface and inside of the model. The generated point cloud data is processed to remove redundant points and noise points. For example, the distance between each point and its neighboring points can be calculated. If the distance is too small, it is considered a redundant point and deleted. Some outliers can also be removed according to the density distribution of the point cloud, which are noise points. After processing, a streamlined point cloud data is obtained, which more accurately reflects the shape and position of the liquid medicine. The bounding box is a minimum cuboid that can contain all the liquid medicine point cloud data. There are many methods to calculate the bounding box, such as AABB (axis-aligned bounding box) or OBB (oriented bounding box). For example, the AABB method can be used to calculate the maximum and minimum values ​​of all points in the liquid medicine point cloud data in the three directions of x, y, and z, so as to determine a cuboid that can contain all points. The size of each grid unit is determined according to the size of the bounding box and the preset grid resolution. For example, if the size of the bounding box is 40x40x40 and the preset grid resolution is 1 mm, the size of each grid cell is 1x1x1 mm. Then, according to the size of the grid cell, the bounding box space is divided into a series of grid cells. For example, the bounding box space can be divided into 40x40x40=64000 grid cells. Each grid cell can be represented by the coordinates of its center, for example, the center coordinates of the first grid cell are (10.5, 20.5, 5.5), the center coordinates of the second grid cell are (11.5, 20.5, 5.5), and so on. Traverse all grid cells to determine whether each grid cell is occupied by the liquid medicine. For example, the distance between the center point of each grid cell and all the points in the liquid medicine point cloud data can be calculated. If there is a point whose distance is less than a preset threshold, such as 0.5 mm, the grid cell is considered to be occupied by the liquid medicine and is marked as 1; otherwise, it is marked as 0. Use the eight-connected or twenty-six-connected algorithm for connected domain analysis.Each connected domain represents an independent liquid medicine area. For example, after analysis, two connected domains are obtained, the first connected domain contains 1000 grid cells, and the second connected domain contains 500 grid cells. The index information of the grid cells contained in each connected domain is saved to generate liquid medicine connected domain data. From the liquid medicine connected domain data, find the connected domain containing the largest number of grid cells. This connected domain is the largest connected domain, representing the main liquid medicine area. For example, in the example of step S246.5, the first connected domain contains 1000 grid cells and is the largest connected domain. According to the number of grid cells contained in the largest connected domain and the volume of each grid cell, the total volume of the liquid medicine is calculated. For example, if the largest connected domain contains 1000 grid cells and the volume of each grid cell is 1 cubic millimeter, the total volume of the liquid medicine is 1000 cubic millimeters, that is, 1 cubic centimeter, that is, 1 milliliter. This volume value is used as the final liquid medicine residual volume data.

[0043] As an example of the present invention, refer to Figure 2 As shown, Figure 1 Detailed implementation steps of step S3 in the flowchart, in this example, step S3 includes: Step S31: judging the infusion risk level according to the infusion status assessment data; when the remaining infusion time is less than the first nursing time threshold and the drug flow rate fluctuates abnormally, marking the infusion status assessment data as a high-risk infusion status, and automatically implementing a high-priority alarm push; In an embodiment of the present invention, the system extracts the remaining infusion time data and the drug flow rate status data from the infusion status assessment data. For example, in the current infusion status assessment data, the remaining infusion time is 15 minutes, and the drug flow rate status is "abnormal fluctuation, low flow rate". Then, the remaining infusion time is compared with the preset first nursing time threshold. For example, the first nursing time threshold is set to 20 minutes. When the remaining infusion time is less than the first nursing time threshold, and the drug flow rate status data contains the keyword "abnormal fluctuation", the infusion status assessment data is marked as a high-risk infusion state. For example, in this case, the remaining infusion time of 15 minutes is less than the first nursing time threshold of 20 minutes, and the drug flow rate status is "abnormal fluctuation, low flow rate", which meets the judgment conditions of the high-risk infusion state, so the infusion status assessment data is marked as "high risk". For the infusion state marked as "high risk", the system will automatically implement a high-priority alarm push. For example, the system can display the patient's bed number, name, and "high risk" mark in a red pop-up window on the monitoring screen of the nurse station, and issue a high-decibel alarm sound, such as a series of rapid beeps. In addition, the system can also send vibration and text prompts through smart devices (such as smart watches) worn by nurses to ensure that nurses can receive alarm information in the first place.

[0044] Step S32: when the remaining infusion time is less than the second nursing time threshold and the drug flow rate fluctuates abnormally, the infusion status assessment data is marked as a medium-risk infusion status, and the electronic medical record system is used to query the patient's infusion order information to obtain the patient's infusion order data; otherwise, the infusion status assessment data is marked as a low-risk infusion status, and no related operations are performed; wherein the second nursing time threshold is greater than the first nursing time threshold; In an embodiment of the present invention, the system extracts the remaining infusion time data and the liquid medicine flow rate state data from the infusion state assessment data. For example, in the current infusion state assessment data, the remaining infusion time is 25 minutes, and the liquid medicine flow rate state is "abnormal fluctuation, high flow rate". Then, the remaining infusion time is compared with the preset second nursing time threshold. For example, the second nursing time threshold is set to 30 minutes. When the remaining infusion time is less than the second nursing time threshold, and the liquid medicine flow rate state data contains the keyword "abnormal fluctuation", the infusion state assessment data is marked as a medium-risk infusion state. For example, in this case, the remaining infusion time of 25 minutes is less than the second nursing time threshold of 30 minutes, and the liquid medicine flow rate state is "abnormal fluctuation, high flow rate", which meets the judgment condition of the medium-risk infusion state, so the infusion state assessment data is marked as "medium risk". For the infusion state marked as "medium risk", the system will automatically trigger the doctor's order query process. The system establishes an interface connection with the hospital's electronic medical record system, and uses the patient's unique identifier (such as hospitalization number or ID number) as an index to query the patient's current infusion doctor's order information. If the infusion status assessment data does not meet the judgment conditions of the medium-risk infusion status, such as the remaining infusion time is greater than or equal to the second nursing time threshold, or the drug flow rate status data does not contain the "abnormal fluctuation" keyword, the infusion status assessment data will be marked as a low-risk infusion status. For example, in the current infusion status assessment data, the remaining infusion time is 45 minutes and the drug flow rate status is "normal", it will be marked as "low risk". For low-risk infusion status, the system will not perform any operation and continue to perform regular infusion status monitoring.

[0045] Step S33: formatting the patient's infusion order data to generate structured patient order data; In an embodiment of the present invention, the system receives the patient infusion order data obtained in step S32, formats it, and generates patient order structured data. Patient infusion order data is usually stored in the electronic medical record system in text or semi-structured form, such as "500ml of normal saline intravenous drip 20 drops / minute". The system needs to convert these unstructured or semi-structured data into a structured data format for subsequent processing and analysis. For example, the system can extract the infusion type as "normal saline", the dosage as "500ml", and the speed as "20 drops / minute", and store this information in different fields of the database. The speed can also be converted into a more standard unit, such as ml / hour.

[0046] Step S34: performing abnormal infusion key state parameter analysis on the real-time drug liquid flow rate characteristic spectrum data through the medium-risk infusion state to obtain abnormal infusion key state parameters; In an embodiment of the present invention, the system extracts all data points in the flow rate curve that exceed the normal flow rate range, and calculates the average overspeed value and the maximum overspeed value. For example, according to the doctor's advice, the normal flow rate range of the patient is 50-70 ml / hour, and in the real-time flow rate map data, there are multiple data points exceeding 70 ml / hour, the average overspeed value is 8 ml / hour, and the maximum overspeed value is 15 ml / hour. Then, the system analyzes the fluctuation of the flow rate curve and calculates its fluctuation amplitude and fluctuation frequency. For example, the standard deviation or coefficient of variation of the flow rate can be calculated to measure the fluctuation amplitude, and the number of times the flow rate curve crosses its average value per unit time can be calculated to measure the fluctuation frequency. For example, the pressure curve of the pressure sensor can be analyzed to determine whether there is a sudden increase or decrease in pressure, which is related to blockage or leakage of the infusion line. For example, it is found that during the time period when the flow rate fluctuates abnormally, the data of the pressure sensor also fluctuates abnormally, and the fluctuation amplitude exceeds 20% of the normal range. According to the above analysis results, the key state parameters of abnormal infusion are extracted, for example: {average overspeed value: 8 ml / hour, maximum overspeed value: 15 ml / hour, flow rate standard deviation: 5 ml / hour, fluctuation frequency: 10 times per minute, pressure fluctuation amplitude: 20%}, and the key state parameters of abnormal infusion are generated.

[0047] Step S35: performing transfer learning on the convolutional neural network model through the preset infusion doctor's order-infusion action mapping rule, and performing spatial probability distribution processing of the control suggestion action on the infusion bag to train the control suggestion action decision model; In an embodiment of the present invention, the system pre-sets an infusion order-infusion action mapping rule library. The rule library defines the corresponding actions that should be taken under different infusion orders and different infusion states. For example, if the infusion rate required by the doctor's order is 20 drops / minute, and the actual flow rate is 10 drops / minute, the corresponding action is "increase the infusion rate". For another example, if bubbles are detected, the corresponding action is "pause the infusion and remove the bubbles". Using this preset rule library, a pre-trained convolutional neural network model can be transferred to learn, so that it can predict the best action to be taken based on the patient's infusion order and the current infusion state. Transfer learning can use pre-trained ResNet or Inception models, replace the output layer of the pre-trained model with a new output layer, and use the data in the infusion order-infusion action mapping rule library to fine-tune the model. After the training is completed, the model is applied to the probability distribution processing of the infusion bag control suggestion action space. The infusion bag control suggestion action space contains all control actions, such as "increase the infusion rate", "reduce the infusion rate", "pause the infusion", "change the infusion set", etc. The model calculates the probability of each action and generates a probability distribution. The action with the highest probability is considered to be the most appropriate regulation suggestion for the current situation.

[0048] Step S36: Use the control suggestion action decision model to perform intelligent control action decision processing on the patient's medical order structured data and abnormal infusion key state parameters, and perform action safety verification to generate intelligent infusion control suggestion data.

[0049] In the embodiment of the present invention, the patient's medical order structured data generated in step S33 and the abnormal infusion key state parameters obtained in step S34 are used as inputs and provided to the control suggestion action decision model trained in step S35. The model will calculate the probability distribution of each control action based on these input data, and select the action with the highest probability as the intelligent infusion control suggestion. For example, if the infusion rate required by the patient's medical order is 20 drops / minute, but the actual flow rate is only 10 drops / minute, and the flow rate is detected to be continuously decreasing, the model will suggest "increase the infusion rate". After generating the intelligent infusion control suggestion, the system will perform a safety check on the suggested action. For example, check whether the suggested infusion rate is within the safety range and whether it will cause potential risks to the patient. For example, if the suggested infusion rate is too high and exceeds the maximum speed that the patient can bear, the system will reject the suggestion and generate a new suggestion, such as "Please check whether the infusion line is blocked". Only the suggestions that pass the safety check will eventually generate the intelligent infusion control suggestion data and pass it to the next step of manual adjustment action instruction coding.

[0050] As an example of the present invention, refer to Figure 3 As shown, Figure 1Detailed implementation steps of step S4 in the embodiment are shown in the flowchart. In this embodiment, step S4 includes: Step S41: encoding manual adjustment action instructions according to the intelligent infusion control suggestion data to obtain manual adjustment action instruction data; In an embodiment of the present invention, a manual adjustment action instruction coding table is established, which contains all manual adjustment actions and their corresponding codes. For example, the code for "reducing the dripping rate" can be defined as "001", the code for "increasing the dripping rate" can be defined as "002", the code for "changing the infusion bag" can be defined as "003", the code for "exhausting" can be defined as "004", and the code for "remaining unchanged" can be defined as "005", etc. The format of the coding can be designed according to actual needs, for example, binary coding, hexadecimal coding or customized coding rules can be used. According to the intelligent infusion control recommendation data, the corresponding coding is searched in the manual adjustment action instruction coding table. For example, if the intelligent infusion control recommendation data is "reducing the dripping rate", the code corresponding to "reducing the dripping rate" is searched in the coding table to obtain the code "001". The code is used as the manual adjustment action instruction data and saved together with the corresponding patient information and timestamp information.

[0051] Step S42: The recipient of the nursing staff is determined by the intelligent infusion control suggestion data, and the manual adjustment action instruction data is visually prompted to obtain visual nursing suggestion data; In an embodiment of the present invention, the nursing staff who receives the control suggestion is determined according to the personnel scheduling information of the current ward and the scope of patients that each nursing staff is responsible for. For example, the work number and name of the nursing staff responsible for the bed can be queried according to the bed number where the patient is located. The intelligent infusion control suggestion data and the manual adjustment action instruction data are sent to the designated nursing staff. For example, it can be sent through the computer system of the nurse station, the personal mobile device of the nursing staff (such as PDA or smart phone), etc. The content sent may include the patient's basic information (such as bed number, name, hospitalization number), intelligent infusion control suggestions (such as "reduce the drip rate"), manual adjustment action instruction code (such as "001"), and the urgency of the suggestion, etc.

[0052] Step S43: Based on the visualized nursing suggestion data, the nursing staff manually adjusts the dripping rate of the liquid medicine using the infusion regulator to obtain the manually adjusted liquid medicine, and at the same time collects the nursing staff's confirmation / adjustment operation instructions to obtain the manual operation instruction data; In an embodiment of the present invention, the nursing staff finds the corresponding patient and infusion device according to the prompt information in the visualized nursing suggestion data, and makes manual adjustments according to the suggested content. For example, if the suggestion is "reduce the drip rate", the nursing staff will find the infusion regulator and rotate the adjustment knob to reduce the drip rate to the suggested range. While making manual adjustments, the system will collect the nursing staff's operation instructions. For example, the rotation angle and direction of the adjustment knob can be detected by a sensor installed on the infusion regulator, so as to determine whether the nursing staff has performed the operation of "increasing the drip rate" or "reducing the drip rate". If the infusion device is equipped with a confirmation button, the system can also collect the operation record of the nursing staff pressing the confirmation button. For example, when the nursing staff completes the adjustment and presses the confirmation button, the system will record the timestamp and operation type ("confirmation") of the operation. For example, the nurse can confirm that the operation has been completed through the button or touch screen on the infusion regulator, or can manually enter the adjusted infusion speed value.

[0053] Step S44: binding the patient's identity according to the manual operation instruction data, and performing manual adjustment execution record to obtain manual adjustment execution record data; In an embodiment of the present invention, the system receives the manual operation instruction data collected in step S43, such as operation records of the nurse confirming the execution of the suggestion, adjusting the infusion speed, or refusing to execute the suggestion. In order to ensure the accuracy and traceability of the operation records, the system needs to bind the manual operation instruction data to the patient's identity. For example, the system can associate the operation record with a specific patient by scanning a barcode or QR code on the patient's wrist, or by entering information such as the patient's hospitalization number. In addition, the system also needs to record information such as the time of operation execution and the operator's ID. All of this information will be recorded in the manual adjustment execution record data.

[0054] Step S45: monitoring the state of the adjusted drug solution based on the nursing infusion monitoring network to obtain the adjusted drug solution fluid dynamics parameters; performing manual adjustment deviation calculation based on the adjusted drug solution fluid dynamics parameters to obtain manual adjustment deviation data; In an embodiment of the present invention, after the nurse completes the manual adjustment operation, the system will monitor the state of the adjusted liquid medicine based on the nursing infusion monitoring network (the monitoring network established in step S1). For example, the system will continue to collect parameters such as pressure, temperature, and flow rate of the liquid medicine to obtain the adjusted liquid medicine fluid dynamics parameters. Then, the system will compare the adjusted liquid medicine fluid dynamics parameters with the abnormal infusion key state parameters analyzed in step S34 to calculate the manual adjustment deviation. For example, if the previous abnormal parameter is "flow rate is too slow, 10 drops / minute", and the flow rate after manual adjustment is 25 drops / minute, the manual adjustment deviation is 15 drops / minute. The system will record the deviation values ​​of all relevant parameters and generate manual adjustment deviation data. These data can reflect the effect of manual adjustment and are used to evaluate whether the adjustment has achieved the expected goal.

[0055] Step S46: Evaluate the manual adjustment result of the manual adjustment deviation data based on the preset deviation threshold, and bind the manual adjustment execution record data to the nursing record to obtain the infusion nursing status record data.

[0056] In an embodiment of the present invention, the manual adjustment deviation data generated in step S45 is evaluated for the manual adjustment result based on a preset deviation threshold, and the nursing record is bound to the manual adjustment execution record data generated in step S44 to obtain the infusion nursing status record data. First, a deviation threshold needs to be set in advance to determine whether the result of the manual adjustment is qualified. For example, the deviation threshold can be set to ±3 drops / minute, that is, if the deviation between the actual flow rate after manual adjustment and the recommended value is within ±3 drops / minute, the adjustment result is considered qualified; otherwise, the adjustment result is considered unqualified. The manual adjustment deviation data is compared with the preset deviation threshold to determine whether the adjustment result is qualified. For example, if the manual adjustment deviation data is 2 drops / minute, which is less than the preset deviation threshold of 3 drops / minute, the adjustment result is considered qualified. If the manual adjustment deviation data is 5 drops / minute, which is greater than the preset deviation threshold, the adjustment result is considered unqualified. In the case of unqualified adjustment results, the system can automatically generate a prompt message, such as "the adjustment deviation is too large, please readjust", and send the prompt message to the nursing staff. The result of the manual adjustment result evaluation ("qualified" or "unqualified"), the manual adjustment deviation data and the related prompt information are bound to the manual adjustment execution record data generated in step S44 to generate a complete infusion care status record.

[0057] Preferably, step S45 includes the following steps: Step S451: estimating the theoretical flow rate change of the manual adjustment action instruction data to obtain theoretical flow rate change data; Step S452: monitoring the state of the adjusted drug solution based on the nursing infusion monitoring network to obtain the fluid dynamics parameters of the adjusted drug solution; Step S453: analyzing the actual change of the flow rate of the drug solution according to the adjusted fluid dynamics parameters of the drug solution to generate actual change data of the flow rate of the drug solution; Step S454: Calculate the manual flow adjustment deviation of the actual liquid medicine flow change data using the theoretical flow change data to obtain manual adjustment deviation data.

[0058] In the embodiment of the present invention, a mapping relationship table between an operation instruction and a theoretical flow change is established. The table can be formulated according to the design parameters of the infusion regulator and clinical experience. For example, for a common spiral infusion regulator, the following mapping relationship can be established according to the relationship between its pitch and rotation angle and the drip rate change: "reducing the drip rate by one gear" corresponds to "theoretical flow reduction by 5 ml / hour", "increasing the drip rate by one gear" corresponds to "theoretical flow increase by 5 ml / hour", "reducing the drip rate by two gears" corresponds to "theoretical flow reduction by 10 ml / hour", and so on. According to the manual operation instruction data, the above mapping relationship table is queried to obtain the corresponding theoretical flow change value. For example, if the manual operation instruction data is "reducing the drip rate", and according to the coding rule in step S43, its code is "3", then the mapping relationship table is queried to obtain the corresponding theoretical flow change of "-5 ml / hour" (assuming that reducing one gear corresponds to 5 ml / hour). The theoretical flow change value is associated with the liquid flow rate before the operation and the patient information, etc., to generate theoretical flow change data. The sensor in the microfluidic chip converts the collected analog signal into a digital signal and performs preliminary processing, such as filtering, amplification, etc. Then, these digital signals are packaged into data frames and transmitted to the central server in real time through the nursing infusion monitoring network (such as a Wi-Fi network). After receiving these data frames, the central server will classify and store them according to the sensor type and number, and perform unit conversion and other processing. For example, the raw data of the flow rate sensor is converted into the standard unit "ml / hour". The processed data is associated with the timestamp, patient information, etc. to generate the adjusted fluid dynamics parameters of the liquid medicine. The liquid medicine flow rate data is extracted from the adjusted liquid medicine flow dynamics parameters and compared with the liquid medicine flow rate data before adjustment. For example, the system can compare the flow values ​​before and after adjustment and calculate the change in flow. In order to more accurately reflect the flow change, the system can calculate the average flow change over a period of time. The system compares the theoretical flow change data calculated in step S451 with the actual liquid medicine flow change data obtained in step S453, and calculates the difference between the actual flow change value and the theoretical flow change value, that is, the manual adjustment deviation.

[0059] The present invention also provides a nursing information visualization management system, which executes the above-mentioned nursing information visualization management method, and the nursing information visualization management system comprises: The drug liquid flow rate monitoring module is used to integrate the microfluidic chip at the outlet of the infusion bag to obtain a nursing infusion monitoring network; the nursing infusion monitoring network is used to collect multi-parameter sensor signals of the drug liquid to generate real-time drug liquid fluid dynamics parameters; the drug liquid flow rate spectrum is characterized and processed according to the real-time drug liquid fluid dynamics parameters to generate real-time drug liquid flow rate characteristic spectrum data; The infusion state evaluation module is used to continuously shoot the liquid surface of the infusion bag at a periodic frame rate to generate corrected infusion bag image sequence data; calculate the remaining volume of the liquid medicine according to the corrected infusion bag image sequence data to obtain the remaining volume data of the liquid medicine; analyze the liquid medicine flow rate state according to the real-time liquid medicine flow rate characteristic spectrum data to obtain adaptive liquid medicine flow rate state data; use the adaptive liquid medicine flow rate state data to perform a comprehensive infusion state evaluation on the remaining volume data of the liquid medicine to generate infusion state evaluation data; The intelligent control decision module is used to perform differentiated processing of infusion risks based on infusion status assessment data, and to analyze the key state parameters of abnormal infusion on the real-time drug flow rate characteristic map data to obtain the key state parameters of abnormal infusion; to use the electronic medical record system to query the infusion doctor's order information and generate patient doctor's order structured data; to build a control suggestion action decision model for the infusion bag; to use the control suggestion action decision model to perform intelligent control action decision processing on the key state parameters of abnormal infusion and the patient doctor's order structured data to generate intelligent infusion control suggestion data; The nursing advice visualization module is used to encode manual adjustment action instructions according to the intelligent infusion control advice data, and provide visual prompts to obtain visualized nursing advice data; it records manual adjustment execution according to the visualized nursing advice data, and evaluates the manual adjustment results, thereby obtaining infusion nursing status record data.

[0060] The present application is to integrate a microfluidic chip at the outlet of the infusion bag, so as to collect multi-parameter sensor signals of the liquid medicine in real time, including flow rate, pressure, temperature, etc. Compared with the traditional manual timed inspection and observation method, the present invention avoids the problems of time lag and inaccurate data, and can continuously and accurately monitor the infusion speed of the patient, thereby effectively preventing medical accidents caused by too fast or too slow infusion, and ensuring the safety of the patient's medication. By continuously shooting the liquid surface of the infusion bag at a periodic frame rate, combined with real-time flow rate data, the remaining volume of the liquid medicine can be accurately calculated, and a comprehensive evaluation of the infusion status can be performed. The system can differentiate the infusion risk according to the evaluation data, timely identify the high-risk infusion status, and automatically implement high-priority alarm push. Through the intelligent infusion management system, the nursing efficiency can be significantly improved and the workload of nursing staff can be reduced. The system can automatically monitor and evaluate the infusion status and generate control suggestions. Nursing staff only need to operate according to the system prompts, avoiding frequent manual inspections and recording work. This not only improves work efficiency, but also reduces the risk of human operation errors.

[0061] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0062] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for visual management of nursing information, characterized in that: The following steps are involved: Step S1: integrating the microfluidic chip at the outlet of the infusion bag to obtain a nursing infusion monitoring network; The nursing infusion monitoring network is used to collect multi-parameter sensor signals of the drug solution to generate real-time drug solution fluid dynamics parameters; the drug solution flow rate spectrum is characterized and processed according to the real-time drug solution fluid dynamics parameters to generate real-time drug solution flow rate characteristic spectrum data; Step S2: continuously photographing the liquid surface of the infusion bag at a periodic frame rate to generate corrected infusion bag image sequence data; calculating the remaining volume of the liquid medicine according to the corrected infusion bag image sequence data to obtain remaining volume data of the liquid medicine; analyzing the flow rate state of the liquid medicine according to the real-time liquid medicine flow rate characteristic spectrum data to obtain adaptive liquid medicine flow rate state data; using the adaptive liquid medicine flow rate state data to perform a comprehensive infusion state evaluation on the remaining volume data of the liquid medicine to generate infusion state evaluation data; Step S3: Perform differentiated infusion risk processing based on the infusion status assessment data, and perform abnormal infusion key state parameter analysis on the real-time drug liquid flow rate characteristic spectrum data to obtain abnormal infusion key state parameters; use the electronic medical record system to query infusion doctor order information and generate patient doctor order structured data; Construct a control suggestion action decision model for the infusion bag; use the control suggestion action decision model to perform intelligent control action decision processing on abnormal infusion key state parameters and patient medical order structured data to generate intelligent infusion control suggestion data; Step S4: encoding manual adjustment action instructions according to the intelligent infusion control suggestion data, and performing visual prompts to obtain visual nursing suggestion data; Manual adjustment execution records are made according to the visualized nursing suggestion data, and the manual adjustment result evaluation is performed to obtain the infusion nursing status record data.

2. The method for visual management of nursing information according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: initializing microfluidic parameters of the infusion bag outlet and integrating the microfluidic chip to obtain a nursing infusion monitoring network; wherein the microfluidic chip includes a piezoresistive sensor, a capacitive sensor and an optical sensor; Step S12: when the infusion starts, the original drug solution multi-parameter sensor signal is collected by using the nursing infusion monitoring network to obtain the original drug solution sensor signal data; Step S13: preprocessing the liquid medicine fluid signal according to the original liquid medicine sensor signal data to generate real-time liquid medicine fluid dynamics parameters; Step S14: Obtaining liquid medicine type coding data; Step S15: calculating the flow rate of the medicine liquid according to the real-time medicine liquid fluid dynamics parameters by using the medicine liquid type coding data to generate real-time medicine liquid flow rate data; Step S16: Performing drug liquid flow rate graph characterization processing according to the real-time drug liquid flow rate data to generate real-time drug liquid flow rate characteristic graph data.

3. The method for visual management of nursing information according to claim 2, characterized in that: Step S16 includes the following steps: Step S161: performing time axis mapping according to the real-time liquid medicine flow rate data to obtain signal acquisition time axis mapping data; Step S162: performing flow rate value axis mapping according to the real-time liquid medicine flow rate data to obtain flow rate value mapping coordinate data; Step S163: Draw a drug flow rate scatter plot according to the signal acquisition time axis mapping data and the flow rate value mapping coordinate data to obtain a preliminary drug solution flow rate characteristic map; Step S164: performing flow rate pattern matching on the real-time fluid dynamics parameters of the drug liquid through a preset normal flow rate pattern library based on the drug liquid type coding data to obtain drug liquid flow rate pattern matching data; Step S165: performing a drug liquid flow rate deviation analysis on the preliminary drug liquid flow rate characteristic map through drug liquid flow rate pattern matching data, and performing a drug liquid flow rate fluctuation abnormality detection to generate drug liquid flow rate abnormality marking data; Step S166: using the abnormal marking data of the drug liquid flow rate to mark abnormal areas of the preliminary drug liquid flow rate characteristic map, thereby obtaining real-time drug liquid flow rate characteristic map data.

4. The method for visual management of nursing information according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: attaching a flexible transparent visual module to the outside of the infusion bag, and continuously photographing the liquid surface of the drug at a periodic frame rate to obtain original infusion bag image sequence data; wherein the flexible transparent visual module has a built-in micro camera; Step S22: performing image distortion correction on the original infusion bag image sequence data to generate corrected infusion bag image sequence data; Step S23: identifying the liquid surface of the drug solution according to the corrected infusion bag image sequence data to obtain liquid surface segmentation mask data; Step S24: calculating the remaining volume of the liquid medicine according to the liquid surface segmentation mask data to obtain the remaining volume data of the liquid medicine; Step S25: analyzing the flow rate state of the drug liquid according to the real-time drug liquid flow rate characteristic spectrum data to obtain adaptive drug liquid flow rate state data; Step S26: Calculating the remaining infusion time of the residual volume data of the drug solution by using the adaptive drug solution flow rate state data to obtain the remaining infusion time data; Step S27: Perform a comprehensive evaluation of the infusion status according to the remaining infusion time data and the adaptive drug liquid flow rate status data to generate infusion status evaluation data.

5. The method for visual management of nursing information according to claim 4, characterized in that: Step S24 includes the following steps: Step S241: extracting liquid surface edge pixel points according to the liquid surface segmentation mask data to obtain liquid surface edge pixel point data; Step S242: performing sub-pixel edge positioning on the liquid surface segmentation mask data using the liquid surface edge pixel point data to obtain floating-point coordinate data of the liquid surface edge point; Step S243: Obtain the geometric model of the inner wall of the infusion bag, Step S244: constructing an image pose space coordinate system according to the geometric model of the inner wall of the infusion bag to generate infusion bag space coordinate system data; Step S245: setting camera shooting parameters according to the flexible transparent vision module, and performing coordinate transformation on the floating point coordinate data of the liquid surface edge point through the spatial coordinate system data of the infusion bag, thereby obtaining the three-dimensional coordinate data of the liquid surface edge point; Step S246: Use the three-dimensional coordinate data of the liquid surface edge point to perform liquid surface curve fitting on the geometric model of the inner wall of the infusion bag, and calculate the remaining volume of the drug liquid to obtain the remaining volume data of the drug liquid.

6. The method for visual management of nursing information according to claim 5, characterized in that: Step S246 includes the following steps: The liquid surface is fitted to the geometric model of the inner wall of the infusion bag using the three-dimensional coordinate data of the liquid surface edge points, and the liquid surface of the infusion bag is reconstructed in three dimensions to obtain real-time three-dimensional model data of the liquid medicine; Perform model point cloud processing according to the real-time liquid medicine three-dimensional model data to obtain liquid medicine point cloud data; Perform bounding box processing on the liquid medicine point cloud data to obtain liquid medicine bounding box data; Perform spatial grid division according to the liquid medicine bounding box data to obtain model spatial grid data; Mark the liquid occupancy according to the model space grid data and the liquid point cloud data, and perform grid connected domain analysis to obtain the liquid connected domain data; The maximum connected domain of the model space grid data is extracted through the drug liquid connected domain data, and the residual volume of the drug liquid is accurately calculated to obtain the residual volume data of the drug liquid.

7. The method for visual management of nursing information according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: judging the infusion risk level according to the infusion status assessment data; when the remaining infusion time is less than the first nursing time threshold and the drug flow rate fluctuates abnormally, marking the infusion status assessment data as a high-risk infusion status, and automatically implementing a high-priority alarm push; Step S32: when the remaining infusion time is less than the second nursing time threshold and the drug flow rate fluctuates abnormally, the infusion status assessment data is marked as a medium-risk infusion status, and the electronic medical record system is used to query the patient's infusion order information to obtain the patient's infusion order data; otherwise, the infusion status assessment data is marked as a low-risk infusion status, and no related operations are performed; wherein the second nursing time threshold is greater than the first nursing time threshold; Step S33: formatting the medical order data of the patient's infusion to generate structured medical order data of the patient; Step S34: performing abnormal infusion key state parameter analysis on the real-time drug liquid flow rate characteristic spectrum data through the medium-risk infusion state to obtain abnormal infusion key state parameters; Step S35: performing transfer learning on the convolutional neural network model through the preset infusion doctor's order-infusion action mapping rule, and performing spatial probability distribution processing of the control suggestion action on the infusion bag to train the control suggestion action decision model; Step S36: Use the control suggestion action decision model to perform intelligent control action decision processing on the patient's medical order structured data and abnormal infusion key state parameters, and perform action safety verification to generate intelligent infusion control suggestion data.

8. The method for visual management of nursing information according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: encoding manual adjustment action instructions according to the intelligent infusion control suggestion data to obtain manual adjustment action instruction data; Step S42: The recipient of the nursing staff is determined by the intelligent infusion control suggestion data, and the manual adjustment action instruction data is visually prompted to obtain visual nursing suggestion data; Step S43: Based on the visualized nursing suggestion data, the nursing staff manually adjusts the dripping rate of the liquid medicine using the infusion regulator to obtain the manually adjusted liquid medicine, and at the same time collects the nursing staff's confirmation / adjustment operation instructions to obtain the manual operation instruction data; Step S44: binding the patient's identity according to the manual operation instruction data, and performing manual adjustment execution records to obtain manual adjustment execution record data; Step S45: monitoring the state of the adjusted drug solution based on the nursing infusion monitoring network to obtain the adjusted drug solution fluid dynamics parameters; performing manual adjustment deviation calculation based on the adjusted drug solution fluid dynamics parameters to obtain manual adjustment deviation data; Step S46: Evaluate the manual adjustment result of the manual adjustment deviation data based on the preset deviation threshold, and bind the manual adjustment execution record data to the nursing record to obtain the infusion nursing status record data.

9. The method for visual management of nursing information according to claim 8, characterized in that: Step S45 includes the following steps: Step S451: estimating the theoretical flow rate change of the manual adjustment action instruction data to obtain theoretical flow rate change data; Step S452: monitoring the state of the adjusted drug solution based on the nursing infusion monitoring network to obtain the fluid dynamics parameters of the adjusted drug solution; Step S453: analyzing the actual change of the flow rate of the drug solution according to the adjusted fluid dynamics parameters of the drug solution to generate actual change data of the flow rate of the drug solution; Step S454: Calculate the manual flow adjustment deviation of the actual liquid medicine flow change data using the theoretical flow change data to obtain manual adjustment deviation data.

10. A visual management system for nursing information, characterized in that: Used to execute the method for visual management of nursing information according to claim 1, the visual management system for nursing information comprises: The drug liquid flow rate monitoring module is used to integrate the microfluidic chip at the outlet of the infusion bag to obtain a nursing infusion monitoring network; the nursing infusion monitoring network is used to collect multi-parameter sensor signals of the drug liquid to generate real-time drug liquid fluid dynamics parameters; the drug liquid flow rate spectrum is characterized and processed according to the real-time drug liquid fluid dynamics parameters to generate real-time drug liquid flow rate characteristic spectrum data; The infusion state evaluation module is used to continuously shoot the liquid surface of the infusion bag at a periodic frame rate to generate corrected infusion bag image sequence data; calculate the remaining volume of the liquid medicine according to the corrected infusion bag image sequence data to obtain the remaining volume data of the liquid medicine; analyze the liquid medicine flow rate state according to the real-time liquid medicine flow rate characteristic spectrum data to obtain adaptive liquid medicine flow rate state data; use the adaptive liquid medicine flow rate state data to perform a comprehensive infusion state evaluation on the remaining volume data of the liquid medicine to generate infusion state evaluation data; The intelligent control decision module is used to perform differentiated processing of infusion risks based on infusion status assessment data, and to analyze the key state parameters of abnormal infusion on the real-time drug flow rate characteristic map data to obtain the key state parameters of abnormal infusion; to use the electronic medical record system to query the infusion doctor's order information and generate patient doctor's order structured data; to build a control suggestion action decision model for the infusion bag; to use the control suggestion action decision model to perform intelligent control action decision processing on the key state parameters of abnormal infusion and the patient doctor's order structured data to generate intelligent infusion control suggestion data; The nursing advice visualization module is used to encode manual adjustment action instructions according to the intelligent infusion control advice data, and provide visual prompts to obtain visualized nursing advice data; it records manual adjustment execution according to the visualized nursing advice data, and evaluates the manual adjustment results, thereby obtaining infusion nursing status record data.

Images