Sickbed emergency call alarm system

Abstract

The invention relates to an artificial intelligence technology, and discloses a sickbed emergency call alarm system, and the system comprises an emotion data extraction module which obtains environment-related data and physiological data of a target patient, carries out the emotion recognition of the target patient through the environment-related data and the physiological data, and obtains emotion data; the data fusion module is used for performing multi-modal data fusion on the data of the target patient to generate a global state of the patient; the health risk assessment module is used for performing health risk assessment on the patient by using the global state of the patient to obtain a health risk score; the alarm signal generation module is used for comparing the health risk score with a risk threshold value, if the health risk score is larger than the risk threshold value, an alarm signal is generated according to the health risk score of the target patient, the environment related data, the emotion data and the physiological data, and the alarm signal is sent to a preset receiver. The sickbed emergency call alarm system can solve the problem that an existing sickbed emergency call alarm system is insufficient in early warning.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to an emergency call alarm system for hospital beds. Background Technology

[0002] In modern healthcare systems, emergency call alarm systems for hospital beds play an indispensable role, rapidly identifying and responding to patient emergencies to safeguard their lives and health. With advancements in medical technology and evolving patient needs, the pursuit of more efficient and intelligent emergency call alarm systems has become increasingly urgent. To improve healthcare services, numerous innovative technologies and methods have emerged.

[0003] However, most existing emergency call alarm systems rely primarily on monitoring a single physiological parameter, neglecting the impact of a patient's psychological state on their overall health. These systems often fail to effectively capture potential risks arising from emotional changes, resulting in limitations in predicting and preventing emergencies. Summary of the Invention

[0004] This invention provides a bedside emergency call alarm system, the main purpose of which is to solve the problem of insufficient early warning in an existing bedside emergency call alarm system.

[0005] To achieve the above objectives, the present invention provides a bedside emergency call alarm system, comprising: an emotion data extraction module, a data fusion module, a health risk assessment module, and an alarm signal generation module, wherein:

[0006] An emotion data extraction module is used to acquire environmental and physiological data of the target patient, use the environmental and physiological data to identify the emotion of the target patient, obtain emotion data, and transmit the emotion data to the data fusion module.

[0007] The data fusion module is used to perform multimodal data fusion on the environmental data, the physiological data and the received emotional data to generate the patient's global state and transmit the global state to the health risk assessment module.

[0008] The health risk assessment module is used to perform a health risk assessment on the target patient using the received global status of the patient, obtain a health risk score, and transmit the health risk score to the alarm signal generation module.

[0009] An alarm signal generation module is used to compare the received health risk score with a preset risk threshold. If the health risk score is greater than the risk threshold, an alarm signal is generated based on the target patient's health risk score, environmental data, emotional data, and physiological data, and the alarm signal is sent to a preset receiver.

[0010] Optionally, when the emotion data extraction module performs the function of using the environmental data and the physiological data to identify the emotion of the target patient and obtain emotion data, it is specifically used for:

[0011] Emotional features are extracted from the environmental and physiological data.

[0012] The emotional features are classified to obtain emotional categories;

[0013] The emotional intensity is calculated by performing an emotional intensity calculation on the emotional characteristics.

[0014] The emotion category and the emotion intensity are combined to obtain emotion data.

[0015] Optionally, when the emotion data extraction module performs the function of classifying the emotion features to obtain the emotion category, it is specifically used for:

[0016] The emotional features are standardized by feature vector normalization to obtain standard features;

[0017] The standard features are segmented to obtain a data subset;

[0018] For each subset of data, emotion labels are mapped to obtain emotion categories.

[0019] Optionally, when the emotion data extraction module performs the function of calculating the emotion intensity of the emotion features to obtain the emotion intensity, it is specifically used for:

[0020] Calculate the amount of change in the emotional characteristics;

[0021] The intensity score is obtained by weighting and summing the changes using preset feature weights.

[0022] The emotional features are classified according to the intensity score to obtain the emotional intensity.

[0023] Optionally, when the data fusion module performs the function of multimodal data fusion of the environmental data, the physiological data, and the received emotional data to generate the patient's global state, it is specifically used for:

[0024] The environmental data, the emotional data, and the physiological data are synchronized to obtain synchronized data.

[0025] The synchronous data is subjected to intermodal correlation screening to obtain correlation results;

[0026] Intermodal interaction features are extracted from the correlation results to obtain synchronization features;

[0027] The patient's global state is obtained by concatenating the feature vectors of the synchronization features.

[0028] Optionally, when the data fusion module performs the function of synchronizing the environmental data, the emotional data, and the physiological data to obtain synchronized data, it is specifically used for:

[0029] The environmental data, the emotional data, and the physiological data are timestamped to obtain time-aligned data.

[0030] The time-aligned data is resampled to obtain sampled data;

[0031] The sampled data is merged to obtain synchronized data.

[0032] Optionally, when the data fusion module performs the function of filtering intermodal correlations of the synchronized data to obtain correlation results, it is specifically used for:

[0033] Calculate the intermodal correlation coefficient of the synchronized data;

[0034] The correlation coefficients between the modes were subjected to a significance test, and the test results were obtained.

[0035] Based on the test results, synchronous data with a correlation greater than a preset correlation threshold are selected from the intermodal correlation coefficients to obtain the correlation results.

[0036] Optionally, when the data fusion module performs the function of extracting inter-modal interaction features from the correlation results to obtain synchronization features, it is specifically used for:

[0037] Calculate the interaction feature index corresponding to the correlation results;

[0038] Based on the interaction feature indicators and the correlation results, feature clustering is performed to obtain feature combinations;

[0039] The feature combination and the correlation results are merged to generate synchronous features.

[0040] Optionally, when the health risk assessment module performs the function of assessing the health risk of the target patient using the received global patient status to obtain a health risk score, it is specifically used for:

[0041] Perform a convolution operation on the patient's global state to obtain a convolutional feature map;

[0042] Max pooling is performed on the convolutional feature map to obtain a smooth feature map;

[0043] The smooth feature map is combined to obtain comprehensive feature information;

[0044] A health risk score is obtained by performing a nonlinear mapping on the comprehensive feature information.

[0045] Optionally, when the alarm signal generation module performs the function of generating an alarm signal based on the target patient's health risk score, environmental data, emotional data, and physiological data, it is specifically used for:

[0046] The health risk score, the environmental data, the emotional data, and the physiological data are structured to obtain an alarm structure;

[0047] The alarm signal is obtained by format transformation based on the alarm structure.

[0048] This invention achieves comprehensive monitoring of patient status by integrating environmental, physiological, and emotional data. This method not only focuses on physiological indicators but also considers the combined impact of external environment and psychological state on patient health, improving the comprehensiveness and accuracy of monitoring. By fusing data from different sources in a multimodal manner to generate a global patient status, this invention can more accurately reflect the patient's actual condition. This fusion technology helps to discover potential health risks that are difficult to reveal from a single data source, providing a richer information foundation for clinical decision-making. Therefore, the bedside emergency call alarm system proposed in this invention can solve the problem of insufficient early warning in existing bedside emergency call alarm systems. Attached Figure Description

[0049] Figure 1 This is a functional block diagram of a hospital bed emergency call alarm system according to an embodiment of the present invention;

[0050] Figure 2 This is a schematic diagram of an emotion recognition process provided in an embodiment of the present invention;

[0051] Figure 3 A schematic diagram of a data synchronization process provided in an embodiment of the present invention;

[0052] Figure 4 This is a flowchart illustrating an emergency call alarm method for hospital beds, provided in an embodiment of the present invention.

[0053] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0054] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0055] Reference Figure 1 The diagram shown is a functional block diagram of a hospital bed emergency call alarm system according to an embodiment of the present invention. In this embodiment, the hospital bed emergency call alarm system 100 can be installed in an electronic device. Depending on the functions implemented, the hospital bed emergency call alarm system 100 may include an emotion data extraction module 101, a data fusion module 102, a health risk assessment module 103, and an alarm signal generation module 104. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0056] In this embodiment of the invention, the emotion data extraction module 101 is used to acquire environmental and physiological data of the target patient, use the environmental and physiological data to identify the emotion of the target patient, obtain emotion data, and transmit the emotion data to the data fusion module.

[0057] In this embodiment of the invention, the data fusion module 102 is used to perform multimodal data fusion on the environmental data, the physiological data and the received emotional data to generate the patient's global state, and transmit the global state to the health risk assessment module.

[0058] In this embodiment of the invention, the health risk assessment module 103 is used to perform a health risk assessment on the target patient using the received global status of the patient, obtain a health risk score, and transmit the health risk score to the alarm signal generation module.

[0059] In this embodiment of the invention, the alarm signal generation module 104 is used to compare the received health risk score with a preset risk threshold. If the health risk score is greater than the risk threshold, an alarm signal is generated based on the target patient's health risk score, environmental data, emotional data, and physiological data, and the alarm signal is sent to a preset receiver.

[0060] In detail, the modules of the emergency call alarm system 100 for hospital beds described in this embodiment of the invention employ the same technical means as the emergency call alarm method for hospital beds described in the accompanying drawings, and can produce the same technical effects, which will not be repeated here.

[0061] The following describes, with reference to specific embodiments, each component and its specific workflow of the aforementioned emergency call alarm system for hospital beds:

[0062] The emotion data extraction module 101 is configured to obtain the environment-related data and physiological data of the target patient, perform emotion recognition on the target patient by using the environment-related data and the physiological data to obtain emotion data, and transmit the emotion data to the data fusion module.

[0063] In the embodiments of the present invention, the environment-related data of the target patient is obtained through sensors or external APIs, including temperature sensors, humidity sensors, air quality sensors, noise sensors, illuminance sensors, etc.

[0064] In the embodiments of the present invention, the physiological data of the target patient is collected by wearable devices (such as smart bracelets and smart watches) for data such as heart rate, blood oxygen saturation, respiratory rate, and blood pressure.

[0065] In the embodiments of the present invention, by combining the environment-related data and the physiological data, the health status of the patient can be evaluated more comprehensively, and the emotional changes of the patient, especially negative emotions (such as anxiety and depression), can be detected in a timely manner.

[0066] See Figure 2 As shown, when the emotion data extraction module executes the function of performing emotion recognition on the target patient by using the environment-related data and the physiological data to obtain emotion data, it is specifically configured to:

[0067] S21. Extract emotion features from the environment-related data and the physiological data;

[0068] S22. Classify the emotion features to obtain emotion categories;

[0069] S23. Calculate the emotion intensity of the emotion features to obtain the emotion intensity;

[0070] S24. Combine the emotion categories and the emotion intensity to obtain emotion data.

[0071] In the embodiments of the present invention, emotion features are information extracted from environment-related data and physiological data, and these features can reflect the emotional state of the patient. Emotion features include physiological data such as heart rate, blood pressure, galvanic skin response (GSR), and respiratory rate, as well as environmental factors such as temperature, humidity, and noise.

[0072] Furthermore, physiological data feature extraction includes: Heart rate variability (HRV): By analyzing heart rate fluctuations, heart rate variability (HRV) can be extracted, which is related to emotions (such as stress and anxiety); Skin conductance (GSR): Increased GSR usually indicates emotional tension or anxiety, and data can be extracted using skin conductance sensors; Breathing patterns: Changes in respiratory rate and depth are often associated with emotional changes (such as anxiety, relaxation, etc.). Environmental data feature extraction includes: Temperature and humidity: Environmental temperature and humidity may affect a patient's mood. For example, a high-temperature environment may trigger irritability or restlessness; Noise: High noise levels are often associated with mood fluctuations (such as irritability or anxiety); Light intensity: Darker environments may trigger depressive moods, while brighter environments may be associated with positive moods.

[0073] In this embodiment of the invention, when the emotion data extraction module performs the function of classifying the emotion features to obtain the emotion category, it is specifically used for:

[0074] The emotional features are standardized by feature vector normalization to obtain standard features;

[0075] The standard features are segmented to obtain a data subset;

[0076] For each subset of data, emotion labels are mapped to obtain emotion categories.

[0077] In detail, the purpose of eigenvector standardization is to transform feature values ​​at different scales to the same order of magnitude. Standardization is obtained by subtracting the corresponding mean from the original value of the emotion feature and then dividing by the variance. Typically, each feature is scaled to a normal distribution with a mean of 0 and a standard deviation of 1.

[0078] In detail, the standardized feature vector is divided into several data subsets, each containing a portion of the feature vector. By dividing the standard features into subsets, the index range of each subset is determined. For example, if the feature vector of the standard features has a length of n, it is divided into m subsets, each with a length of n / m. Subsets are extracted from the standardized standard features according to the index range to obtain the data subsets.

[0079] In detail, the emotion label mapping is performed using the following formula:

[0080] L i =argmax k (∑j∈S i w jk ·F norm [j])

[0081] Among them, L i Let w be the sentiment category corresponding to the i-th data subset. jkLet F be the weight of the j-th standard feature in the k-th sentiment label. norm [j] represents the j-th element in the standard feature, S i For the i-th data subset, argmax k () indicates selecting one emotion label k from all emotion labels such that the weighted sum is maximized.

[0082] In this embodiment of the invention, when the emotion data extraction module performs the function of calculating the emotion intensity of the emotion features to obtain the emotion intensity, it is specifically used for:

[0083] Calculate the amount of change in the emotional characteristics;

[0084] The intensity score is obtained by weighting and summing the changes using preset feature weights.

[0085] The emotional features are classified according to the intensity score to obtain the emotional intensity.

[0086] In this embodiment of the invention, the intensity of emotion is typically determined by analyzing the amplitude or amount of fluctuation in characteristics. Dramatic changes in certain physiological characteristics (such as a sudden increase in heart rate or significant fluctuations in skin conductance) may indicate a strong emotional response.

[0087] In specific terms, for example, a sudden increase in heart rate variability (HRV) may indicate a stronger emotional response. A sharp change in respiratory rate may indicate a higher intensity of anxiety.

[0088] In detail, different weights can be assigned to different emotional features, and a weighted sum can be made based on their influence on the intensity of the emotion. Important features (such as heart rate or skin conductance) will have a greater impact on the intensity calculation.

[0089] In this embodiment of the invention, emotion category and emotion intensity are combined into a structured data format. Common data structures include dictionaries, lists, or data frames.

[0090] In this embodiment of the invention, by combining environmental and physiological data, the accuracy and reliability of emotion recognition are improved. It can monitor the patient's emotional state in real time, detect emotional abnormalities in a timely manner, and improve the personalization and adaptability of recognition.

[0091] The data fusion module 102 is used to perform multimodal data fusion on the environmental data, the physiological data and the received emotional data to generate the patient's global state, and transmit the global state to the health risk assessment module.

[0092] In the embodiments of the present invention, through data fusion, data redundancy and conflicts can be reduced, the consistency and reliability of data can be improved, a comprehensive patient status view can be formed, and more comprehensive information can be provided.

[0093] When the data fusion module in the embodiments of the present invention executes the function of performing multi-modal data fusion on the environment-related data, the physiological data, and the received emotion data to generate the global patient status, it is specifically used for:

[0094] Synchronize the environment-related data, the emotion data, and the physiological data to obtain synchronized data;

[0095] Perform inter-modal correlation screening on the synchronized data to obtain a correlation result;

[0096] Extract the inter-modal interaction features from the correlation result to obtain synchronized features;

[0097] Perform feature vector splicing on the synchronized features to obtain the global patient status.

[0098] See Figure 3 As shown, when the data fusion module executes the function of synchronizing the environment-related data, the emotion data, and the physiological data to obtain synchronized data, it is specifically used for:

[0099] S31. Align the timestamps of the environment-related data, the emotion data, and the physiological data to obtain timestamp-aligned data;

[0100] S32. Resample the timestamp-aligned data to obtain sampled data;

[0101] S33. Merge the sampled data to obtain synchronized data.

[0102] Specifically, by reading the environment-related data, the emotion data, and the physiological data, then parsing the timestamps in the environment-related data, the emotion data, and the physiological data into a unified format, and finally aligning the timestamps of the data to ensure that the timestamp formats of all data are consistent.

[0103] Specifically, data resampling refers to the process of converting time series data from one sampling frequency to another. By setting the timestamp column as the index of the data frame, using the preset resampling frequency to perform data conversion on the timestamp-aligned data to obtain converted frequency data, and then calculating the average value of all data within each time period for the converted frequency data to obtain sampled data, where data conversion can be performed by calling the resample method. <C

[0104] In detail, data from different data sources are merged into a single data frame based on timestamps, and then missing values ​​are processed in the merged data to obtain synchronized data.

[0105] Furthermore, the merging method can be selected as follows: inner join: only retains timestamps that exist in both data frames; left join: retains all timestamps in the left data frame, and the values ​​corresponding to timestamps that are not in the right data frame are missing values; right join: retains all timestamps in the right data frame, and the values ​​corresponding to timestamps that are not in the left data frame are missing values; outer join: retains all timestamps in both data frames, and fills in the missing values ​​where there are no corresponding values.

[0106] In this embodiment of the invention, when the data fusion module performs the function of filtering intermodal correlations of the synchronized data to obtain correlation results, it is specifically used for:

[0107] Calculate the intermodal correlation coefficient of the synchronized data;

[0108] The correlation coefficients between the modes were subjected to a significance test, and the test results were obtained.

[0109] Based on the test results, synchronous data with a correlation greater than a preset correlation threshold are selected from the intermodal correlation coefficients to obtain the correlation results.

[0110] In detail, Pearson correlation coefficients were calculated for each feature of environmental, emotional, and physiological data to obtain a linear correlation coefficient matrix, which represents the degree of linear dependence between the two modes.

[0111] In detail, a significance test is performed on each correlation coefficient (e.g., using p-values) to determine the statistical significance of each correlation result.

[0112] Specifically, based on a pre-set correlation threshold (such as 0.5 or 0.7), synchronous data combinations with a correlation greater than the correlation threshold are filtered out, and only highly correlated synchronous data combinations are retained.

[0113] In this embodiment of the invention, when the data fusion module performs the function of extracting inter-modal interaction features from the correlation results to obtain synchronization features, it is specifically used for:

[0114] Calculate the interaction feature index corresponding to the correlation results;

[0115] Based on the interaction feature indicators and the correlation results, feature clustering is performed to obtain feature combinations;

[0116] The feature combination and the correlation results are merged to generate synchronous features.

[0117] In detail, the interaction feature index is generated by applying methods such as product, ratio, and difference to the correlation results to generate new features.

[0118] For example, the interaction feature index can be calculated using the following formula.

[0119] Interaction n，k =Feature n *Feature k

[0120] Interaction n，k The interaction feature metric calculated for the nth and kth correlation results. n The feature value corresponding to the nth correlation result. k This is the feature value corresponding to the kth correlation result.

[0121] In detail, cluster analysis is performed on interaction items and original features to discover the aggregation relationship of different modal features in specific contexts (such as using K-means or hierarchical clustering) and identify common feature combination patterns.

[0122] In detail, by vectorizing the correlation results and the feature combination to form an interaction feature vector, and then using a preset feature importance score to perform a weighted summation of the interaction feature vector, a synchronization feature is obtained.

[0123] In this embodiment of the invention, through data fusion, the system can process and analyze data more efficiently, improve response speed and accuracy, make full use of information from various data sources, and avoid information silos.

[0124] The health risk assessment module 103 is used to perform a health risk assessment on the target patient using the received global status of the patient, obtain a health risk score, and transmit the health risk score to the alarm signal generation module.

[0125] In this embodiment of the invention, the global patient status generated based on multimodal data fusion can be used for more accurate health risk assessment.

[0126] In this embodiment of the invention, when the health risk assessment module performs the function of assessing the health risk of the target patient using the received global patient status to obtain a health risk score, it is specifically used for:

[0127] Perform a convolution operation on the patient's global state to obtain a convolutional feature map;

[0128] Max pooling is performed on the convolutional feature map to obtain a smooth feature map;

[0129] The smooth feature map is combined to obtain comprehensive feature information;

[0130] A health risk score is obtained by performing a nonlinear mapping on the comprehensive feature information.

[0131] In this embodiment of the invention, the patient's global state data is typically a multidimensional array, such as a two-dimensional image or a time-series data. One or more convolutional kernels (also called filters) are selected; these kernels are small multidimensional arrays used to slide across the input data. Convolution operation: the convolutional kernels are slid across the input data, and the dot product between the convolutional kernel and the corresponding region of the input data is calculated each time the kernel slides, generating a new value. These new values ​​form a convolutional feature map.

[0132] In this embodiment of the invention, a preset pooling window (usually a small square region, such as 2x2) is used to slide across the convolutional feature map. Each time the window slides, the maximum value within the window is taken to generate a new value. These new values ​​form a smooth feature map. Max pooling reduces the spatial size of the feature map while retaining the most important features, which helps to reduce computational cost and the risk of overfitting.

[0133] In this embodiment of the invention, if multiple convolutional kernels generate multiple convolutional feature maps, these feature maps will generate multiple smooth feature maps after max pooling. These smooth feature maps are then stacked together according to the channel dimension to form a multi-channel feature map. Typically, one or more fully connected layers (DenseLayer) are used to flatten the multi-channel feature map and perform feature combination to generate a fixed-length feature vector. Through linear combination and activation function of the fully connected layer, the information of different feature maps is fused together to form comprehensive feature information.

[0134] In this embodiment of the invention, one or more fully connected (dense) layers are used to perform nonlinear mapping on the comprehensive feature information. These fully connected layers typically contain activation functions (such as ReLU, Sigmoid, Tanh, etc.) to introduce nonlinear characteristics. The last fully connected layer usually has only one neuron, used to output a health risk score. An appropriate activation function can be selected based on the task requirements. For example, if the score range is between 0 and 1, a Sigmoid activation function can be used; if the score range is real values, a linear activation function can be used. The final output value is the patient's health risk score, which can be used to assess the patient's health status.

[0135] In this embodiment of the invention, based on health risk scores, patients' health problems can be managed more effectively, thereby improving treatment outcomes.

[0136] The alarm signal generation module 104 is used to compare the received health risk score with a preset risk threshold. If the health risk score is greater than the risk threshold, an alarm signal is generated based on the target patient's health risk score, environmental data, emotional data, and physiological data, and the alarm signal is sent to a preset receiver.

[0137] In this embodiment of the invention, high-risk situations can be detected in a timely manner by comparing with a preset risk threshold; an alarm signal is generated and automatically sent to a preset receiver to achieve automated management.

[0138] In this embodiment of the invention, the preset receiver refers to, but is not limited to, the following types: mobile devices of medical staff: such as smartphones, tablets, or smartwatches, which can be used by medical staff to receive alarm signals in real time and thus take swift action; hospital monitoring systems: the central monitoring system inside the hospital can receive alarm signals and display them on the screens at the nurses' station or doctors' offices, helping medical staff to understand the patient's condition in a timely manner.

[0139] In this embodiment of the invention, when the alarm signal generation module performs the function of generating an alarm signal based on the target patient's health risk score, environmental data, emotional data, and physiological data, it is specifically used for:

[0140] The health risk score, the environmental data, the emotional data, and the physiological data are structured to obtain an alarm structure;

[0141] The alarm signal is obtained by format transformation based on the alarm structure.

[0142] In this embodiment of the invention, health risk scores, environmental data, emotional data, and physiological data are organized into a structured form, such as a dictionary or class object.

[0143] In this embodiment of the invention, structured data is converted into a selected format, such as JSON, XML, or a custom format.

[0144] In this embodiment of the invention, the alarm signal generation module, in performing the step of sending the alarm signal to a preset receiver, includes:

[0145] Generate a receiver address based on the receiver;

[0146] A sending request is generated based on the receiver address and the alarm signal;

[0147] The alarm signal is sent to the receiver using the sending request.

[0148] In this embodiment of the invention, the generated sending request is sent out via a network or other communication method to reach a predetermined receiver. This may involve using different communication protocols, such as HTTP, SMTP (for email), SMS (short message service), etc., depending on the nature of the receiver and the configuration of the alarm system.

[0149] In this embodiment of the invention, timely alarms and interventions can detect and handle high-risk situations, reducing the risk of patients developing serious health problems.

[0150] like Figure 4 The diagram shown is a flowchart illustrating a bedside emergency call alarm method according to an embodiment of the present invention. In this embodiment, the bedside emergency call alarm method includes:

[0151] S401, Emotional data extraction module, used to acquire environmental and physiological data of the target patient, use the environmental and physiological data to identify the emotion of the target patient, obtain emotional data, and transmit the emotional data to the data fusion module;

[0152] S402, Data fusion module, used to perform multimodal data fusion on the environmental data, the physiological data and the received emotional data to generate the patient's global state, and transmit the global state to the health risk assessment module;

[0153] S403, Health Risk Assessment Module, used to perform health risk assessment on the target patient using the received global status of the patient, obtain a health risk score, and transmit the health risk score to the alarm signal generation module;

[0154] S404, Alarm signal generation module, used to compare the received health risk score with a preset risk threshold. If the health risk score is greater than the risk threshold, an alarm signal is generated based on the target patient's health risk score, environmental data, emotional data, and physiological data, and the alarm signal is sent to a preset receiver.

[0155] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0156] Furthermore, it is clear that the word "including" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or systems described in the system can also be implemented by a single unit or system through software or hardware. Terms such as "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0157] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A hospital bed emergency call alarm system, characterized in that, The system includes an emotion data extraction module, a data fusion module, a health risk assessment module, and an alarm signal generation module, wherein: An emotion data extraction module is used to acquire environmental and physiological data of the target patient, use the environmental and physiological data to identify the emotion of the target patient, obtain emotion data, and transmit the emotion data to the data fusion module. The data fusion module is used to perform multimodal data fusion on the environmental data, the physiological data and the received emotional data to generate the patient's global state and transmit the global state to the health risk assessment module. The health risk assessment module is used to perform a health risk assessment on the target patient using the received global status of the patient, obtain a health risk score, and transmit the health risk score to the alarm signal generation module. An alarm signal generation module is used to compare the received health risk score with a preset risk threshold. If the health risk score is greater than the risk threshold, an alarm signal is generated based on the target patient's health risk score, environmental data, emotional data, and physiological data, and the alarm signal is sent to a preset receiver.

2. The emergency call alarm system for hospital beds as described in claim 1, characterized in that, When the emotion data extraction module performs the function of identifying emotions in the target patient using the environmental data and the physiological data to obtain emotion data, it is specifically used for: Emotional features are extracted from the environmental and physiological data. The emotional features are classified to obtain emotional categories; The emotional intensity is calculated by performing an emotional intensity calculation on the emotional characteristics. The emotion category and the emotion intensity are combined to obtain emotion data.

3. The emergency call alarm system for hospital beds as described in claim 2, characterized in that, When the emotion data extraction module performs the function of classifying the emotion features to obtain the emotion category, it is specifically used for: The emotional features are standardized by feature vector normalization to obtain standard features; The standard features are segmented to obtain a data subset; For each subset of data, emotion labels are mapped to obtain emotion categories using the following formula: Among them, L i Let w be the sentiment category corresponding to the i-th data subset. jk Let F be the weight of the j-th standard feature in the k-th sentiment label. norm [j] represents the j-th element in the standard feature, S i For the i-th data subset, argmax k () indicates selecting one emotion label k from all emotion labels such that the weighted sum is maximized.

4. The emergency call alarm system for hospital beds as described in claim 2, characterized in that, When the emotion data extraction module performs the function of calculating the emotion intensity of the emotion features to obtain the emotion intensity, it is specifically used for: Calculate the amount of change in the emotional characteristics; The intensity score is obtained by weighting and summing the changes using preset feature weights. The emotional features are classified according to the intensity score to obtain the emotional intensity.

5. The emergency call alarm system for hospital beds as described in claim 1, characterized in that, When the data fusion module performs the function of multimodal data fusion of the environmental data, the physiological data, and the received emotional data to generate the patient's global state, it is specifically used for: The environmental data, the emotional data, and the physiological data are synchronized to obtain synchronized data. The synchronous data is subjected to intermodal correlation screening to obtain correlation results; Intermodal interaction features are extracted from the correlation results to obtain synchronization features; The patient's global state is obtained by concatenating the feature vectors of the synchronization features.

6. The emergency call alarm system for hospital beds as described in claim 5, characterized in that, When the data fusion module performs the function of synchronizing the environmental data, the emotional data, and the physiological data to obtain synchronized data, it is specifically used for: The environmental data, the emotional data, and the physiological data are timestamped to obtain time-aligned data. The time-aligned data is resampled to obtain sampled data; The sampled data is merged to obtain synchronized data.

7. The emergency call alarm system for hospital beds as described in claim 5, characterized in that, When the data fusion module performs the function of filtering intermodal correlations of the synchronized data to obtain correlation results, it is specifically used for: Calculate the intermodal correlation coefficient of the synchronized data; The correlation coefficients between the modes were subjected to a significance test, and the test results were obtained. Based on the test results, synchronous data with correlation greater than a preset correlation threshold are selected from the intermodal correlation coefficients to obtain the correlation results.

8. A bedside emergency call alarm system as described in claim 5, characterized in that, When the data fusion module performs the function of extracting inter-modal interaction features from the correlation results to obtain synchronization features, it is specifically used for: The interaction feature index corresponding to the correlation result is calculated, and the interaction feature index can be calculated using the following formula: Interaction n，k =Feature n *Feature k Interaction n，k The interaction feature metric calculated for the nth and kth correlation results. n The feature value corresponding to the nth correlation result. k This is the feature value corresponding to the k-th correlation result; Based on the interaction feature indicators and the correlation results, feature clustering is performed to obtain feature combinations; The feature combination and the correlation results are merged to generate synchronous features.

9. A bedside emergency call alarm system as described in claim 1, characterized in that, When the health risk assessment module performs the function of assessing the health risk of the target patient using the received global patient status and obtaining a health risk score, it is specifically used for: Perform a convolution operation on the patient's global state to obtain a convolutional feature map; Max pooling is performed on the convolutional feature map to obtain a smooth feature map; The smooth feature map is combined to obtain comprehensive feature information; A health risk score is obtained by performing a nonlinear mapping on the comprehensive feature information.

10. A bedside emergency call alarm system as described in claim 1, characterized in that, When the alarm signal generation module performs the function of generating an alarm signal based on the target patient's health risk score, environmental data, emotional data, and physiological data, it is specifically used for: The health risk score, the environmental data, the emotional data, and the physiological data are structured to obtain an alarm structure; The alarm signal is obtained by format transformation based on the alarm structure.

Images