Transfusion speed alarm method and system

Abstract

The invention relates to an infusion speed alarm method and system. The infusion speed alarm method comprises the steps that a multi-mode sensing data set is formed; according to the multi-mode sensing data set, whether the conditions of dripping speed abnormity, pressure sudden increase, too large bubbles, pipeline blockage or backflow exist or not is judged, and corresponding abnormal feature information is generated; inputting the abnormal feature information into a prediction model and a fuzzy logic controller of a long short-term memory network, outputting a prediction result of a future infusion speed in an infusion process, and dynamically adjusting an alarm threshold according to the prediction result; executing graded early warning based on the prediction result and the dynamic adjustment alarm threshold value to obtain a graded early warning result of infusion speed alarm; uploading the grading early warning result to a centralized monitoring end of a nurse station through wireless communication to obtain a nursing response result; and performing iterative optimization on the prediction model and the fuzzy logic controller based on reinforcement learning. The accuracy of infusion speed alarm can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to an infusion speed alarm method and system. BACKGROUND

[0002] Nowadays, in clinical infusion treatment, for drugs that need to be accurately controlled in infusion speed or special patient groups, two main methods are used to realize flow rate monitoring. One is manual observation of infusion drops, which is dependent on nursing staff to regularly patrol and manually adjust the infusion rate. This method is low in cost but relies on manpower, is easily affected by factors such as fatigue and heavy workload, and has problems such as discontinuous monitoring and delayed response. The other is to use electronic infusion pumps or injection pumps for automatic control, which has high flow rate control accuracy and real-time monitoring function, and is widely used in high-risk scenarios such as intensive care and operating rooms.

[0003] However, electronic infusion pump equipment is high in cost, large in size, and needs to be repeatedly disinfected for use, which not only increases the burden of hospital procurement and maintenance, but also is in short supply in the scenario of multiple patients using at the same time. More importantly, for some sensitive liquids (such as blood products), the extrusion action of the mechanical pump may damage the blood cell components, leading to clinical disuse. The traditional manual observation method cannot realize real-time early warning and is prone to cause excessive liquid load or slow infusion speed due to negligence, which has a great safety hazard. Although there are alarm devices based on sensors, most of them only realize single threshold alarm and lack comprehensive judgment ability for dynamic changes in the infusion process (such as pressure abnormalities, bubbles, blockage, and gradual trend of flow rate), and the system has poor connectivity, and alarm information is difficult to reach mobile medical staff in time, resulting in delayed response. SUMMARY

[0004] The present application provides an infusion speed alarm method and system to solve the technical problems in the prior art.

[0005] The technical solution of the present application to solve the above technical problems is as follows: The present application provides an infusion speed alarm method, which comprises: Based on the photoelectric sensor, the drop speed data is obtained, the pressure and flow rate data are obtained in combination with the pressure sensor and the flow sensor, and the bubble volume information is obtained through the ultrasonic sensor to form a multi-modal perception data set; According to the multi-modal perception data set, it is determined whether there is a drop speed anomaly, a sudden increase in pressure, a large bubble, a pipeline blockage or a backflow condition, and corresponding abnormal feature information is generated; The abnormal feature information is input into the prediction model of the long short-term memory network and the fuzzy logic controller, and the prediction result of the future infusion speed in the infusion process is output, and the alarm threshold is dynamically adjusted according to the prediction result; perform hierarchical early warning based on the prediction result and the dynamically adjusted alarm threshold to obtain a hierarchical early warning result of the infusion speed alarm; upload the hierarchical early warning result to a centralized monitoring end of a nurse station through wireless communication and synchronously push the hierarchical early warning result to a mobile terminal of a nursing staff to trigger a quick response to obtain a nursing response result; store the multi-modal perception data set, the prediction result, records of the hierarchical early warning result, and the nursing response result to a database, and iteratively optimize the prediction model and the fuzzy logic controller based on reinforcement learning to optimize the hierarchical early warning result of future infusion speed alarms.

[0006] Optionally, the multi-modal perception data set is formed by: calculating a drop frequency from the drop speed pulse signal obtained by the photoelectric sensor to obtain the drop speed data; generating pressure-flow rate joint feature data based on the pressure change value detected by the pressure sensor and the flow rate detection result of the flow sensor; obtaining bubble volume information from the echo signal intensity detected by the ultrasonic sensor, and fusing the bubble volume information with the drop speed data and the pressure-flow rate joint feature data to form the multi-modal perception data set.

[0007] Optionally, the multi-modal perception data set is formed by: obtaining a first determination result of drop speed anomaly based on the deviation of the drop speed data in the multi-modal perception data set; obtaining a second determination result of pressure surge and pipeline blockage based on the change trend of the pressure-flow rate joint feature data; obtaining a third determination result of excessive bubbles based on threshold comparison of the bubble volume information; generating the abnormal feature information in combination with the first determination result, the second determination result, and the third determination result.

[0008] Optionally, the abnormal feature information is input into a prediction model and a fuzzy logic controller of a long short-term memory network to output a prediction result of future infusion speed in the infusion process, including: inputting the time series of the abnormal feature information into the prediction model to obtain a future infusion speed prediction value; inputting the instantaneous fluctuation of the abnormal feature information into the fuzzy logic controller to obtain a prediction correction value after noise filtering; fusing the future infusion speed prediction value and the prediction correction value to output the prediction result.

[0009] Optionally, the hierarchical early warning based on the prediction result and the dynamically adjusted alarm threshold comprises: triggering a low-priority prompt when the prediction result is mild abnormality, triggering automatic clamping and voice alarm when the prediction result is severe abnormality, and triggering sound and light alarm and backup resource calling when the prediction result is emergency abnormality.

[0010] Optionally, the low-priority prompt triggered when the prediction result is mild abnormality comprises: generating prompt information and uploading the prompt information to a centralized monitoring end of the nurse station when the prediction result is mild abnormality; pushing a low-priority notification to a mobile terminal of the nursing staff according to the prompt information to prompt the nursing staff to pay attention to the infusion state.

[0011] Optionally, the automatic clamping and voice alarm triggered when the prediction result is severe abnormality comprises: controlling an electromagnetic clamping device to perform automatic clamping operation of an infusion pipeline when the prediction result is severe abnormality; triggering voice broadcast prompt of the nurse station and synchronously pushing a high-priority alarm to the mobile terminal of the nursing staff when the automatic clamping operation is performed.

[0012] Optionally, the sound and light alarm and backup resource calling triggered when the prediction result is emergency abnormality comprises: controlling a sound and light device in a ward to trigger sound and light alarm when the prediction result is emergency abnormality; sending an emergency notification to the nurse station, automatically calling backup infusion equipment resources, and assigning a disposal task to backup nursing staff.

[0013] Optionally, the prediction model and the fuzzy logic controller are iteratively optimized based on reinforcement learning, comprising: calculating a prediction deviation of the prediction model according to the multimodal perception data set, the prediction result and the nursing response result stored in the database; updating parameters of the long short-term memory network prediction model by using a back propagation algorithm according to the prediction deviation; adjusting fuzzy rules of the fuzzy logic controller according to the updated prediction result and hierarchical early warning accuracy to optimize future hierarchical early warning results.

[0014] The application further provides an infusion speed alarm system, which comprises: an abnormality detection module configured to determine whether there is a drip speed anomaly, a sudden pressure surge, a large bubble, a pipeline blockage, or a backflow condition based on the multi-modal perception dataset, and generate corresponding abnormal feature information; a dynamic self-adaptive analysis module configured to input the abnormal feature information into a prediction model of a long short-term memory network and a fuzzy logic controller, output a prediction result of a future infusion speed in an infusion process, and dynamically adjust an alarm threshold based on the prediction result; a hierarchical early warning module configured to perform hierarchical early warning based on the prediction result and the dynamically adjusted alarm threshold, and obtain a hierarchical early warning result of the infusion speed alarm; an intelligent linkage module configured to upload the hierarchical early warning result to a centralized monitoring end of a nurse station through wireless communication, and synchronously push the hierarchical early warning result to a mobile terminal of a nursing staff to trigger a quick response, and obtain a nursing response result; a data archiving and learning module configured to store the multi-modal perception dataset, the prediction result, records of the hierarchical early warning result, and the nursing response result into a database, and iteratively optimize the prediction model and the fuzzy logic controller based on reinforcement learning, so as to optimize the hierarchical early warning result of the future infusion speed alarm.

[0015] In addition, to achieve the above-mentioned purpose, the application further provides an electronic device, which comprises a memory for storing a computer software program, and a processor for reading and executing the computer software program, thereby realizing the infusion speed alarm method as described above.

[0016] In addition, to achieve the above-mentioned purpose, the application further provides a non-transitory computer readable storage medium, wherein the storage medium stores a computer software program, and the computer software program is executed by a processor to realize the infusion speed alarm method as described above.

[0017] The application has the following beneficial effects: (1) The application can cover multiple risk scenarios such as drip speed anomaly, pipeline blockage, bubble entry, and backflow in the infusion process through the fusion of multiple sensors such as photoelectric drip speed, pressure, flow rate, ultrasonic bubble, and weight detection, thereby avoiding the problems of single sensor missed detection and misjudgment, and greatly improving the comprehensiveness and accuracy of detection.

[0018] (2) The application introduces a long short-term memory network (LSTM) and a fuzzy logic controller to predict and denoise the multi-parameter time series, rather than relying on a fixed threshold, but making a judgment based on trends and context data, thereby realizing early warning of infusion anomalies and reducing the probability of delayed alarm or false alarm.

[0019] (3) The application establishes a hierarchical early warning mechanism, differentiates the processing of mild, severe and urgent abnormalities, realizes multi-level response from low priority screen prompt to automatic clamping, sound and light alarm and backup resource calling, guarantees that nursing staff can make quick decisions according to different risk levels, and improves the clinical safety.

[0020] To sum up, the application significantly improves the safety, accuracy and nursing efficiency of the infusion process through the synergistic innovation of multi-modal sensing, dynamic algorithm prediction, hierarchical early warning linkage and AI closed-loop optimization, is especially suitable for intensive care units, operating rooms and sensitive drug infusion scenes, and can effectively solve the problems of high false alarm rate, response delay and lack of individualized adaptation existing in the existing infusion alarm scheme. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 A flowchart of an infusion speed alarm method provided by the application is shown in the figure. Figure 2 A structural schematic diagram of an infusion speed alarm system provided by the application is shown in the figure. Figure 3 A hardware structural schematic diagram of a possible electronic device provided by the application is shown in the figure. Figure 4 A hardware structural schematic diagram of a possible computer readable storage medium provided by the application is shown in the figure. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the application.

[0023] In the description of the application, the terms "first", "second" are only used for description purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0024] In the description of the present application, the term "for example" is used to indicate "serving as an example, instance, or illustration." Any embodiment described as "for example" in the present application is not necessarily to be construed as preferred or advantageous over other embodiments. The following description is presented to enable any person skilled in the art to make and use the present application. In the following description, for the purpose of explanation, details are set forth. It is apparent to those skilled in the art that the present application can be practiced without the use of these specific details. In other instances, well-known structures and processes are not elaborated in order not to obscure the description of the present application with unnecessary details. Thus, the present application is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed.

[0025] Referring to Figure 1 , a flowchart of a transfusion speed alarm method of the present application is provided, comprising the following steps: Step 201, acquiring drop speed data based on a photoelectric sensor, acquiring pressure and flow speed data in combination with a pressure sensor and a flow sensor, and acquiring bubble volume information through an ultrasonic sensor to form a multi-modal perception data set.

[0026] In this embodiment, a photoelectric sensor is arranged at the drip chamber of the transfusion device to acquire the falling frequency of liquid drops in real time, thereby obtaining drop speed data; a pressure sensor and a flow sensor are arranged in the transfusion pipeline to detect the pressure change and flow speed information in the pipeline, respectively; and an ultrasonic sensor is arranged at the outlet of the transfusion pipeline to detect the bubble volume. Through the fusion of the data of the above-mentioned multiple sensors, a multi-modal perception data set containing drop speed, pressure, flow speed, and bubble volume is formed, thereby providing a basis for subsequent abnormality determination.

[0027] In some embodiments, step 201 can comprise: According to the drop speed pulse signal acquired by the photoelectric sensor, the falling frequency of liquid drops is calculated and drop speed data is obtained; According to the pressure change value detected by the pressure sensor, in combination with the flow speed detection result of the flow sensor, pressure-flow speed joint feature data is generated; According to the echo signal intensity detected by the ultrasonic sensor, bubble volume information is obtained, and is fused with the drop speed data and the pressure-flow speed joint feature data to form a multi-modal perception data set.

[0028] In this embodiment, the system first collects the pulse signal of the falling of liquid drops through the photoelectric sensor arranged at the transfusion drip chamber. The frequency of the pulse signal corresponds to the falling speed of the liquid drops. The system counts and analyzes the time interval of the pulse signal, calculates the falling frequency of the liquid drops, and thereby obtains drop speed data.

[0029] Further, the system is equipped with a micro pressure sensor in the infusion pipeline to detect the pressure change value in the pipeline; at the same time, a flow sensor (such as an ultrasonic Doppler flowmeter) is arranged to detect the instantaneous flow rate of the liquid in the pipeline. By combining the pressure change value with the flow rate detection result, the pressure-flow combined feature data is generated to reflect the comprehensive state of pressure and flow rate in the infusion process.

[0030] In addition, the system is equipped with an ultrasonic sensor at the outlet of the infusion pipeline to detect whether there are bubbles in the liquid. The ultrasonic sensor can calculate the size and volume information of the bubbles through the change of the echo signal intensity. By feature extraction and amplitude analysis of the echo signal, the bubble volume information is obtained.

[0031] Finally, the system fuses the drop rate data, pressure-flow combined feature data and bubble volume information to form a multi-modal perception data set. The multi-modal perception data set can comprehensively represent the drop rate, pressure, flow rate and bubble state in the infusion process, and provide a reliable data basis for subsequent anomaly judgment and prediction analysis.

[0032] Step 202, according to the multi-modal perception data set, it is judged whether there is a drop rate anomaly, a pressure surge, a bubble that is too large, a pipeline blockage or a backflow situation, and corresponding abnormal feature information is generated.

[0033] In this embodiment, the system first compares the drop rate data according to the multi-modal perception data set to obtain a drop rate anomaly judgment result; analyzes the trend of the pressure and flow rate data to determine whether there is a pressure surge or a pipeline blockage; compares the bubble volume data to determine whether there is a bubble that is too large; and identifies whether backflow occurs in combination with the weight information. Finally, the system generates corresponding abnormal feature information according to the above determination results for subsequent intelligent prediction.

[0034] In some embodiments, step 202 can include: According to the deviation of the drop rate data in the multi-modal perception data set, a first determination result of the drop rate anomaly is obtained; According to the trend of the pressure-flow combined feature data, a second determination result of the pressure surge and the pipeline blockage is obtained; According to the threshold comparison of the bubble volume information, a third determination result of the bubble that is too large is obtained; In combination with the first determination result, the second determination result and the third determination result, the abnormal feature information is generated.

[0035] In this embodiment, the system first analyzes the drop rate data contained in the multi-modal perception data set. By comparing the real-time drop rate data with the set reference value, if the deviation exceeds the preset range and lasts for a certain time, it is determined that the drop rate is abnormal, and the first determination result of the drop rate abnormality is obtained accordingly.

[0036] Further, the system performs time series analysis on the pressure-flow rate combined feature data. When detecting that the pressure value rises rapidly in a short time and the corresponding flow rate data decreases significantly or tends to zero, the system determines that the pressure of the infusion pipeline suddenly increases or is blocked, thereby obtaining the second determination result.

[0037] In addition, the system performs threshold comparison on the bubble volume information. When detecting that the bubble volume exceeds the preset critical threshold, it is determined that the bubble is too large, and the third determination result is obtained.

[0038] Finally, the system logically fuses the first determination result, the second determination result, and the third determination result to comprehensively generate abnormal feature information. The abnormal feature information can accurately represent various abnormal states in the infusion process, providing a reliable basis for the input of subsequent prediction models and the execution of hierarchical early warning.

[0039] Step 203, input the abnormal feature information into the prediction model of the long short-term memory network and the fuzzy logic controller, output the prediction result of the future infusion speed in the infusion process, and dynamically adjust the alarm threshold according to the prediction result.

[0040] In this embodiment, the abnormal feature information is input as input into the long short-term memory network prediction model and the fuzzy logic controller, the former outputs the infusion speed prediction value in the future period, and the latter performs noise filtering and correction on the instantaneous fluctuation. By fusing the prediction value and the correction value, a more accurate future infusion speed prediction result is obtained, and the alarm threshold is dynamically adjusted according to the prediction result to adapt to the differentiated needs of different patients and drugs.

[0041] The system can not only detect the flow rate of the liquid (unit: ml / h), but also automatically calculate the remaining infusion time after inputting the total amount of the liquid (such as 500ml). For example, when the system detects that the current flow rate is 100ml / h, if the total infusion amount is 500ml, the remaining time is 5 hours. This function enables medical staff to plan to replace new liquid, avoiding the emergency caused by the completion of liquid infusion.

[0042] Through the above design, medical staff can actively master the infusion situation of all patients in the ward, thereby realizing more efficient and accurate nursing management. Patients or their families no longer need to frequently observe whether the liquid is finished or worry about the problem of blockage of the infusion needle for a long time, greatly reducing their anxiety and burden.

[0043] In some embodiments, step 203 can include: According to the time sequence of the abnormal feature information, input into the prediction model to obtain a future infusion speed prediction value; According to the instantaneous fluctuation of the abnormal feature information, input into a fuzzy logic controller to obtain a noise-filtered prediction correction value; Fusing the future infusion speed prediction value and the prediction correction value to output a prediction result.

[0044] In this embodiment, the system first models the time sequence of the generated abnormal feature information, and inputs the time sequence data into a long short-term memory (LSTM) prediction model. The prediction model can output a future infusion speed prediction value according to the time sequence characteristics of the historical data, thereby reflecting the possible infusion abnormal trend in advance.

[0045] Further, the system extracts the instantaneous fluctuation in the abnormal feature information, and inputs the fluctuation information into a fuzzy logic controller. The fuzzy logic controller can perform noise filtering and correction processing on the input instantaneous fluctuation according to the preset fuzzy rules, to obtain a filtered prediction correction value, so as to eliminate the misjudgment caused by occasional bubbles, environmental interference and other factors.

[0046] Finally, the system fuses and calculates the future infusion speed prediction value and the prediction correction value to obtain a final prediction result. In this way, the stability of the time sequence trend and the instantaneous fluctuation can be considered, so that the prediction result can reflect the future trend and has strong robustness, thereby providing more accurate data support for subsequent dynamic threshold adjustment and graded warning.

[0047] Step 204, based on the prediction result and the dynamically adjusted alarm threshold, performing graded warning to obtain a graded warning result of the infusion speed alarm.

[0048] In this embodiment, the system performs a graded warning strategy according to the prediction result and the dynamically adjusted alarm threshold. When the prediction result shows a mild abnormality, a low-priority prompt is triggered; when the prediction result is a serious abnormality, automatic clamping and voice alarm are triggered; when the prediction result is an emergency abnormality, sound and light alarms and backup resource calling are triggered. In this way, a graded response to different degrees of infusion abnormality is realized, and the pertinence and disposal efficiency of the warning are improved.

[0049] Through the above design, medical staff can actively master the infusion situation of all patients in infusion in the whole ward, thereby realizing more efficient and accurate nursing management. Patients or their families no longer need to frequently observe whether the liquid is dropped or not, or worry about the problem of needle blockage after a long time, which greatly reduces their anxiety and burden.

[0050] In some embodiments, step 204 can include: When the prediction result is mild abnormality, a low-priority prompt is triggered; when the prediction result is severe abnormality, automatic clamping and voice alarm are triggered; when the prediction result is emergency abnormality, audible and light alarms and backup resource calling are triggered.

[0051] In the embodiment, the system executes a hierarchical early warning strategy based on the prediction result and the dynamically adjusted alarm threshold.

[0052] Specifically, when the prediction result represents mild abnormality, the system triggers a low-priority prompt. The low-priority prompt includes displaying a flashing mark on a centralized monitoring terminal interface of a nurse station or pushing a reminder message to a nursing staff mobile terminal, so as to prompt the nursing staff to pay attention to the infusion state without immediately interrupting the infusion process.

[0053] When the prediction result represents severe abnormality, the system triggers automatic clamping and voice alarm. The automatic clamping clamps the infusion pipeline by controlling an electromagnetic locking device to prevent the abnormality from continuing to develop; at the same time, the centralized monitoring terminal of the nurse station generates voice broadcast information (for example, prompting “infusion blockage of a certain bed”), so that the nursing staff can quickly locate and dispose.

[0054] When the prediction result represents emergency abnormality, the system triggers audible and light alarms and backup resource calling. The audible and light alarms include starting a buzzer and an LED flashing device in the ward to form a significant local warning effect; at the same time, the system automatically calls backup infusion resources (such as a backup infusion pump or manual disposal personnel) through the internal communication network of the hospital to ensure the safety of patient infusion.

[0055] Through the above hierarchical early warning strategy, the system can trigger different intensity of alarms and disposal measures according to the degree of abnormality, realize a complete closed loop from prompting, intervention to forced guarantee, and improve the safety of the infusion process and the nursing response efficiency.

[0056] In some embodiments, when the prediction result is mild abnormality, a prompt message is generated and uploaded to the centralized monitoring terminal of the nurse station; according to the prompt message, a low-priority notification is pushed to the mobile terminal of the nursing staff to prompt him / her to pay attention to the infusion state.

[0057] In the embodiment, when the prediction result represents mild abnormality, the system first generates a prompt message. The prompt message includes bed number, current drop rate deviation range, and prompt level, etc. Subsequently, the system uploads the prompt message to the centralized monitoring terminal of the nurse station through a wireless communication module to form a visual prompt on the monitoring interface, such as a flashing mark or color change, so that the nursing staff can intuitively perceive the infusion abnormality of the corresponding bed.

[0058] Further, the system pushes a low-priority notification to the mobile terminal (such as a PDA or a smart watch) of the nursing staff based on the prompt information. The low-priority notification can be in the form of a vibration reminder, a pop-up window on the screen, or a message push to prompt the nursing staff to pay attention to the infusion state of the corresponding patient. This step does not force the nursing staff to immediately perform intervention, but provides additional information support, so that the nursing staff can reasonably arrange the inspection time according to the work load.

[0059] In the above manner, the system can realize non-intrusive early warning and reminder in the mild abnormal stage, avoid waste of nursing resources due to excessive alarm, and ensure that the infusion abnormality can be discovered and paid attention to in time, thereby improving the safety of the infusion process and the overall efficiency of the nursing work.

[0060] In some embodiments, when the prediction result is a serious abnormality, the electromagnetic clamping device is controlled to perform automatic clamping operation on the infusion pipeline; at the same time of performing the automatic clamping operation, voice broadcast prompt of the nurse station is triggered, and a high-priority alarm is pushed to the mobile terminal of the nursing staff.

[0061] In this embodiment, when the prediction result represents a serious abnormality, the system first controls the electromagnetic clamping device to perform automatic clamping operation on the infusion pipeline. Through automatic clamping, the abnormal infusion process can be quickly blocked without the immediate presence of the nursing staff, thereby avoiding the risk of drug liquid extravasation, backflow, or abnormal drug dosage input due to continuous infusion.

[0062] At the same time of performing the automatic clamping operation, the system triggers voice broadcast prompt of the nurse station. The voice broadcast broadcasts abnormal information in real time in the form of voice synthesis, such as prompting “3 bed infusion is blocked, please handle in time”, so that the on-duty staff in the nurse station can quickly identify the abnormal patient bed and respond immediately.

[0063] Further, the system also pushes high-priority alarm information to the mobile terminal of the nursing staff through a wireless communication module. The high-priority alarm includes patient information, abnormality category, abnormality time, and automatic clamping execution state, and adopts a forced reminding mode (such as continuous vibration, alarm sound, or eye-catching screen prompt) to ensure that the nursing staff can learn about the abnormal situation in time when they are busy or not in the nurse station.

[0064] In the above manner, when a serious abnormality is detected, the system can realize a multi-channel linkage mechanism of “automatic handling + voice prompt + remote push”, which not only ensures quick and safe handling in the abnormal state, but also significantly improves the response efficiency of the nursing staff.

[0065] In some embodiments, when the prediction result is an emergency exception, the sound and light device in the ward is controlled to trigger a sound and light alarm; at the same time, an emergency notification is sent to the nurse station, a backup infusion device resource is automatically called, and a disposal task is assigned to a backup nursing staff.

[0066] In the present embodiment, when the prediction result is an emergency exception, the system first controls the sound and light device in the ward to trigger a sound and light alarm. The sound and light alarm includes starting a buzzer to emit a high-decibel sound and driving an LED indicator light to flash, thereby forming a strong local warning effect in the ward and ensuring that the patient and surrounding personnel can immediately perceive the exception even if no nursing staff is immediately present.

[0067] At the same time, the system sends an emergency notification to the nurse station through a wireless communication module. The emergency notification includes the identification information of the abnormal bed, the type of the exception, the alarm time, and the current disposal state, and is highlighted on the centralized monitoring interface with the highest priority, so as to ensure that the nursing staff can pay attention and take emergency measures at the first time.

[0068] Further, the system automatically calls a backup infusion device resource, such as a backup infusion pump or a replacement pipeline module, so as to complete resource scheduling preparation before the nursing staff arrives at the scene, thereby shortening the time of subsequent disposal. At the same time, the system automatically assigns the disposal task to a backup nursing staff, and when the primary nursing staff fails to respond within a set time window, the backup nursing staff can receive a forced task push, so as to ensure that the exception is intervened in time.

[0069] In the above manner, the system can realize a multi-level linkage disposal mechanism of “local forced warning + central notification + device resource calling + nursing task allocation” in an emergency exception scenario, thereby not only improving the safety guarantee capability of the infusion process, but also significantly reducing the clinical risk caused by response delay.

[0070] Step 205, uploading the hierarchical early warning result to the centralized monitoring end of the nurse station through wireless communication, and synchronously pushing to the mobile terminal of the nursing staff to trigger a quick response, to obtain a nursing response result.

[0071] In the present embodiment, the hierarchical early warning result is uploaded to the centralized monitoring end of the nurse station in real time through a wireless communication module, and is synchronously pushed to the mobile terminal (such as a PDA or a smart watch) of the nursing staff. The nursing staff can take corresponding disposal measures immediately after receiving the alarm information, to obtain a nursing response result. This manner can significantly shorten the time between exception discovery and disposal, thereby improving the clinical safety.

[0072] In this embodiment, in order to ensure that the nurses can grasp the infusion situation of the patients in real time and respond in time, the system designs a centralized monitoring function of the nurse station and a PDA (Personal Digital Assistant) linkage function. Since the nurses will carry the PDA device with them during work, they can also obtain the infusion state information of the patients in real time through the PDA even if they are not at the nurse station.

[0073] The transmitter is installed on the infusion clamp and is responsible for collecting and sending various parameter data in the infusion process; the receiver includes the screen of the nurse station and the PDA device carried by the nurses. This design enables the nurses to understand the infusion progress and abnormal conditions of the patients at any time regardless of where they are in the ward.

[0074] It is worth noting that the PDA device of the present application has network connection function and can transmit data in real time and keep synchronization with the nurse station. Not only can it improve the efficiency of data transmission, but also it can ensure the accuracy and timeliness of the information.

[0075] Step 206, store the multi-modal perception data set, prediction result, record of hierarchical early warning result and nursing response result to the database, and iteratively optimize the prediction model and fuzzy logic controller based on reinforcement learning to optimize the hierarchical early warning result of future infusion speed alarm.

[0076] In this embodiment, the system stores the multi-modal perception data set, prediction result, hierarchical early warning result and nursing response result to the database. Based on the historical data, the long short-term memory network prediction model and the fuzzy logic controller are iteratively optimized by using the reinforcement learning algorithm, so that the system can gradually improve the prediction accuracy and the reliability of the hierarchical early warning in subsequent operation, thereby realizing the intelligent and adaptive long-term optimization effect.

[0077] In some embodiments, step 206 can include: According to the multi-modal perception data set, prediction result and nursing response result stored in the database, calculate the prediction deviation of the prediction model; According to the prediction deviation, update the parameters of the long short-term memory network prediction model by using the back propagation algorithm; According to the updated prediction result and hierarchical early warning accuracy, adjust the fuzzy rules of the fuzzy logic controller to optimize the future hierarchical early warning result.

[0078] In this embodiment, the system first calculates the prediction deviation of the long short-term memory network prediction model under different scenarios according to the multi-modal perception data set, historical prediction result and actual nursing response result stored in the database. The prediction deviation can be quantified in the form of mean square error or cross entropy loss, etc., which is used to represent the difference between the prediction result and the actual infusion state.

[0079] Further, the system updates the internal parameters of the long short-term memory network using a backpropagation algorithm based on the predicted deviation. Through iterative correction of network weights and biases, the prediction model can gradually reduce the prediction deviation, thereby improving the prediction accuracy of future infusion rate trends.

[0080] Meanwhile, the system adjusts the fuzzy rules of the fuzzy logic controller based on the updated prediction results and the actual accuracy of the hierarchical warning. For example, when the prediction results have frequent false positives or false negatives, the system redefines the judgment interval of the abnormal feature value by optimizing the membership function or modifying the fuzzy reasoning rules, in order to improve the robustness of the fuzzy logic in the case of transient fluctuations.

[0081] In one specific embodiment of the present application, the system not only realizes accurate monitoring and alarm of infusion rate, but also specially designs a humanized service module. This module provides personalized nursing suggestions and services through intelligent algorithm analysis of the specific situation and needs of the patient. For example, the system can automatically adjust the light, sound and other parameters of the infusion environment according to the patient's condition, age, psychological state and other factors, to create a more comfortable and reassuring treatment environment.

[0082] In another specific embodiment, the present application proposes an intelligent management system based on liquid remaining time detection. The system can monitor the infusion progress of patients in multiple wards in real time, and automatically generate an optimal nurse work plan according to the preset priority and time arrangement.

[0083] For example, at a certain time, the system detects that patients with bed numbers 1, 5, 9 and 3 are all undergoing infusion, and their respective remaining times are different. The system will calculate the most reasonable liquid replacement order and time point according to these data, guiding the nurses to enter each room in an orderly manner to replace the liquid. This not only avoids the frequent back-and-forth and time waste that may occur in the traditional call bell mode, but also ensures that each patient can receive timely care at the best time.

[0084] Specifically, when the system predicts that a patient room will soon need to replace the liquid, it will notify the nurse in advance to prepare, and guide him / her to the designated room at the appropriate time. If multiple beds in the same room need to replace the liquid at the same time, the system will also automatically coordinate to ensure that the nurse can complete all operations in the shortest time, reducing unnecessary waiting and moving, so as to leave more time for the patient and provide higher quality of care.

[0085] In this way, the present application not only improves the efficiency and accuracy of nursing work, but also effectively relieves the work pressure of nurses, so that they can focus more on the personalized needs and emotional support of patients, further improving the overall nursing level and social satisfaction.

[0086] Through the above manner, the system can continuously optimize the prediction model and the fuzzy logic controller based on historical data and actual nursing feedback in a long-term running process, form a self-learning and self-adaptive closed-loop optimization mechanism, and thus significantly improve the accuracy and stability of future grading early warning results, and guarantee the safety of the infusion process and the efficiency of the nursing work.

[0087] Please refer to Figure 2 , Figure 2 The structure diagram of the infusion speed alarm system provided by the present application is shown.

[0088] As Figure 2 shown, the infusion speed alarm system provided by the embodiment of the present application comprises: A multi-modal perception acquisition module 301 is configured to acquire drop speed data based on a photoelectric sensor, acquire pressure and flow speed data in combination with a pressure sensor and a flow sensor, and acquire bubble volume information through an ultrasonic sensor to form a multi-modal perception data set; An abnormality detection module 302 is configured to determine whether there is a drop speed anomaly, a sudden increase in pressure, excessively large bubbles, pipeline blockage or backflow based on the multi-modal perception data set, and generate corresponding abnormality feature information; A dynamic self-adaptive analysis module 303 is configured to input the abnormality feature information into a prediction model and a fuzzy logic controller of a long short-term memory network, output a prediction result of future infusion speed in an infusion process, and dynamically adjust an alarm threshold value according to the prediction result; A grading early warning module 304 is configured to perform grading early warning based on the prediction result and the dynamically adjusted alarm threshold value, and obtain a grading early warning result of infusion speed alarm; An intelligent linkage module 305 is configured to upload the grading early warning result to a centralized monitoring end of a nurses' station through wireless communication, and synchronously push the grading early warning result to a mobile terminal of a nursing staff to trigger a quick response, and obtain a nursing response result; A data archiving and learning module 306 is configured to store the multi-modal perception data set, the prediction result, the record of the grading early warning result and the nursing response result to a database, and iteratively optimize the prediction model and the fuzzy logic controller based on reinforcement learning, so as to optimize the grading early warning result of future infusion speed alarm.

[0089] Please refer to Figure 3 , Figure 3 The embodiment of the electronic device provided by the present application is shown. As Figure 3 shown, the electronic device 400 provided by the embodiment of the present application comprises a memory 410, a processor 420 and a computer program 411 stored in the memory 410 and capable of running on the processor 420, and the processor 420 implements the following steps when executing the computer program 411: The drop speed data is acquired based on a photoelectric sensor, the pressure and flow speed data are acquired in combination with a pressure sensor and a flow sensor, and the bubble volume information is acquired through an ultrasonic sensor to form a multi-modal perception data set; According to the multi-modal perception data set, it is determined whether there is a drop speed anomaly, a sudden increase in pressure, excessively large bubbles, pipeline blockage or backflow, and corresponding abnormal feature information is generated; The abnormal feature information is input into a prediction model of a long short-term memory network and a fuzzy logic controller, a prediction result of future infusion speed in an infusion process is output, and an alarm threshold is dynamically adjusted according to the prediction result; Based on the prediction result and the dynamically adjusted alarm threshold, hierarchical early warning is performed to obtain a hierarchical early warning result of infusion speed alarm; The hierarchical early warning result is uploaded to a centralized monitoring end of a nurse station through wireless communication and is synchronously pushed to a mobile terminal of a nursing staff to trigger a rapid response to obtain a nursing response result; The multi-modal perception data set, the prediction result, the record of the hierarchical early warning result and the nursing response result are stored into a database, and the prediction model and the fuzzy logic controller are iteratively optimized based on reinforcement learning to optimize the hierarchical early warning result of future infusion speed alarm.

[0090] Please refer to Figure 4 , Figure 4 An embodiment of a computer readable storage medium provided for the embodiment of the application is shown in the figure. Figure 4 As shown in the figure, the embodiment provides a computer readable storage medium 500, which stores a computer program 411, and the computer program 411 is executed by a processor to implement the following steps: The drop speed data is acquired based on a photoelectric sensor, the pressure and flow speed data are acquired in combination with a pressure sensor and a flow sensor, and the bubble volume information is acquired through an ultrasonic sensor to form a multi-modal perception data set; According to the multi-modal perception data set, it is determined whether there is a drop speed anomaly, a sudden increase in pressure, excessively large bubbles, pipeline blockage or backflow, and corresponding abnormal feature information is generated; The abnormal feature information is input into a prediction model of a long short-term memory network and a fuzzy logic controller, a prediction result of future infusion speed in an infusion process is output, and an alarm threshold is dynamically adjusted according to the prediction result; Based on the prediction result and the dynamically adjusted alarm threshold, hierarchical early warning is performed to obtain a hierarchical early warning result of infusion speed alarm; The hierarchical early warning result is uploaded to a centralized monitoring end of a nurse station through wireless communication and is synchronously pushed to a mobile terminal of a nursing staff to trigger a rapid response to obtain a nursing response result; The multi-modal perception data set, the prediction result, the record of the hierarchical early warning result and the nursing response result are stored to a database, and the prediction model and the fuzzy logic controller are iteratively optimized based on reinforcement learning to optimize the hierarchical early warning result of future infusion speed alarm.

[0091] It should be noted that in the above embodiments, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0092] Those skilled in the art will appreciate that embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer usable program code.

[0093] The present application is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 a system with the function specified in one or more blocks.

[0094] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction system that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 a system with the function specified in one or more blocks.

[0095] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable data processing apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable data processing apparatus provide a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1steps of the functions specified in the block or blocks.

[0096] While the preferred embodiments of the application have been described, additional variations and modifications can be made to the preferred embodiments by those skilled in the art once they learn of the basic inventive concepts. Therefore, the appended claims are intended to encompass within their scope all such variations and modifications as are included within the scope of the application.

[0097] It is clear that the application can be subject to many modifications and variations without departing from the scope of the application as defined by the appended claims. The application is therefore intended to cover all technical and structural equivalents.

Claims

1. A method for alarming infusion rate, characterized in that, The method includes: A multimodal sensing dataset is formed by acquiring drop rate data using photoelectric sensors, combining pressure and flow sensor data to acquire pressure and flow rate data, and acquiring bubble volume information through ultrasonic sensors. Based on the multimodal sensing dataset, determine whether there are abnormal drip rates, sudden pressure increases, excessively large bubbles, pipe blockages, or backflow, and generate corresponding abnormal feature information; The abnormal feature information is input into the prediction model of the long short-term memory network and the fuzzy logic controller, and the prediction result of the future infusion rate during the infusion process is output. The alarm threshold is dynamically adjusted according to the prediction result. Based on the prediction results and the dynamically adjusted alarm threshold, a graded early warning is executed to obtain a graded early warning result for the infusion rate alarm; The tiered early warning results are uploaded to the centralized monitoring terminal of the nurse station via wireless communication and simultaneously pushed to the mobile terminals of nursing staff to trigger a rapid response and obtain nursing response results; The multimodal perception dataset, the prediction results, the records of the graded early warning results, and the nursing response results are stored in the database. The prediction model and the fuzzy logic controller are iteratively optimized based on reinforcement learning to optimize the graded early warning results for future infusion rate alarms.

2. The infusion rate alarm method according to claim 1, characterized in that, The formation of the multimodal sensing dataset includes: The droplet falling frequency is calculated and the droplet speed data is obtained based on the droplet speed pulse signal acquired by the photoelectric sensor. Based on the pressure change value detected by the pressure sensor and the flow velocity detection result of the flow sensor, pressure-flow velocity joint feature data is generated. Based on the echo signal intensity detected by the ultrasonic sensor, the bubble volume information is obtained, and then fused with the drop rate data and the pressure-flow rate joint feature data to form the multimodal sensing dataset.

3. The infusion rate alarm method according to claim 2, characterized in that, The process involves determining, based on the multimodal sensing dataset, whether there are abnormal drip rates, sudden pressure increases, excessively large bubbles, pipe blockages, or backflow, and generating corresponding abnormal feature information, including: Based on the deviation of the drip rate data in the multimodal sensing dataset, a first determination result of drip rate anomaly is obtained; Based on the changing trend of the pressure-flow velocity combined characteristic data, a second determination result of pressure surge and pipeline blockage is obtained; Based on the threshold comparison of the bubble volume information, a third determination result is obtained that the bubble is too large; The abnormal feature information is generated by combining the first determination result, the second determination result, and the third determination result.

4. The infusion rate alarm method according to claim 3, characterized in that, The step of inputting the abnormal feature information into the prediction model of the long short-term memory network and the fuzzy logic controller, and outputting the prediction result of the future infusion rate during the infusion process, includes: Based on the time series of the abnormal feature information, the prediction model is input to obtain the predicted value of the future infusion rate; Based on the instantaneous fluctuations of the abnormal feature information, the data is input to the fuzzy logic controller to obtain the predicted correction value after noise filtering. The predicted future infusion rate is fused with the predicted correction value to output the prediction result.

5. The infusion rate alarm method according to claim 4, characterized in that, The step of executing tiered early warnings based on the prediction results and the dynamically adjusted alarm thresholds includes: When the prediction result is a minor anomaly, a low-priority prompt is triggered; when the prediction result is a severe anomaly, automatic clamping and voice alarm are triggered; when the prediction result is an emergency anomaly, audible and visual alarms and backup resource activation are triggered.

6. The infusion rate alarm method according to claim 5, characterized in that, When the prediction result is slightly abnormal, a low-priority prompt is triggered, including: When the prediction result is slightly abnormal, a prompt message is generated and uploaded to the centralized monitoring terminal of the nurse station; Based on the prompt information, a low-priority notification is pushed to the caregiver's mobile terminal to remind them to pay attention to the infusion status.

7. The infusion rate alarm method according to claim 6, characterized in that, When the prediction result is severely abnormal, automatic clamping and voice alarm are triggered, including: When the predicted result is severely abnormal, the electromagnetic clamping device is controlled to perform an automatic clamping operation on the infusion line. While performing the automatic clamping operation, the voice broadcast prompt at the nurse station is triggered, and a high-priority alarm is simultaneously pushed to the nurse's mobile terminal.

8. The infusion rate alarm method according to claim 7, characterized in that, When the prediction result indicates an emergency anomaly, an audible and visual alarm and backup resource activation are triggered, including: When the predicted result is an emergency anomaly, the sound and light device in the ward is controlled to trigger an sound and light alarm. Simultaneously, an emergency notification is sent to the nursing station, automatically calling up backup infusion equipment resources and assigning the treatment task to the backup nursing staff.

9. The infusion rate alarm method according to claim 8, characterized in that, Iterative optimization of the prediction model and the fuzzy logic controller based on reinforcement learning includes: The prediction bias of the prediction model is calculated based on the multimodal perception dataset stored in the database, the prediction results, and the nursing response results. Based on the prediction bias, the parameters of the Long Short-Term Memory Network prediction model are updated using the backpropagation algorithm; Based on the updated prediction results and the accuracy of the graded early warning, the fuzzy rules of the fuzzy logic controller are adjusted to optimize future graded early warning results.

10. An infusion rate alarm system, characterized in that, The system includes: The multimodal sensing and acquisition module is used to acquire drop rate data based on photoelectric sensors, acquire pressure and flow rate data by combining pressure and flow sensors, and acquire bubble volume information through ultrasonic sensors to form a multimodal sensing dataset. The anomaly detection module is used to determine whether there are abnormal drip rates, sudden pressure increases, excessively large bubbles, pipeline blockages, or backflows based on the multimodal sensing dataset, and to generate corresponding anomaly feature information. The dynamic adaptive analysis module is used to input the abnormal feature information into the prediction model of the long short-term memory network and the fuzzy logic controller, output the prediction result of the future infusion rate during the infusion process, and dynamically adjust the alarm threshold according to the prediction result. The graded early warning module is used to perform graded early warning based on the prediction results and the dynamically adjusted alarm threshold, so as to obtain the graded early warning result of the infusion rate alarm; The intelligent linkage module is used to upload the graded early warning results to the centralized monitoring terminal of the nurse station via wireless communication, and simultaneously push them to the mobile terminals of nursing staff to trigger a rapid response and obtain nursing response results; The data archiving and learning module is used to store the multimodal perception dataset, the prediction results, the records of the graded early warning results, and the nursing response results in the database, and to iteratively optimize the prediction model and the fuzzy logic controller based on reinforcement learning, so as to optimize the graded early warning results for future infusion rate alarms.

Images