Ward call-for-help method and system based on smart medical treatment

Abstract

The invention relates to the technical field of medical informatization, in particular to a ward distress call method and system based on intelligent medical treatment, and the method comprises the steps: obtaining patient state data collected by a plurality of sensors in a ward; receiving a ward help-calling instruction; constructing a multi-modal data fusion matrix based on the patient state data; executing random matrix transformation and feature extraction according to the multi-modal data fusion matrix; based on the extracted features, topological feature mapping and anomaly detection are carried out; executing group theory base risk assessment according to the topological features; calculating a chaotic power system early warning index based on a group theory risk assessment result; determining a call-for-help judgment result according to the early warning index; a corresponding response strategy is executed based on the call-for-help judgment result, the system can comprehensively grasp the state of the patient through the multi-modal data fusion technology, and the evaluation accuracy is greatly improved. And secondly, the robustness of the system is enhanced by random matrix transformation, so that the system can cope with a complex and changeable medical environment.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and more specifically, to a ward emergency call method and system based on smart medical care. Background Art

[0002] With the continuous advancement of medical technology and the increasing aging of the population, the efficient utilization of medical resources and ensuring patient safety have become major challenges facing the current medical system. In this context, the importance of ward call systems, as the key link between patients and medical staff, is self-evident. However, existing ward call systems still have many shortcomings and cannot meet the needs of modern medical care.

[0003] Traditional ward emergency call systems rely primarily on bedside call buttons and regular patrols. While simple and straightforward, this approach suffers from significant drawbacks such as long response times and a tendency to miss reports. Patients may be unable to press the call button for various reasons, while regular patrols struggle to detect sudden changes in a patient's condition. This not only increases the workload of medical staff but can also result in critical situations being delayed, threatening patients' lives.

[0004] In recent years, some hospitals have begun implementing electronic emergency call systems based on vital sign monitoring. These systems continuously monitor patients' heart rate, blood pressure, and other indicators, automatically sounding an alarm when a specific indicator exceeds a preset threshold. While an improvement over traditional methods, this single-indicator, fixed-threshold monitoring approach still presents numerous challenges. First, it fails to fully reflect a patient's true condition, as many dangerous situations are the result of the combined effects of multiple indicators. Second, fixed thresholds are difficult to adapt to individual patient differences, making them prone to false alarms or missed alerts. Furthermore, these systems often only respond passively to existing anomalies and lack predictive and preventative capabilities.

[0005] Recent research attempts to incorporate artificial intelligence (AI) technologies to optimize emergency call systems, such as using machine learning algorithms to analyze patient data. However, these approaches still have limitations: they typically focus on a single type of data (such as physiological indicators), overlooking important information such as environmental factors and patient behavior; secondly, these systems are often black-box systems, making it difficult for medical staff to understand and trust their decision-making processes; and finally, they lack a theoretical foundation, making it difficult to ensure stability and reliability in complex and changing medical environments. Summary of the Invention

[0006] The present invention aims to solve the above-mentioned technical problems existing in the existing ward call system and provide a comprehensive, accurate, reliable and predictive smart medical ward call method and system.

[0007] The present invention provides a ward emergency call method based on smart medical care, comprising:

[0008] The acquisition steps include:

[0009] Obtain patient status data collected by multiple sensors in the ward;

[0010] Receive emergency calls from the ward;

[0011] Processing steps include:

[0012] constructing a multimodal data fusion matrix based on the patient status data;

[0013] Performing random matrix transformation and feature extraction according to the multimodal data fusion matrix;

[0014] Based on the extracted features, topological feature mapping and anomaly detection are performed;

[0015] Perform group-theoretic risk assessment based on topological features;

[0016] Calculate the early warning index of chaotic dynamical systems based on the results of group theory risk assessment;

[0017] Output steps include:

[0018] Determining a call for help judgment result according to the warning index;

[0019] Based on the distress judgment result, a corresponding response strategy is executed.

[0020] Preferably, the construction of the multimodal data fusion matrix specifically includes:

[0021] The vital sign data, action recognition data, speech recognition data and environmental monitoring data are respectively constructed into vectors;

[0022] Assign a corresponding weight vector to each data type;

[0023] Combine the data vector and weight vector into a fusion matrix.

[0024] Preferably, the random matrix transformation and feature extraction specifically include:

[0025] Generate a random matrix;

[0026] Multiplying the random matrix by the multimodal data fusion matrix;

[0027] Apply a nonlinear activation function to the product result to obtain the feature matrix.

[0028] Preferably, the topological feature mapping and anomaly detection specifically include:

[0029] Define a topological space;

[0030] Mapping the characteristic matrix to the topological space;

[0031] Based on the mapping results, abnormal patterns are detected.

[0032] Preferably, the group-based risk assessment specifically includes:

[0033] Define group action operator;

[0034] Applying group action operators to the components of the topological features;

[0035] Calculate the product of group actions to obtain the risk assessment result.

[0036] Preferably, the calculation of the chaotic dynamic system warning index specifically includes:

[0037] Introducing a small perturbation;

[0038] Observe the evolution of the disturbance over time;

[0039] Calculate the Lyapunov exponent as a warning index.

[0040] Preferably, the determination of the call for help judgment result specifically includes:

[0041] Set early warning index thresholds;

[0042] Comparing the calculated warning index with the threshold value;

[0043] Based on the comparison results, determine whether to call for help.

[0044] Preferably, the execution of the response strategy specifically includes:

[0045] When it is determined that there is no need to call for help, continue to monitor the patient's condition;

[0046] When it is determined that a call for help is needed, a call for help instruction is sent;

[0047] If emergency assistance is determined to be necessary, notify the nearest medical personnel.

[0048] As an option, it also includes:

[0049] Adaptively optimize the early warning model based on historical data;

[0050] Regularly update the random matrix to improve system robustness.

[0051] A ward emergency call system based on smart medical care for executing the method includes:

[0052] A data acquisition module is used to obtain patient status data collected by multiple sensors in the ward;

[0053] A distress request receiving module is used to receive ward distress instructions;

[0054] The data processing module is used to construct a multimodal data fusion matrix, perform random matrix transformation and feature extraction, perform topological feature mapping and anomaly detection, perform group theory-based risk assessment, and calculate the chaotic dynamic system warning index;

[0055] A distress call judgment module is used to determine a distress call judgment result based on the warning index;

[0056] The response execution module is used to execute the corresponding response strategy based on the distress call judgment result.

[0057] Specifically, the beneficial effects of the present invention are mainly reflected in the following aspects:

[0058] This method achieves comprehensive assessment and accurate prediction of patient status by innovatively combining advanced mathematical tools such as multimodal data fusion, random matrix theory, topology, group theory, and chaos theory. This approach not only comprehensively considers multidimensional information such as physiological indicators, environmental factors, and patient behavior, but also captures the complex nonlinear relationships between these factors.

[0059] The method of the present invention exhibits significant advantages and innovations in multiple aspects. First, multimodal data fusion technology enables the system to fully grasp the patient's condition, greatly improving the accuracy of the assessment. Second, random matrix transformation enhances the robustness of the system, enabling it to cope with complex and changing medical environments. Furthermore, topological feature mapping and group theory risk assessment provide a solid mathematical foundation for the system, enabling it to capture the inherent structure and complex relationships in the data. Finally, the introduction of chaotic dynamics models gives the system powerful predictive capabilities, enabling it to identify potential dangerous situations in advance.

[0060] These innovative technical features work together and complement each other, forming a highly integrated intelligent system. For example, multimodal data fusion provides rich input for topological feature mapping, which in turn lays the foundation for group-theoretic risk assessment. The results of group-theoretic risk assessment are further utilized by chaotic dynamics models to predict future trends. This synergistic effect not only improves the system's overall performance but also enhances its adaptability and scalability.

[0061] The method of the present invention not only solves many problems in the prior art, but also brings about a series of significant beneficial effects. It greatly improves the accuracy of calls for help and reduces the false alarm rate and missed alarm rate to an unprecedented low level. This not only reduces the workload of medical staff, but also ensures that patients who really need help can receive timely treatment. Secondly, the early warning capability of the present invention provides medical staff with valuable reaction time, which may have a decisive impact when dealing with acute conditions and is expected to significantly improve patient prognosis. Furthermore, the adaptability of the system enables it to be optimized for different patients and different medical environments to provide personalized monitoring plans. Finally, the method of the present invention has good interpretability, which helps to enhance the trust of medical staff in the system and promote its widespread application in clinical practice.

[0062] In general, the smart medical ward distress call method and system provided by the present invention represents a major breakthrough in this field. It can not only improve the quality and efficiency of medical care, but also is expected to promote the development of the entire medical system in a smarter and more precise direction, providing patients with safer and higher-quality medical services. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 Flow chart of the method of the present invention.

[0064] Figure 2 It is a data processing and decision-making flow chart of the present invention. DETAILED DESCRIPTION

[0065] Please refer to Figure 1-2 The present invention provides a ward call for help method and system based on smart medical care. This method achieves comprehensive monitoring and accurate assessment of patient status through multimodal data fusion and advanced mathematical models, thereby improving the accuracy and timeliness of ward call for help.

[0066] Specifically, the method of the present invention comprises the following steps:

[0067] First, in the acquisition step, the method acquires patient status data collected by multiple sensors within the ward and receives a call for help from the ward. These sensors may include, but are not limited to, vital signs monitoring devices, motion sensors, voice recognition devices, and environmental monitoring devices. For example, vital signs monitoring devices can collect data such as a patient's heart rate, blood pressure, body temperature, and blood oxygen saturation; motion sensors can capture changes in the patient's body position and abnormal movements; voice recognition devices can identify a patient's voice commands or abnormal sounds; and environmental monitoring devices can record parameters such as temperature, humidity, and light intensity within the ward.

[0068] Next, in the processing step, the method first constructs a multimodal data fusion matrix based on the acquired patient status data. The purpose of this step is to integrate data from different sources and types into a unified mathematical representation to facilitate subsequent analysis and processing. Specifically, the multimodal data fusion matrix It can be expressed as:

[0069] ,

[0070] in, is the vital sign data vector, is the action recognition data vector, is the speech recognition data vector, is the environmental monitoring data vector. are the corresponding weight vectors. These weights can be adjusted according to the importance of different data types. For example, vital signs data can be given a higher weight because they directly reflect the patient's health status.

[0071] Suppose there is a patient in a hospital room and the sensors collect the following data:

[0072] Vital signs data vector (heart rate, blood pressure, body temperature, blood oxygen saturation); action recognition data vector (whether there has been any abnormal movement such as falling); speech recognition data vector (whether the sound of emergency call is detected); environmental monitoring data vector (temperature, humidity, light intensity).

[0073] The corresponding weight vector is:

[0074] ,

[0075] ,

[0076] ,

[0077] ,

[0078] Then the multimodal data fusion matrix It can be expressed as:

[0079] ,

[0080] This representation enables subsequent processing to consider multiple types of signals simultaneously and weight them according to their importance, thereby improving the accuracy and robustness of the system.

[0081] Then, this method performs random matrix transformation and feature extraction based on the constructed multimodal data fusion matrix. The purpose of this step is to enhance the robustness of the system and extract the most representative features. Random matrix transformation can be achieved by the following formula:

[0082] ,

[0083] in, is a random matrix used to increase the robustness of the system; is a nonlinear activation function, such as ReLU or Sigmoid, is the extracted feature matrix. Random matrix The introduction of can help the system better cope with noise and unknown interference and improve the generalization ability of the model.

[0084] Assuming a random matrix For a simple Matrices, for example:

[0085] ,

[0086] Applying the nonlinear activation function ReLU (i.e. taking the larger of the maximum value 0 and the input value), the feature matrix F can be calculated:

[0087] ,

[0088] Assume that after calculation:

[0089] ,

[0090] By introducing random matrix transformations, the system's resistance to noise and unknown interference is enhanced, and the model's generalization ability is improved. For example, in the presence of equipment noise, the system can still effectively identify abnormal patient conditions.

[0091] Next, this method performs topological feature mapping and anomaly detection based on the extracted features. This step leverages topological concepts to map features into a predefined topological space to capture the inherent structure and relationships of the data. Building on this topological feature mapping, this method further performs group-theoretic risk assessment. The introduction of group theory enables the system to account for complex interactions between features, providing a more comprehensive risk assessment. This approach captures nonlinear relationships and underlying symmetries in the data, providing deeper risk insights.

[0092] Finally, based on the group theory risk assessment results, this method calculates the early warning index of the chaotic dynamical system. The purpose of this step is to quantify the instability of the system and provide a theoretical basis for the early warning mechanism.

[0093] In the output step, this method first determines the call for help based on the calculated warning index. For example, the warning index threshold can be set at 0.5. When the calculated warning index exceeds this threshold, the system determines that a call for help is necessary. This threshold is selected based on extensive clinical data and expert experience and can be fine-tuned based on the specific medical environment.

[0094] Finally, this method executes the corresponding response strategy based on the determined call for help result. For example, if it is determined that no call for help is needed, the system will continue to monitor the patient's condition; if it is determined that a call for help is needed, the system will immediately send a call for help to the medical station; if it is determined that emergency rescue is required, the system will directly notify the nearest medical staff and initiate the emergency plan.

[0095] The method of the present invention achieves a comprehensive, accurate and timely assessment of the patient's condition by comprehensively applying advanced mathematical tools such as multimodal data fusion, random matrix theory, topology, group theory and chaos theory. This method can not only improve the accuracy of calls for help and reduce false alarms and missed alarms, but also predict potential dangerous situations and provide medical staff with more reaction time. In addition, the adaptability and robustness of this method enable it to adapt to different types of patients and changing medical environments, and has broad application prospects. In a preferred embodiment of the present invention, the topological feature mapping and anomaly detection steps are further refined. Specifically, the method first defines a suitable topological space. The selection of this topological space is crucial for subsequent anomaly detection because it determines how to interpret and analyze the structure of the data. In a medical monitoring scenario, a topological space can be selected that can reflect the normal range and change pattern of physiological parameters.

[0096] Next, this method maps the feature matrix to the defined topological space. This mapping process can be achieved by the following formula:

[0097] ,

[0098] in, is the topological mapping function, is the feature matrix, is a predefined topological space, is the topological feature after mapping. Topological mapping function The choice of should take into account the intrinsic structure and medical significance of the data. For example, the theory of persistent homology can be used to construct this mapping function to capture the persistent features in the data.

[0099] Assuming topological mapping function Map the feature matrix F to a predefined topological space For the above feature matrix , assuming that the topological features after mapping as follows:

[0100] ,

[0101] Each row represents a topological feature, such as the changing trend in different dimensions.

[0102] This approach can help uncover potentially unusual patterns. For example, if the data for a particular row deviates significantly from the other rows, this could indicate an anomaly in that dimension. In practical applications, this can help identify worsening conditions or sudden illnesses early on.

[0103] Based on the mapping results, the method further detects abnormal patterns. Anomaly detection can be achieved by comparing current topological features with those of historical data. Preferably, a persistence diagram can be used to represent and compare topological features. If the current topological features differ significantly from historical normal patterns, this may indicate an abnormal situation.

[0104] In another embodiment of the present invention, the group-theoretic risk assessment process is further refined. First, the method defines a set of group action operators. The design of these operators should take into account the interactions and potential symmetries between different physiological parameters. For example, one operator can be defined to describe the relationship between heart rate and blood pressure, and another operator can be defined to describe the relationship between respiratory rate and blood oxygen saturation.

[0105] Then, this method applies these group action operators to each component of the topological feature. This process can be expressed by the following formula:

[0106] ,

[0107] in, is the i-th group action operator, is the i-th component of the topological feature, is the group theory risk assessment result. In this way, the proposed method can capture the complex nonlinear relationship between different physiological parameters.

[0108] Assuming the group action operator For simple linear transformations, such as:

[0109] ,

[0110] Topological features Apply these operators to each component of :

[0111] ,

[0112] The calculation results are:

[0113] ,

[0114] After simplification, we get:

[0115] ,

[0116] This approach can capture nonlinear relationships and underlying symmetries in the data, providing deeper risk insights. For example, when assessing the risk of heart disease patients, combining information from multiple dimensions can provide a more accurate risk assessment.

[0117] Finally, the method calculates the product of the group effects to obtain the final risk assessment result. This result comprehensively considers the status of various physiological parameters and their interactions, providing a comprehensive risk assessment.

[0118] In another preferred embodiment of the present invention, the calculation process of the chaotic dynamic system early warning index is described in detail. First, the method introduces a small disturbance. This disturbance can be understood as a small change in the patient's condition, such as a slight fluctuation in heart rate or a subtle change in blood pressure.

[0119] Next, the method observes the evolution of this disturbance over time. In a medical monitoring scenario, this is equivalent to tracking the changing trend of the patient's status over a short period of time. Preferably, the following formula can be used to describe the evolution of the disturbance:

[0120] ,

[0121] in, is the group theory risk assessment result when time is is the group theory risk assessment result of the initial state. Finally, this method calculates the Lyapunov index as a warning index. The calculation formula of the Lyapunov index is as follows:

[0122] ,

[0123] in, is the Lyapunov exponent, which indicates the system's warning level. In practical applications, an appropriate time window can be selected to approximate this limit. For example, a 15-minute window can be chosen, which is long enough to capture most acute changes in the condition without incurring excessive computational overhead.

[0124] Assume a small perturbation at the initial moment , after a period of time, the small perturbation of the group theory risk assessment result becomes . Then the Lyapunov exponent The calculation is as follows:

[0125] ,

[0126] Assumed time seconds, then:

[0127] ,

[0128] The calculation of the Lyapunov exponent helps the system detect potentially dangerous situations early, providing valuable decision-making support to medical staff. For example, if a patient in intensive care shows a rapid deterioration in their physiological parameters, i.e., a rapid increase in the Lyapunov exponent, this indicates that the patient may be facing an emergency and requires immediate action.

[0129] The method of the present invention effectively quantifies the instability of a patient's condition by introducing the Lyapunov exponent. A positive Lyapunov exponent indicates a chaotic system, potentially foreshadowing a rapid deterioration in the patient's condition. A negative Lyapunov exponent indicates a stable system, potentially implying improvement in the patient's condition.

[0130] The method of the present invention employs a dynamic threshold mechanism to determine the outcome of a call for help. Specifically, the method sets a baseline warning index threshold, such as 0.5. However, this threshold is not fixed but dynamically adjusted based on the patient's specific condition and historical data.

[0131] For example, for patients with chronic diseases, their physiological parameters may fluctuate more significantly than those of the average person, so the threshold can be appropriately raised, such as to 0.6 or 0.7. For patients who have just undergone surgery, their condition may be more fragile, so the threshold can be lowered, such as to 0.4 or 0.3. This dynamic threshold mechanism effectively reduces false positives and false negatives, improving the accuracy and reliability of the emergency call system.

[0132] When the calculated warning index exceeds the set threshold, the method will determine that a call for help is needed. In this case, the system will immediately trigger the corresponding response strategy, such as notifying medical staff or activating an emergency plan.

[0133] Through the above steps, the method of the present invention realizes accurate assessment of the patient's status and timely warning, and provides strong theoretical support and technical implementation for the smart medical ward emergency call system. In a preferred embodiment of the present invention, the execution steps of the response strategy are further refined. When it is determined that there is no need to call for help, the method continues to monitor the patient's status. This continuous monitoring is crucial for timely detection of potential changes in the patient's condition. Preferably, the method can adjust the frequency of data collection to reduce the system load while ensuring the quality of monitoring. For example, for patients in a stable state, the frequency of collecting vital signs can be adjusted from once every 5 minutes to once every 10 minutes.

[0134] When a call for help is determined, this method sends a distress call. This distress call contains not only basic patient information but also the specific cause of the call and relevant physiological data. For example, a distress call might include the following: "Patient Zhang in Ward 12 has a heart rate exceeding 120 beats per minute and blood pressure dropping to 90 / 60 mmHg. Heart failure is suspected. Please address it immediately." This detailed distress call helps medical staff quickly understand the situation and prepare.

[0135] When emergency assistance is determined to be necessary, this method notifies the nearest medical personnel. Preferably, this method can leverage the hospital's positioning system to locate the appropriate medical personnel closest to the patient. For example, if the system determines that the patient may be experiencing a cardiac issue, it will prioritize notifying a nearby cardiologist or nurse experienced in cardiac emergency care. Simultaneously, the system will automatically unlock all electronic doors leading to the patient's room, clearing access for medical personnel and saving valuable time for emergency treatment.

[0136] In another embodiment of the present invention, the method further includes the step of adaptively optimizing the early warning model based on historical data. This adaptive optimization enables the system to continuously learn and improve, adapting to the characteristics of different patients and changes in the medical environment. Specifically, the method can periodically (e.g., weekly or monthly) analyze system performance, including metrics such as call accuracy and response time. Based on these analysis results, the system can automatically adjust various parameters, such as weights in the data fusion matrix, topological feature mapping methods, and group theory risk assessment operators.

[0137] For example, if the system detects a high frequency of certain types of false alarms, it might adjust the weights of related parameters. For example, if the system frequently misclassifies normal movements of elderly patients as falls, it might reduce the weight of the motion recognition data in the fusion matrix or adjust the sensitivity of the motion recognition algorithm. This adaptive optimization mechanism enables the system to continuously improve its performance, providing more accurate and reliable patient monitoring.

[0138] Preferably, the method also includes a step of regularly updating the randomization matrix to improve system robustness. Regularly updating the randomization matrix prevents the system from overfitting to specific data patterns, thereby improving its ability to cope with a variety of unknown situations. For example, the system could automatically generate a new randomization matrix every day at 3:00 AM (typically the hospital's least busy time). This new matrix would be tested on a small scale, and if it performs well, it would be fully implemented the following day. If any issues are detected, the system would automatically roll back to the previous matrix and send an alert to technical staff.

[0139] Finally, the present invention also provides a ward emergency call system based on smart medical care. The system includes a data acquisition module 1, an emergency request receiving module 2, a data processing module 3, an emergency call judgment module 4, and a response execution module 5.

[0140] Data Acquisition Module 1 is used to acquire patient status data collected by multiple sensors within the ward. These sensors may include medical devices such as electrocardiographs, blood pressure monitors, oximeters, and respiratory monitors, as well as auxiliary equipment such as cameras, microphones, and environmental sensors. Data Acquisition Module 1 is not only responsible for collecting this data, but also for preliminary cleaning and preprocessing to ensure the quality of subsequent analysis.

[0141] The distress request receiving module 2 is used to receive distress calls from the ward. This module can receive distress requests from multiple channels, including but not limited to bedside call buttons, distress calls captured by the voice recognition system, and automatic alarms issued by smart wearable devices. The distress request receiving module 2 is also responsible for preliminary classification and prioritization of these requests.

[0142] Data Processing Module 3 is the core of the system. It is responsible for constructing a multimodal data fusion matrix, performing random matrix transformations and feature extraction, performing topological feature mapping and anomaly detection, performing group-theoretic risk assessment, and calculating the chaotic dynamical system warning index. This module integrates all the advanced mathematical processing steps in the proposed method, transforming raw data into meaningful risk assessment results.

[0143] The call-for-help decision module 4 determines the call-for-help decision based on the early warning index. This module implements a dynamic threshold mechanism, adjusting the decision criteria based on the patient's characteristics and current situation. The call-for-help decision module 4 not only provides a binary judgment on whether a call for help is necessary but also provides a more detailed risk level assessment, such as low risk, medium risk, and high risk.

[0144] Response Execution Module 5 is responsible for executing the appropriate response strategy based on the distress call judgment result. This module is responsible for translating the system's judgment into specific actions, such as continuing monitoring, sending an alarm, and notifying medical personnel. Response Execution Module 5 also includes a feedback mechanism that records the results of each response, providing data support for continuous system optimization.

[0145] Through the collaborative work of these modules, the system of the present invention can achieve comprehensive monitoring, accurate assessment and timely response to patients, providing hospitals with an efficient and reliable smart ward emergency solution.

[0146] To verify the superiority of the proposed smart healthcare-based ward call method and system, a simulation experiment was designed to simulate the operation of a medical ward in a medium-sized hospital with 100 wards. The experiment lasted 30 days and involved 300 patients of different types.

[0147] The experimental setup is as follows:

[0148] Example 1: A smart medical ward emergency call system using the present invention.

[0149] Comparative Example 1: Using a traditional nurse call system that relies solely on bedside buttons and scheduled inspections.

[0150] Comparative Example 2: A basic electronic monitoring system is used with simple vital signs monitoring and threshold alarm functions.

[0151] The following key indicators were selected to evaluate system performance:

[0152] 1. Help call accuracy: The percentage of situations that require medical intervention that are correctly identified.

[0153] 2. Average response time: The average time from the occurrence of an abnormal situation to the arrival of medical personnel.

[0154] 3. False alarm rate: The percentage of false alarms triggered as a percentage of the total number of alarms.

[0155] 4. Underreporting rate: The percentage of situations requiring intervention that were not identified.

[0156] 5. Warning lead time: The average time it takes for the system to issue a warning of a potential dangerous situation.

[0157] The detection methods of these indicators are as follows:

[0158] 1. Accuracy of call for help: A team of senior medical staff reviews each call for help to determine its necessity.

[0159] 2. Average response time: calculated by the system automatically recording the time when the abnormality occurred and the time when the medical staff arrived.

[0160] 3. False alarm rate: The medical team determines the necessity of each call for help and calculates the proportion of unnecessary calls for help.

[0161] 4. Missed-report rate: By reviewing patient medical records and surveillance footage, we can identify dangerous situations that the system failed to detect in time.

[0162] 5. Warning lead time: Compare the difference between the system warning time and the actual deterioration time.

[0163] The experimental results are shown in the following table:

[0164] index Example 1 Comparative Example 1 Comparative Example 2 Accuracy of calling for help 95.8% 78.2% 85.6% Average response time 2.3 minutes 8.7 minutes 5.1 minutes False positive rate 3.2% 18.5% 12.7% False negative rate 1.0% 15.3% 8.9% Warning lead time 12.5 minutes 0 minutes 3.2 minutes

[0165] The analysis and discussion are as follows:

[0166] 1. Accuracy of calling for help: The system of the present invention significantly outperforms the other two solutions. This is mainly due to the multimodal data fusion and advanced mathematical model, which enables the system to comprehensively assess the patient's condition and reduce misjudgment.

[0167] 2. Average Response Time: The system of the present invention significantly shortens response time. This is not only because the system can detect problems in a timely manner, but also because it can accurately locate the nearest suitable medical personnel and automatically open the channel.

[0168] 3. False Alarm Rate: The system of the present invention significantly reduces the false alarm rate. This is due to the system's adaptive learning capabilities and dynamic threshold mechanism, which can adapt to the characteristics of different patients.

[0169] 4. Missing alarm rate: The system of the present invention almost eliminates the phenomenon of missing alarms. This is due to the system's comprehensive monitoring and prediction capabilities, which can capture subtle abnormal changes.

[0170] 5. Early Warning Time: The system demonstrated excellent early warning capabilities, detecting potential risks an average of 12.5 minutes in advance. This is due to the system's chaotic dynamics model, which can predict future trends in patient conditions.

[0171] These results fully demonstrate the superiority of the proposed system. It not only significantly improves the accuracy and timeliness of emergency calls, but also demonstrates robust early warning capabilities. This means medical staff can intervene earlier in potentially dangerous situations, significantly improving patient safety. Furthermore, the low false alarm rate means medical resources can be used more efficiently, reducing unnecessary disruptions.

[0172] Notably, the system's advantage in early warning time is particularly significant. An average warning time of 12.5 minutes could have a decisive impact on the treatment of certain acute conditions, such as cardiac arrest and stroke. This could not only save lives but also significantly improve patient outcomes.

[0173] Overall, the proposed smart medical ward emergency call system demonstrates comprehensive advantages, not only improving medical quality and efficiency but also potentially generating substantial economic benefits (such as reduced medical disputes and shorter hospital stays). This system has the potential to become a standard feature of future smart hospitals, providing patients with safer and more precise medical services.

[0174] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modification, replacement, or improvement made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A ward call for help method based on smart medical care, characterized in that: include: The acquisition steps include: Obtain patient status data collected by multiple sensors in the ward; Receive emergency calls from the ward; Processing steps include: constructing a multimodal data fusion matrix based on the patient status data; Performing random matrix transformation and feature extraction according to the multimodal data fusion matrix; Based on the extracted features, topological feature mapping and anomaly detection are performed; Perform group-theoretic risk assessment based on topological features; Calculate the early warning index of chaotic dynamical systems based on the results of group theory risk assessment; Output steps include: Determining a call for help judgment result according to the warning index; Based on the distress judgment result, a corresponding response strategy is executed.

2. The method according to claim 1, characterized in that The construction of the multimodal data fusion matrix specifically includes: The vital sign data, action recognition data, speech recognition data and environmental monitoring data are respectively constructed into vectors; Assign a corresponding weight vector to each data type; Combine the data vector and weight vector into a fusion matrix.

3. The method according to claim 1, characterized in that The random matrix transformation and feature extraction specifically include: Generate a random matrix; Multiplying the random matrix by the multimodal data fusion matrix; Apply a nonlinear activation function to the product result to obtain the feature matrix.

4. The method according to claim 1, wherein The topological feature mapping and anomaly detection specifically include: Define a topological space; Mapping the characteristic matrix to the topological space; Based on the mapping results, abnormal patterns are detected.

5. The method according to claim 1, characterized in that The group-theory-based risk assessment specifically includes: Define group action operator; Applying group action operators to the components of the topological features; Calculate the product of group actions to obtain the risk assessment result.

6. The method according to claim 1, characterized in that The calculation of the chaotic dynamic system early warning index specifically includes: Introducing a small perturbation; Observe the evolution of the disturbance over time; Calculate the Lyapunov exponent as a warning index.

7. The method according to claim 1, characterized in that The determination of the call for help judgment result specifically includes: Set early warning index thresholds; Comparing the calculated warning index with the threshold value; Based on the comparison results, determine whether to call for help.

8. The method according to claim 1, characterized in that The execution of the response strategy specifically includes: When it is determined that there is no need to call for help, continue to monitor the patient's condition; When it is determined that a call for help is needed, a call for help instruction is sent; If emergency assistance is determined to be necessary, notify the nearest medical personnel.

9. The method according to claim 1, characterized in that Also includes: Adaptively optimize the early warning model based on historical data; Regularly update the random matrix to improve system robustness.

10. A ward emergency call system based on smart medical care for executing the method according to any one of claims 1 to 9, characterized in that: include: A data acquisition module is used to obtain patient status data collected by multiple sensors in the ward; A distress request receiving module is used to receive ward distress instructions; The data processing module is used to construct a multimodal data fusion matrix, perform random matrix transformation and feature extraction, perform topological feature mapping and anomaly detection, perform group theory-based risk assessment, and calculate the chaotic dynamic system warning index; A distress call judgment module is used to determine a distress call judgment result based on the warning index; The response execution module is used to execute the corresponding response strategy based on the distress call judgment result.

Images