Remote monitoring system with intelligent voice

Abstract

The invention relates to the technical field of medical health monitoring, and discloses a remote monitoring system with intelligent voice, which comprises an intelligent monitoring module, a communication network module, a cloud platform module and a user terminal module, the intelligent monitoring module comprises a multi-mode sensor unit, a voice interaction unit and an edge calculation unit; by arranging the intelligent monitoring module, intelligent recognition and analysis of user voice are facilitated, natural voice instructions are supported, the problem that a traditional system is complex in operation is solved, meanwhile, multi-modal data are fused, a user personalized baseline is dynamically adjusted based on historical data of a user, and the user experience is improved. Therefore, the method can adapt to users of different ages and disease types, generates the environmental factors to correct the personalized baseline, and can adjust the physiological parameter weight according to the users of different disease types, thereby improving the accuracy of the personalized baseline, effectively reducing the probability of missing report and false report, and improving the intelligence.

Description

Technical Field

[0001] This invention relates to the field of medical and health monitoring technology, and more specifically to a remote monitoring system with intelligent voice control. Background Technology

[0002] With the development of technology, remote monitoring systems have been widely used in fields such as medical care, smart homes, and elderly care. Most existing remote monitoring systems use monitoring devices such as sensors and cameras to collect physiological parameters and transmit the data to the monitoring platform via the network to achieve real-time monitoring of the monitored object.

[0003] However, existing remote monitoring systems still have shortcomings in terms of interactivity, convenience, and intelligence. Traditional systems rely on physical buttons or mobile apps for manual operation, which results in a poor interactive experience and low convenience for users with limited mobility or declining vision, making it difficult for them to independently complete operations such as device startup and parameter query. At the same time, the health baselines of different users vary greatly, and existing systems are unable to dynamically adjust, leading to some users experiencing missed or over-reported health status. The intelligence of these systems still needs to be improved.

[0004] In view of this, the present invention proposes a remote monitoring system with intelligent voice, which solves the problems of difficult interaction, high false alarm rate and slow response of traditional monitoring systems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies in the prior art, this invention provides a remote monitoring system with intelligent voice, which lowers the operational threshold through intelligent voice interaction, improves early warning accuracy by combining multimodal physiological and environmental data fusion, achieves low-latency response through edge-cloud collaborative computing, and supports dynamic threshold adjustment based on the user's health status, thereby solving the problems existing in the background art.

[0006] The present invention provides the following technical solution: a remote monitoring system with intelligent voice, comprising an intelligent monitoring module, a communication network module, a cloud platform module, and a user terminal module; The intelligent monitoring module is used to collect multimodal data and realize voice interaction and intelligent monitoring; the intelligent monitoring module includes a multimodal sensor unit, a voice interaction unit, and an edge computing unit; The communication network module is used for data transmission and supports Bluetooth pairing. Dual-link; The cloud platform module is used to store data in the cloud; The user terminal module is used for remote viewing and intervention by the guardian; The multimodal sensor unit is used to acquire multimodal data through the multimodal sensor group; The voice interaction unit is used for intelligent voice interaction through a voice device; the voice device includes a microphone array, a voice recognition chip, and a voice synthesis chip. The edge computing unit is used to fuse multimodal data and perform anomaly detection based on the fused multimodal data.

[0007] Preferably, the multimodal sensor group includes a photoelectric heart rate sensor, a blood pressure sensor, a blood oxygen sensor, a triaxial accelerometer, and a temperature and humidity sensor; the multimodal data includes the physiological parameters and environmental parameters of the monitored person; the physiological parameters include heart rate, systolic blood pressure, blood oxygen saturation, and accelerometer modulus; the environmental parameters include ambient temperature and ambient humidity.

[0008] Preferably, the specific process of the voice interaction unit performing intelligent voice interaction through the voice device is as follows: The system wakes up the intelligent voice; it collects ambient sound through a microphone array, filters out non-user voice through voiceprint recognition, and activates the system when a wake word is detected and the confidence level is greater than or equal to a certain threshold. At this time, activate the voice interaction function and wake up the intelligent voice; The voice commands are parsed; the input voice signal is converted into text by a voice recognition chip, the intent is extracted by natural language processing and matched with a preset command library, thereby obtaining the parsed command; The parsed instructions are executed and feedback is provided; if the parsed instruction is a query instruction, i.e. an instruction to query data, the query instruction is executed to perform the query and a voice response is generated; if the parsed instruction is an operation instruction, i.e. an instruction that requires an operation, the operation instruction is executed.

[0009] Preferably, the specific process by which the edge computing unit performs anomaly detection based on the fused multimodal data is as follows: Obtain historical multimodal data of the ward over the previous 30 days to establish a personalized baseline for the user; Based on real-time multimodal data, the weighted sum of deviations of each physiological parameter from the baseline is obtained to form an abnormality score; Set thresholds and obtain the warning level and physiological parameters that trigger the warning based on the anomaly score.

[0010] Preferably, the user personalization baseline is expressed by the formula: ;in, In multimodal data, the first... The baseline of each physiological parameter; In multimodal data, the first... Historical mean values ​​of several physiological parameters; In multimodal data, the first... The historical standard deviation of a physiological parameter; the user is the person under guardianship.

[0011] Preferably, the anomaly score is expressed by the formula: ;in, Indicates abnormal rating; Indicates the first The weights of each physiological parameter; Indicates the first Real-time values ​​of several physiological parameters This represents the total number of physiological parameters; ; This represents the confidence score for a fall event; if a fall event is detected, then... If no fall event is detected, then .

[0012] Preferably, the step of obtaining the early warning level based on anomaly scoring specifically involves: If abnormal rating satisfy If the situation is abnormal, the warning level is considered mild; at this time, a voice alert will be given regarding the abnormal situation. If abnormal rating satisfy If the warning level is abnormal, the warning level is moderate; at this time, a warning message is pushed to the user terminal module to inform the guardian. If abnormal rating satisfy or If the warning level is severe, immediately call emergency services and send location information to the user's terminal. The and These are the first threshold and the second threshold, respectively.

[0013] Preferably, the user-personalized baseline is corrected based on real-time environmental parameters, specifically as follows: After real-time collection of environmental parameters, the degree of environmental deviation from the baseline is obtained, thereby generating environmental factors. ; The environmental factors are expressed by the following formula: ;in, Indicates environmental factors; Indicates the first One environmental parameter; This indicates the total number of environmental parameters; Indicates the first The weights of each environmental parameter; Indicates the first The values ​​of each environmental parameter; Indicates the first The system preset baseline values ​​for each environmental parameter; based on The user personalization baseline has been revised as follows: The standard deviation is corrected, and the corrected standard deviation is expressed as follows: ;in, This represents the corrected standard deviation. This represents the standard deviation before correction.

[0014] Preferably, the weights of the physiological parameters are adjusted according to the user's disease type, specifically in the following manner: Labels are set for the disease types of the wards; the wards fill out a questionnaire during registration, and the disease types are labeled by the questionnaire and historical multimodal data, and a correspondence table between diseases and sensitive parameters is established; the correspondence table between diseases and sensitive parameters is a table consisting of disease types and corresponding physiological parameters, and the corresponding physiological parameters are sensitive parameters, that is, physiological parameters that can characterize disease types. The initial weights of physiological parameters are set based on disease type labels; the disease type of the ward is selected, the corresponding physiological parameters are obtained, and the weights of the corresponding physiological parameters are adjusted accordingly. ; Weights are optimized using reinforcement learning based on historical multimodal data.

[0015] Preferably, the detection of the fall event specifically includes: Acceleration data is acquired using a triaxial accelerometer, including the resultant acceleration, attitude angle, and duration; the duration refers to the length of time the resultant acceleration exceeds a threshold value; the threshold value is selected as follows. ,in, The acceleration due to Earth's gravity is expressed as: ;in, Indicates the resultant acceleration. Indicates acceleration at Values ​​on the axis Indicates acceleration at Values ​​on the axis Indicates acceleration at Values ​​on the axis; If conditions one, two, and three are met simultaneously, then it is determined to be a fall. Otherwise, it is determined that no fall occurred. ; The first condition is The second condition is: ,in, This represents the change in attitude angle; condition three is... ,in, Indicates duration.

[0016] The technical effects and advantages of this invention are as follows: This invention, by incorporating an intelligent monitoring module, facilitates intelligent recognition and analysis of user voice, supports natural voice commands, and thus solves the problem of complex operation in traditional systems. Simultaneously, it fuses multimodal data and dynamically adjusts the user's personalized baseline based on historical data, adapting to users of different ages and disease types. It also generates environmental factors to correct the personalized baseline and adjusts physiological parameter weights according to different disease types to improve the accuracy of the personalized baseline, effectively reducing the probability of missed and false alarms, enhancing intelligence. Furthermore, it employs a communication network module and a user terminal module for timely response, thus effectively improving the accuracy of early warnings. Attached Figure Description

[0017] Figure 1 This is a structural diagram of a remote monitoring system with intelligent voice control according to the present invention.

[0018] Figure 2 This is a structural diagram of the intelligent monitoring module of the present invention. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The remote monitoring system with intelligent voice involved in the present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] like Figure 1 As shown, the present invention provides a remote monitoring system with intelligent voice, including an intelligent monitoring module, a communication network module, a cloud platform module, and a user terminal module; The intelligent monitoring module is used to collect multimodal data and realize voice interaction and intelligent monitoring; the intelligent monitoring module includes a multimodal sensor unit, a voice interaction unit, and an edge computing unit; The communication network module is used for data transmission and supports Bluetooth pairing. Dual links ensure communication redundancy and smooth operation; The cloud platform module is used to store data in the cloud; The user terminal module is used by the guardian to remotely view and intervene in order to ensure the safety of the ward; The multimodal sensor unit is used to acquire multimodal data through a multimodal sensor group; the multimodal sensor group includes, but is not limited to, photoelectric heart rate sensors, blood pressure sensors, blood oxygen sensors, triaxial accelerometers, and temperature and humidity sensors; the multimodal data includes the physiological parameters and environmental parameters of the monitored person; the physiological parameters include, but are not limited to, heart rate, systolic blood pressure, blood oxygen saturation, and accelerometer modulus; the environmental parameters include, but are not limited to, ambient temperature and ambient humidity; The voice interaction unit is used for intelligent voice interaction through a voice device; the voice device includes a microphone array, a voice recognition chip, and a voice synthesis chip. The edge computing unit is used to fuse multimodal data and perform anomaly detection based on the fused multimodal data.

[0021] In this embodiment, it should be specifically explained that the specific process of the voice interaction unit performing intelligent voice interaction through the voice device is as follows: The system wakes up the intelligent voice; it collects ambient sound through a microphone array, filters out non-user voice through voiceprint recognition, and activates the system when a wake word is detected and the confidence level is greater than or equal to a certain threshold. When activated, the voice interaction function is turned on to wake up the intelligent voice; the wake-up word can be set by the guardian or the ward. The voice commands are parsed; the input voice signal is converted into text by a voice recognition chip, the intent is extracted by natural language processing and matched with a preset command library, thereby obtaining the parsed command; The parsed instructions are executed and feedback is provided; if the parsed instruction is a query instruction, that is, an instruction to query data, the query instruction is executed to perform the query and a voice response is generated; if the parsed instruction is an operation instruction, that is, an instruction to perform an operation, the operation instruction is executed. Its purpose is to enable the ward to directly complete information inquiry and control operations through voice without manual operation, which is especially suitable for the wards with visual or physical disabilities, and effectively improves convenience.

[0022] In this embodiment, it should be specifically explained that the specific process of the edge computing unit performing anomaly detection based on the fused multimodal data is as follows: Obtain historical multimodal data of the ward over the previous 30 days to establish a personalized baseline for the user; The user personalization baseline is expressed by the formula: ;in, In multimodal data, the first... The baseline of each physiological parameter; In multimodal data, the first... Historical mean values ​​of several physiological parameters; In multimodal data, the first... The historical standard deviation of each physiological parameter is used to characterize the range of normal physiological fluctuations for a user; for example, if a personalized baseline for a user's heart rate is established, then... ;in, This represents the baseline of a user's heart rate in multimodal data; This represents the historical mean of a user's heart rate in multimodal data; This represents the historical standard deviation of a user's heart rate in multimodal data; the user is the person under guardianship. Based on real-time multimodal data, the weighted sum of deviations of each physiological parameter from the baseline is obtained to form an abnormality score; The anomaly score is expressed by the formula: ;in, Indicates abnormal rating; Indicates the first The weights of each physiological parameter can be adjusted according to the user's disease type; Indicates the first Real-time values ​​of several physiological parameters This represents the total number of physiological parameters; ; This represents the confidence score for a fall event; if a fall event is detected, then... If no fall event is detected, then ; Set thresholds and obtain the warning level and physiological parameters that trigger the warning based on the anomaly score; The specific method for obtaining the early warning level based on anomaly scoring is as follows: If abnormal rating satisfy If the situation is abnormal, the warning level is considered mild; at this time, a voice alert will be given regarding the abnormal situation. If abnormal rating satisfy If the warning level is abnormal, the warning level is moderate; at this time, a warning message is pushed to the user terminal module to inform the guardian. If abnormal rating satisfy or If the warning level is abnormal, the alert level is severe; in this case, an emergency call should be made immediately and the location information should be sent to the user's terminal. The and The first threshold and the second threshold are respectively. and The value can be set by those skilled in the art based on the actual disease condition of the ward. This embodiment does not limit the specific value. If the actual disease condition of the ward is more serious, the threshold value can be set to a smaller value. If the actual disease condition of the ward is less serious, the threshold value can be set normally. Its purpose is to reduce the false alarm rate caused by a single indicator by fusing multiple parameters, and to dynamically adjust the weights according to the actual situation of the user, thereby effectively improving the accuracy of the early warning.

[0023] In this embodiment, it should be specifically noted that the user-personalized baseline is corrected based on real-time environmental parameters to avoid false positives caused by environmental changes. Specifically: After real-time collection of environmental parameters, the degree of environmental deviation from the baseline is obtained, thereby generating environmental factors. ; The environmental factors are expressed by the following formula: ;in, Indicates environmental factors; Indicates the first One environmental parameter; This indicates the total number of environmental parameters; Indicates the first The weights of each environmental parameter; Indicates the first The values ​​of each environmental parameter; Indicates the first The system preset baseline values ​​for each environmental parameter; the range of environmental factors is as follows: , The larger the value, the more significant the environmental deviation from the baseline; based on The user personalization baseline has been revised as follows: The standard deviation is corrected, and the corrected standard deviation is expressed as follows: ;in, This represents the corrected standard deviation. This represents the standard deviation before correction; The weights of the environmental parameters are determined by the correlation coefficient. Set the correlation coefficient The higher The larger the value; ; ;in, This indicates the baseline correlation between environmental parameters and physiological parameters; This represents the mean value of environmental parameters; This represents the baseline mean of physiological parameters. Time indicates a positive correlation. Time indicates a negative correlation. The higher the value, the higher the correlation.

[0024] In this embodiment, it should be specifically explained that the process of filtering non-user voice through voiceprint recognition is as follows: After collecting ambient sound through a microphone array, preprocessing is performed to reduce noise interference. The preprocessing operation uses Wiener filtering for noise reduction and detects effective speech segments by short-time energy and frequency detection to remove non-speech segments such as silence and cough. Voiceprint features are extracted from the preprocessed speech segment; the speech segment is divided into frames, and the features of each frame are extracted. Features are combined to form a feature sequence, which is then processed by a time-delay neural network. Sequence encoding is a fixed length Vectors to characterize user voiceprint features; The probabilistic linear discriminant analysis score of the voiceprint features and the pre-stored user voiceprint model is obtained, and the confidence level is determined based on the probabilistic linear discriminant analysis score; the pre-stored user voiceprint model is the user's voiceprint feature database, which stores the user's voiceprint embedding vector features for authentication during real-time voice interaction. The confidence level decision is specifically as follows: After obtaining voiceprint feature extraction The similarity between the vector and the voiceprint features of the pre-stored user voiceprint model is used to obtain the probabilistic linear discriminant analysis score, which is expressed by the formula: ;in, This represents the probabilistic linear discriminant analysis score. This indicates the voiceprint feature extraction process. vector; Indicates user, Indicates background noise. express Belongs to the user The posterior probability, based on the user The mean and covariance are obtained; express Background noise The probability of; Set confidence threshold ,like If so, it is determined to be user voice; if If the voice is not a user's voice, it is determined to be non-user voice; the confidence threshold. The confidence threshold can be set by those skilled in the art; in this embodiment, a confidence threshold is selected. Confidence threshold The specific values ​​can be adjusted according to user preferences.

[0025] In this embodiment, it should be specifically explained that the step of extracting intent using natural language processing and matching it with a preset instruction library specifically involves: use The model extracts user intent from text, including query intent, control intent, and help intent; through The model outputs an intent probability distribution, and the intent with the highest probability is selected as the user's intent. Extract key information from the intent, and search for and match the corresponding operation in a preset instruction library based on the intent and the extracted key information; the preset instruction library consists of intent, key information, and execution action; the corresponding operation that is successfully matched in the preset instruction library is used as the parsed instruction.

[0026] In this embodiment, it should be specifically explained that the specific method for adjusting the weights of physiological parameters according to the user's disease type is as follows: Labels are set for the disease types of the wards; the wards fill out a questionnaire during registration, and the disease types are labeled by the completed questionnaire and historical multimodal data, and a correspondence table between diseases and sensitive parameters is established; the disease types include, but are not limited to, hypertension, coronary heart disease, and diabetes; the labels are the labels that represent the disease types; the correspondence table between diseases and sensitive parameters is a table consisting of disease types and corresponding physiological parameters, and the corresponding physiological parameters are the sensitive parameters, which are physiological parameters that can represent the disease types. For example, the physiological parameter corresponding to coronary heart disease is heart rate; the physiological parameter corresponding to hypertension is systolic blood pressure. The initial weights of physiological parameters are set based on disease type labels; the disease type of the ward is selected, the corresponding physiological parameters are obtained, and the weights of the corresponding physiological parameters are adjusted accordingly. If multiple corresponding physiological parameters exist, the sum of the weights of the multiple physiological parameters is: The sum of the weights of the remaining physiological parameters is The sum of the weights of all physiological parameters is ; Weights are optimized using reinforcement learning based on historical multimodal data; if a user recently triggers multiple alerts due to a certain physiological parameter and is subsequently diagnosed, the weight corresponding to that physiological parameter is increased by an increment of [value missing]. If a user recently triggered an alert due to an excess of a certain physiological parameter but was not diagnosed, the weight corresponding to that physiological parameter will be reduced by a certain amount. .

[0027] In this embodiment, it should be specifically explained that the detection of the fall event is as follows: Acceleration data is acquired using a triaxial accelerometer, including the resultant acceleration, attitude angle, and duration; the duration refers to the length of time the resultant acceleration exceeds a threshold value; the threshold value is selected as follows. ,in, The acceleration due to Earth's gravity is expressed as: ;in, Indicates the resultant acceleration. Indicates acceleration at Values ​​on the axis Indicates acceleration at Values ​​on the axis Indicates acceleration at Values ​​on the axis; If conditions one, two, and three are met simultaneously, then it is determined to be a fall. Otherwise, it is determined that no fall occurred. ; The first condition is The second condition is: ,in, This represents the change in attitude angle; condition three is... ,in, Indicates duration.

[0028] In this embodiment, it should be specifically noted that the physiological parameter triggering the early warning is the physiological parameter corresponding to the summation term with the largest value in the abnormality scoring formula. The physiological parameter triggering the early warning is labeled as... , ,in, This represents the maximum value function; therefore, it makes The largest value These are the physiological parameters that trigger the early warning.

[0029] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0030] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A remote monitoring system with intelligent voice control, characterized in that: It includes an intelligent monitoring module, a communication network module, a cloud platform module, and a user terminal module; The intelligent monitoring module is used to collect multimodal data and realize voice interaction and intelligent monitoring; the intelligent monitoring module includes a multimodal sensor unit, a voice interaction unit, and an edge computing unit; The communication network module is used for data transmission and supports Bluetooth pairing. Dual-link; The cloud platform module is used to store data in the cloud; The user terminal module is used for remote viewing and intervention by the guardian; The multimodal sensor unit is used to acquire multimodal data through the multimodal sensor group; The voice interaction unit is used for intelligent voice interaction through a voice device; the voice device includes a microphone array, a voice recognition chip, and a voice synthesis chip. The edge computing unit is used to fuse multimodal data and perform anomaly detection based on the fused multimodal data.

2. The remote monitoring system with intelligent voice as described in claim 1, characterized in that: The multimodal sensor group includes a photoelectric heart rate sensor, a blood pressure sensor, a blood oxygen sensor, a triaxial accelerometer, and a temperature and humidity sensor; the multimodal data includes the physiological parameters and environmental parameters of the monitored person; the physiological parameters include heart rate, systolic blood pressure, blood oxygen saturation, and accelerometer modulus; the environmental parameters include ambient temperature and ambient humidity.

3. The remote monitoring system with intelligent voice as described in claim 2, characterized in that: The specific process by which the voice interaction unit performs intelligent voice interaction through the voice device is as follows: The system wakes up the intelligent voice; it collects ambient sound through a microphone array, filters out non-user voice through voiceprint recognition, and activates the system when a wake word is detected and the confidence level is greater than or equal to a certain threshold. At this time, activate the voice interaction function and wake up the intelligent voice; The voice commands are parsed; the input voice signal is converted into text by a voice recognition chip, the intent is extracted by natural language processing and matched with a preset command library, thereby obtaining the parsed command; The parsed instructions are executed and feedback is provided; if the parsed instruction is a query instruction, i.e. an instruction to query data, the query instruction is executed to perform the query and a voice response is generated; if the parsed instruction is an operation instruction, i.e. an instruction that requires an operation, the operation instruction is executed.

4. A remote monitoring system with intelligent voice as described in claim 3, characterized in that: The specific process by which the edge computing unit performs anomaly detection based on the fused multimodal data is as follows: Obtain historical multimodal data of the ward over the previous 30 days to establish a personalized baseline for the user; Based on real-time multimodal data, the weighted sum of deviations of each physiological parameter from the baseline is obtained to form an abnormality score; Set thresholds and obtain the warning level and physiological parameters that trigger the warning based on the anomaly score.

5. A remote monitoring system with intelligent voice as described in claim 4, characterized in that: The user personalization baseline is expressed by the formula: ;in, In multimodal data, the first... The baseline of each physiological parameter; In multimodal data, the first... Historical mean values ​​of several physiological parameters; In multimodal data, the first... The historical standard deviation of a physiological parameter; the user is the person under guardianship.

6. A remote monitoring system with intelligent voice as described in claim 5, characterized in that: The anomaly score is expressed by the formula: ;in, Indicates abnormal rating; Indicates the first The weights of each physiological parameter; Indicates the first Real-time values ​​of several physiological parameters This represents the total number of physiological parameters; ; This represents the confidence score for a fall event; if a fall event is detected, then... If no fall event is detected, then .

7. A remote monitoring system with intelligent voice as described in claim 6, characterized in that: The specific method for obtaining the early warning level based on anomaly scoring is as follows: If abnormal rating satisfy If the situation is abnormal, the warning level is considered mild; at this time, a voice alert will be given regarding the abnormal situation. If abnormal rating satisfy If the warning level is abnormal, the warning level is moderate; at this time, a warning message is pushed to the user terminal module to inform the guardian. If abnormal rating satisfy or If so, the warning level is severe anomaly; Immediately call emergency services and send location information to the user's device. The and These are the first threshold and the second threshold, respectively.

8. A remote monitoring system with intelligent voice as described in claim 7, characterized in that: The user-personalized baseline is adjusted based on real-time environmental parameters, specifically as follows: After real-time collection of environmental parameters, the degree of environmental deviation from the baseline is obtained, thereby generating environmental factors. ; The environmental factors are expressed by the following formula: ;in, Indicates environmental factors; Indicates the first One environmental parameter; This indicates the total number of environmental parameters; Indicates the first The weights of each environmental parameter; Indicates the first The values ​​of each environmental parameter; Indicates the first The system preset baseline values ​​for each environmental parameter; based on The user personalization baseline has been revised as follows: The standard deviation is corrected, and the corrected standard deviation is expressed as follows: ;in, This represents the corrected standard deviation. This represents the standard deviation before correction.

9. A remote monitoring system with intelligent voice as described in claim 8, characterized in that: The weights of the physiological parameters are adjusted according to the user's disease type, specifically in the following manner: Labels are set for the disease types of the wards; the wards fill out a questionnaire during registration, and the disease types are labeled by the questionnaire and historical multimodal data, and a correspondence table between diseases and sensitive parameters is established; the correspondence table between diseases and sensitive parameters is a table consisting of disease types and corresponding physiological parameters, and the corresponding physiological parameters are sensitive parameters, that is, physiological parameters that can characterize disease types. The initial weights of physiological parameters are set based on disease type labels; Select the disease type of the ward, obtain the corresponding physiological parameters, and adjust the weights of the corresponding physiological parameters accordingly. ; Weights are optimized using reinforcement learning based on historical multimodal data.

10. A remote monitoring system with intelligent voice as described in claim 9, characterized in that: The detection of the fall event specifically involves: Acceleration data is acquired using a triaxial accelerometer, including the resultant acceleration, attitude angle, and duration; the duration refers to the length of time the resultant acceleration exceeds a threshold value; the threshold value is selected as follows. ,in, The acceleration due to Earth's gravity is expressed as: ;in, Indicates the resultant acceleration. Indicates acceleration at Values ​​on the axis Indicates acceleration at Values ​​on the axis Indicates acceleration at Values ​​on the axis; If conditions one, two, and three are met simultaneously, then it is determined to be a fall. Otherwise, it is determined that no fall occurred. ; The first condition is The second condition is: ,in, This represents the change in attitude angle; condition three is... ,in, Indicates duration.

Images