Iot device unified communication and data integration platform for smart ward

Abstract

The application belongs to the technical field of digital information communication, and discloses an Internet of Things equipment unified communication and data integration platform for a smart ward; comprising: a data acquisition module, which acquires medical equipment analog operation data, performs data cleaning, and obtains a multi-source medical equipment data set; a delay compensation module, which performs data delay estimation and delay compensation on the multi-source medical equipment data set; an association correction module, which constructs an association event chain for a time sequence correction data set; an abnormality filtering module, which performs transmission signal abnormality filtering and outputs a fluctuation elimination data set; an inhibition adjustment module, which acquires ward broadcast information, performs inhibition correction on the ward broadcast information, and obtains inhibition correction broadcast information; and a unified transmission module, which integrates the fluctuation elimination data set and the inhibition correction broadcast information into a transmission data stream, and uses the transmission data stream for equipment communication; the accuracy and reliability of the smart ward Internet of Things equipment communication are improved.

Description

Technical Field

[0001] This invention relates to the field of digital information and communication technology, and more specifically, to a unified communication and data integration platform for Internet of Things (IoT) devices in smart wards. Background Technology

[0002] As an important component of modern medical informatization, smart wards are widely used in medical institutions at all levels due to their ability to achieve intelligent interconnection of medical equipment and real-time monitoring of patient status. Among them, unified communication and data integration of equipment are the core technical links in the construction of smart wards, which directly affect the accuracy of medical data and the timeliness of system response. At present, traditional communication and data integration systems can realize basic equipment communication functions, but they are often limited in more complex ward communication scenarios and still face many technical bottlenecks.

[0003] In the multi-source device data acquisition phase of medical scenarios, different types of IoT medical devices upload data through different communication paths. For example, heart rate data is transmitted directly via WiFi, while body temperature data is relayed via Zigbee. However, traditional communication and data integration systems only focus on device clock synchronization and ignore the differences in transmission path delays, easily causing pseudo-timing problems where events that occur first arrive later. Furthermore, in smart wards, certain patient behaviors often trigger multiple device responses. For example, a patient getting up will sequentially cause changes in mattress pressure, triggering of a position sensor, and heart rate fluctuations. However, traditional communication and data integration systems process these causally related data streams independently, resulting in the loss of causal association markers after the data reaches the platform, making it impossible to identify which data stream is causing the causal relationship. These data belong to the same event chain; at the same time, there are a large number of objects such as metal bed frames and IV stands in the ward environment. The movement of medical staff will change the multipath propagation of wireless signals, resulting in excessive fluctuations in RSSI signals. Traditional communication and data integration systems determine the location of devices based on RSSI signals, which can easily misinterpret signal fluctuations caused by environmental factors as device movement, triggering unnecessary relocation processes. On the other hand, when multiple wards trigger emergency calls at the same time, a broadcast storm will be generated. Traditional communication and data integration systems use broadcast storm suppression mechanisms to suppress subsequent calls as repeated broadcasts. However, these are actually independent emergency events from different locations, causing some emergency calls to be covered, affecting the communication reliability of the smart ward.

[0004] In view of this, the present invention proposes a unified communication and data integration platform for IoT devices in smart wards to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a unified communication and data integration platform for IoT devices in smart wards, comprising:

[0006] The data acquisition module collects simulated operating data from medical equipment and performs data cleaning to obtain a multi-source medical equipment dataset.

[0007] The delay compensation module estimates data delay based on a multi-source medical device dataset, and then compensates for the delay based on the estimation results to obtain a time-corrected dataset.

[0008] The correlation correction module constructs a correlation event chain for the time-series correction dataset, outputs the event chain identifier field, and labels the time-series correction dataset to obtain the correlation labeled dataset.

[0009] The anomaly filtering module performs transmission signal anomaly filtering on the associated labeled dataset and outputs a fluctuation-free dataset.

[0010] The suppression adjustment module collects ward broadcast information, performs oversuppression correction on the ward broadcast information, and obtains suppressed and corrected broadcast information;

[0011] The unified transmission module integrates the fluctuation cancellation dataset and suppression correction broadcast information into a transmission data stream, and uses this transmission data stream for device communication; the modules are connected to each other via wired and / or wireless means.

[0012] Furthermore, the method for performing data latency estimation includes:

[0013] The multi-source medical device dataset is grouped based on the serial number of each medical device to obtain a subset of device data for each medical device.

[0014] Identify the upload path type corresponding to each device data subset, and calculate the time difference between the collection timestamp and the reception timestamp for each upload path in a single period to obtain a time difference sequence;

[0015] For each upload path, the time difference sequence of consecutive periods is traversed, and the mean time difference, the maximum time difference, and the variance of the time difference are calculated to construct the delay features of the corresponding upload path.

[0016] Calculate the first derivative of the delay feature of continuous periods and construct the derivative sequence. Statistically calculate the proportion of the number of signs of each component in the delay feature of each period in the derivative sequence to obtain the change distribution vector. Calculate the similarity between all change distribution vectors and filter the change distribution vectors with similarity higher than a preset similarity threshold to obtain a similar vector set.

[0017] The mean time difference of each period corresponding to the change distribution vector in the similar vector set is sorted, and the ratio of the difference between the median value and the mean value is calculated. If the ratio is less than the preset deviation threshold, the mean value is used as the estimation result; otherwise, the median value is used as the estimation result.

[0018] Furthermore, the method for performing delay compensation includes:

[0019] Based on the estimation results, the receiving timestamps of the corresponding upload paths in the multi-source medical device dataset are adjusted to obtain the compensated receiving timestamps;

[0020] The receiving time series of each device's data subset is constructed based on the compensated receiving timestamp; the time interval between adjacent compensated receiving timestamps is calculated to obtain the actual interval sequence;

[0021] A sliding window is constructed to traverse the actual interval sequence, the interval fluctuation value of each sliding window is calculated, and the sliding window with an interval fluctuation value higher than a preset fluctuation threshold is marked as an unstable window.

[0022] Linear interpolation is performed on the timestamps corresponding to the device data subset containing the unstable window to obtain the corrected data subset; all corrected data subsets are integrated to obtain the time-series corrected dataset.

[0023] Furthermore, the method of constructing the associated event chain includes:

[0024] Set the time window and spatial range for association recognition, and extract data records from the time-series corrected dataset that conform to the time window and spatial range for association recognition; identify the event labels and spatial identifiers corresponding to continuous data records;

[0025] Data records whose parameter changes are higher than the corresponding parameter thresholds for the selected event tags are used as candidate trigger record sets, and the candidate trigger record sets are divided into spatial groups based on spatial identifiers;

[0026] Extract continuous data records of spatial groups within the subsequent association identification time window to obtain a response record set; construct a path structure for each spatial group and its corresponding response record set based on the preset medical device type combination order and operation logic order, and add an event chain number to the path structure to construct an event chain identifier field.

[0027] Furthermore, the spatial identifier includes: medical device number and matching bed number and ward number.

[0028] Furthermore, the method for filtering out transmission signal anomalies includes:

[0029] Extract the signal strength value and corresponding medical device number of each data record in the associated labeled dataset. Sort the signal strength values ​​belonging to the same medical device number based on the path structure of the event chain to obtain the signal strength time series.

[0030] The difference between the maximum and minimum signal strength values ​​belonging to each period in the signal strength time series is calculated to obtain the period intensity range; based on the period intensity range, abnormal periods in the signal strength time series are identified, and abnormal signal fluctuation segments are screened out;

[0031] Extract the communication path field and device location field of the abnormal signal fluctuation segment and the adjacent period before and after it to form the communication path sequence and the location change sequence, respectively; identify the communication ID of the adjacent period in the communication path sequence, and mark it as a path change if there is a change; identify the location information of the adjacent period in the location change sequence, and mark it as a location change if there is a change.

[0032] Identify the signal strength direction of the abnormal signal fluctuation segment, and determine whether the signal strength direction is accompanied by a path change or a location change. If neither occurs, the abnormal signal fluctuation segment is determined to be a false record segment, and a fluctuation filtering label is added. At the same time, the device location field in the false record segment is kept to be the value of the previous cycle.

[0033] The associated labeled dataset after adding labels is used to obtain the fluctuation-removed dataset.

[0034] Furthermore, the method for identifying abnormal cycles includes:

[0035] Calculate the standard deviation of signal strength in the time series of signal strength, construct an anomaly scoring function based on the standard deviation of signal strength and the range of periodic strength, calculate the anomaly score using the anomaly scoring function, and determine the time segment with the anomaly score higher than the preset scoring threshold as the signal fluctuation anomaly segment.

[0036] Furthermore, the oversuppression correction method includes:

[0037] Extract the event type field, device number field, broadcast timestamp field, and spatial location field from each ward broadcast record;

[0038] The event type field, device number field, and spatial location field are concatenated into a broadcast vector, and the broadcast fingerprint value of the broadcast vector is calculated using a hash function;

[0039] Based on the broadcast timestamp field, all ward broadcast records are divided into consecutive broadcast batches, and ward broadcast records with the same broadcast fingerprint value in each broadcast batch are clustered into similar type broadcast clusters.

[0040] The ward broadcast records in each similar type broadcast cluster are grouped based on the spatial location field to obtain spatial group sub-clusters. If the number of spatial group sub-clusters is higher than the preset number threshold, the corresponding similar type broadcast cluster is determined to be a multi-source broadcast cluster.

[0041] Sort each spatial sub-cluster according to the broadcast timestamp field to obtain the broadcast subsequence; perform independent broadcast filtering on the broadcast subsequence and output independent trigger tags;

[0042] By using independent trigger tags to mark ward broadcast information, suppressed and corrected broadcast information can be obtained.

[0043] Furthermore, the method for performing independent broadcast filtering includes:

[0044] Construct a sliding broadcast window to traverse the broadcast subsequence, count the number of ward broadcast messages in each sliding broadcast window, and calculate the difference in the number of ward broadcast messages between adjacent sliding broadcast windows to obtain the broadcast density change.

[0045] If the change in broadcast density is higher than the preset density fluctuation threshold, and the direction of broadcast density change is the same in two or more consecutive sliding broadcast windows, then the segment formed by the consecutive sliding broadcast windows is determined to be a broadcast behavior mutation segment; the ward broadcast information in the broadcast behavior mutation segment is taken as an independent broadcast candidate record.

[0046] Ward broadcast information in independent broadcast candidate records where adjacent device number fields and spatial location fields are different is marked as an independent broadcast record, and an independent trigger field is constructed as an independent trigger label.

[0047] Furthermore, the method of performing device communication includes:

[0048] The system matches the transmitted data stream with the preset communication template, matches the corresponding communication protocol for different upload path types, and performs multi-channel device communication scheduling; the transmitted data stream is then distributed according to the corresponding communication sub-channels allocated by the schedule.

[0049] The technical effects and advantages of this invention, a unified communication and data integration platform for IoT devices in smart wards, are as follows:

[0050] A multi-source medical device dataset was obtained by collecting simulated operational data from medical equipment and performing data cleaning. Based on this dataset, data latency estimation and compensation, event chain construction, signal anomaly filtering, and broadcast oversuppression correction were performed, forming a unified communication and data integration platform for IoT devices in smart wards. Compared with existing experience, the pseudo-time sequence error of multi-source data caused by the difference in latency of heterogeneous communication paths was reduced by constructing a changing distribution vector for similarity screening and dynamically selecting estimation results. By constructing an event chain path structure based on spatial range and time window, the causal relationship of concurrent data streams was avoided from being destroyed. By calculating anomaly scores and judging path and location changes, misjudgment of signal fluctuations caused by multipath effects in the ward environment was identified. By performing cluster analysis and identifying abrupt broadcast behavior segments, independently suppressed broadcast records were recovered, ensuring that multi-source broadcast alarm information was not erroneously discarded. The accuracy and reliability of IoT device communication in smart wards were improved. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the unified communication and data integration platform for IoT devices in smart wards according to the present invention;

[0052] Figure 2 This is a schematic diagram of the unified communication and data integration method for IoT devices in smart wards according to the present invention. Detailed Implementation

[0053] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] Example 1

[0055] Please see Figure 1 As shown in this embodiment, the unified communication and data integration platform for IoT devices in smart wards includes:

[0056] The data acquisition module collects simulated operating data from medical equipment and performs data cleaning to obtain a multi-source medical equipment dataset.

[0057] The delay compensation module estimates data delay based on a multi-source medical device dataset, and then compensates for the delay based on the estimation results to obtain a time-corrected dataset.

[0058] The correlation correction module constructs a correlation event chain for the time-series correction dataset, outputs the event chain identifier field, and labels the time-series correction dataset to obtain the correlation labeled dataset.

[0059] The anomaly filtering module performs transmission signal anomaly filtering on the associated labeled dataset and outputs a fluctuation-free dataset.

[0060] The suppression adjustment module collects ward broadcast information, performs oversuppression correction on the ward broadcast information, and obtains suppressed and corrected broadcast information;

[0061] The unified transmission module integrates the fluctuation cancellation dataset and suppression correction broadcast information into a transmission data stream, and uses this transmission data stream for device communication; the modules are connected to each other via wired and / or wireless means.

[0062] In this embodiment, the medical equipment simulation operation data collected is based on medical scenario simulation, including relevant parameters of IoT medical devices used in wards, such as blood pressure monitors, injection control devices, and ventilators. By filling missing values ​​and standardizing the data, the data is cleaned to obtain a higher quality multi-source medical equipment dataset.

[0063] Methods for estimating data latency include:

[0064] The multi-source medical device dataset is grouped based on the device ID of each medical device to obtain a subset of device data for each medical device. The device ID is an identifier associated with each medical device. Based on the device ID, the relevant data belonging to each medical device in the multi-source medical device dataset is identified, which is the device data subset.

[0065] Identify the upload path type corresponding to each device data subset, and calculate the time difference between the collection timestamp and the reception timestamp for each upload path within a single period to obtain a time difference sequence.

[0066] The upload path type refers to the communication data transmission path of each medical device, which is determined based on the upload gateway number and communication protocol type field, such as Zigbee, WiFi, and BLE, which are commonly used communication paths for IoT communication. In the upload path corresponding to each medical device, the time difference between the acquisition timestamp and the reception timestamp within a single cycle is calculated to obtain the delay, and a time difference sequence is constructed based on the time difference value. The specific time length of the cycle is set based on historical cycle settings to ensure that it can conform to the operating conditions of each device in the medical scenario simulation.

[0067] For each upload path, the time difference sequence of consecutive periods is traversed, and the mean, maximum difference, and variance of the time difference are calculated to construct the latency features of the corresponding upload path.

[0068] The method involves iterating through the time difference sequence of each upload path in a continuous period, calculating the mean time difference, the maximum time difference, and the variance of the time difference for each period, and combining them into a feature vector, which is the delay feature used to reflect the transmission behavior pattern; the continuous period covers the entire time length of the medical scenario simulation.

[0069] Calculate the first derivative of the delay feature of continuous periods and construct the derivative sequence. Statistically calculate the proportion of the number of signs of each component in the delay feature of each period in the derivative sequence to obtain the change distribution vector.

[0070] The first derivative values ​​of the mean time difference, the maximum time difference, and the variance of the time difference in the delay features of continuous periods are obtained, and these first derivative values ​​are sorted according to the order of the delay features to obtain the derivative sequence. The number of signs of the first derivative value of each component is counted, including positive, negative, and 0, and the number of each type of sign corresponding to the delay feature of each period is combined into a change distribution vector, which reflects the change of transmission state.

[0071] Calculate the similarity between all change distribution vectors, and filter the change distribution vectors with similarity higher than a preset similarity threshold to obtain a set of similar vectors. In this embodiment, calculate the cosine similarity between all change distribution vectors, set a similarity threshold based on historical screening experience, and combine the change distribution vectors with cosine similarity higher than the similarity threshold into a set of similar vectors.

[0072] The mean time difference of each period corresponding to the change distribution vector in the similar vector set is sorted, and the ratio of the difference between the median value and the mean value is calculated. The ratio of the difference between the mean value and the median value is calculated by dividing the absolute value of the difference between the mean value and the median value by the mean value.

[0073] If the proportion is less than the preset deviation threshold, the average value is used as the estimation result; otherwise, the median is used as the estimation result. The deviation threshold is set based on historical estimation experience. If the difference proportion is less than the deviation threshold, it indicates that the data distribution is symmetrical and the change is normal. In this case, the average value can be used as the estimation result of the delayed estimation. Conversely, it indicates that the degree of change is high and the mean is easily affected by extreme values. Therefore, the median is selected as the estimation result of the delayed estimation.

[0074] Methods for delay compensation include:

[0075] Based on the estimation results, the receiving timestamps of the corresponding upload paths in the multi-source medical device dataset are adjusted to obtain the compensated receiving timestamps. The difference between the original receiving timestamp field and the estimation result of the upload path is calculated to obtain a logically reasonable receiving time, which is then used as the compensated receiving timestamp.

[0076] The receiving time series of each device's data subset is constructed based on the compensated receiving timestamps, where all compensated receiving timestamps are sorted in chronological order to obtain the receiving time series.

[0077] The time interval between adjacent compensated receiving timestamps is calculated to obtain the actual interval sequence. For the receiving time sequence, the time interval between two adjacent records is calculated to obtain the actual receiving time interval, which reflects the real signal receiving rhythm. All time intervals are sorted to obtain the actual interval sequence as the data basis for subsequent operations.

[0078] A sliding window is constructed to traverse the actual interval sequence, and the interval fluctuation value of each sliding window is calculated. Sliding windows with interval fluctuation values ​​higher than a preset fluctuation threshold are marked as unstable windows. The interval fluctuation value refers to the difference between the maximum and minimum time intervals within the corresponding sliding window. A fluctuation threshold is set based on historical compensation experience, and the time intervals corresponding to sliding windows with interval fluctuation values ​​higher than the preset fluctuation threshold are designated as unstable windows.

[0079] Linear interpolation is performed on the timestamps corresponding to the device data subset containing unstable windows to obtain the corrected data subset; all corrected data subsets are integrated to obtain the time-series correction dataset, wherein linear interpolation is performed on the timestamps within the corresponding time interval of the device data subset containing unstable windows. In this embodiment, the linear interpolation algorithm is used to perform linear interpolation to calibrate the timestamps within the time interval that needs to be compensated.

[0080] Methods for constructing a chain of related events include:

[0081] Set the association identification time window range and spatial range, and extract data records in the time-series correction dataset that conform to the association identification time window range and spatial range. The association identification time window range and spatial range are set based on historical experience to ensure that the time range can cover the duration of each operating cycle of all medical devices, and the spatial range can cover the range of all wards in the simulated medical scenario.

[0082] Identify event tags and spatial identifiers corresponding to continuous data records. Event tags refer to the tasks performed by medical devices corresponding to parameter changes in continuous data records, such as heart rate measurement, infusion volume control, or body position change sensing. Set event tags for corresponding medical devices based on simulated medical scenarios.

[0083] Data records whose parameter changes are higher than the corresponding parameter thresholds for the selected event tags are used as candidate trigger record sets, and the candidate trigger record sets are divided into spatial groups based on spatial identifiers.

[0084] The process involves setting parameter thresholds for corresponding dimension parameters based on parameter changes for event tags in historical records. The change in parameters corresponding to the current event tag is then statistically analyzed. If the change exceeds the parameter threshold for the corresponding dimension, it indicates that the task corresponding to the event tag has been effectively triggered. Data records are then used as candidate trigger records and combined into a candidate trigger record set. Simultaneously, candidate trigger records belonging to the same spatial range are integrated into spatial groups according to spatial identifiers to ensure that the spatial range corresponding to subsequent operations is not confused.

[0085] Extract continuous data records of spatial groups within the subsequent association identification time window range to obtain the response record set. Based on the association identification time window range corresponding to each spatial group, continue to extract all continuous data records within the association identification time window range, and use these data records as the response record set of medical devices in the current spatial group.

[0086] For each spatial group and its corresponding set of response records, a path structure is constructed based on the preset combination order of medical device types and the order of operation logic. An event chain number is added to this path structure to construct an event chain identifier field.

[0087] The process integrates each spatial group and its corresponding response record set to construct an ordered event path structure. In this embodiment, based on medical device specification knowledge and known device operation knowledge, the type combination order and operation logic order of medical devices are obtained. For example, the type combination order is that the mattress sensor, body position sensor, and heart rate monitoring terminal are combined and data is transmitted in sequence according to medical specifications. The operation logic order is that when a patient gets up, pressure changes may occur first, then body position changes, and then heart rate fluctuations occur. A corresponding event chain number is added to each path structure, and this number is associated with the data records of each medical device in the event chain to form an event chain identifier field.

[0088] The spatial identifier includes: medical device number and matching bed number and ward number. Since this embodiment is used for smart wards, the spatial identifier is set as the medical device and bed information carried in each ward to indicate the deployment location of the device and its logical space.

[0089] Methods for filtering out transmission signal anomalies include:

[0090] Extract the signal strength value and corresponding medical device number of each data record in the associated labeled dataset. Sort the signal strength values ​​belonging to the same medical device number based on the path structure of the event chain to obtain the signal strength time sequence.

[0091] The signal strength value refers to the value of the Received Signal Strength Indicator (RSSI) field. Data records uploaded by the same medical device within a certain period are filtered out, and the signal strength values ​​are sorted according to the order of the device in the event chain path structure to obtain a signal strength time sequence, which is used to reflect the signal fluctuation of the corresponding medical device in the event chain path.

[0092] The difference between the maximum and minimum signal strength values ​​in each period of the signal strength time series is calculated to obtain the periodic intensity range, where the period refers to the operating cycle of the medical equipment, which has the same meaning as the period in the data delay estimation part; the difference between the extreme values ​​in the period is calculated to obtain the periodic intensity range, which is used to reflect the signal fluctuation amplitude within the period.

[0093] Based on the period intensity range, abnormal periodicity is identified in the signal intensity time series, and abnormal signal fluctuation segments are screened.

[0094] The communication path field and device location field of the abnormal signal fluctuation segment and the adjacent periods before and after it are extracted to form the communication path sequence and the location change sequence, respectively. Specifically, the communication path sequence and the location change sequence are obtained by extracting the communication path field and the device location field of the time segment where the abnormal signal fluctuation segment is located and the adjacent periods before and after it, and sorting them in chronological order. The communication path field includes information such as the access gateway number and the communication protocol type, and the device location field includes information such as the bed number and the ward number.

[0095] Identify the communication IDs of adjacent cycles in the communication path sequence, and mark any changes as path changes; identify the location information of adjacent cycles in the location change sequence, and mark any changes as location changes.

[0096] In this embodiment, the communication ID refers to the access gateway number. The access gateway number field of adjacent periods is compared. If the numbers are different (for example, GATEWAY A becomes GATEWAY B), it is marked as a path change behavior. If the bed number or ward number of the data record in adjacent periods changes, it indicates that the device has undergone physical displacement in the simulated scene space, and it is marked as a location change. The two change situations are used as the judgment conditions.

[0097] Identify the signal strength direction of abnormal signal fluctuation segments, and determine whether the signal strength direction is accompanied by path changes or location changes. If neither occurs, the abnormal signal fluctuation segment is determined to be a false record segment, a fluctuation filtering label is added, and the device location field in the false record segment is kept to the value of the previous cycle.

[0098] The signal strength direction is determined by calculating the first derivative of the RSSI field. If the signal strength direction remains unidirectional and the communication path field and location field do not change, it indicates that the signal fluctuation does not originate from the device's movement, but may be caused by interference from factors such as metal reflection in the environment. A fluctuation filtering label is added to this data segment for identification during communication. At the same time, in order to maintain the continuity of the device positioning information, the device location field in the original record is filled back with the position of the previous cycle.

[0099] The associated labeled dataset after adding labels is used to obtain the fluctuation-removed dataset.

[0100] Methods for identifying abnormal cycles include:

[0101] Calculate the standard deviation of signal strength in the time series of signal strength, construct an anomaly scoring function based on the standard deviation of signal strength and the range of periodic strength, calculate the anomaly score using the anomaly scoring function, and determine the time segment with the anomaly score higher than the preset scoring threshold as the signal fluctuation anomaly segment.

[0102] The formula for calculating the anomaly scoring function is as follows: ;in, Indicates abnormal rating; Indicates the standard deviation of signal strength; This indicates a very poor periodic intensity. and These are the weights of the signal strength standard deviation and the periodic strength range, respectively, set based on historical function design experience. It should be noted that the signal strength standard deviation and periodic strength range in the anomaly scoring function have been normalized to the same order of magnitude and to eliminate dimensional differences. At the same time, a scoring threshold is set based on historical function design experience. If the anomaly score is higher than the preset scoring threshold, it indicates that the signal jitter measured by the combined standard deviation and range is relatively high. Therefore, the corresponding data segment is judged as a signal fluctuation anomaly segment.

[0103] Methods of suppression correction include:

[0104] Extract the event type field, device number field, broadcast timestamp field, and spatial location field from each ward broadcast record.

[0105] The event type field indicates the type of task event the broadcast belongs to, such as an emergency call, reset, or respiratory rate alarm; the broadcast timestamp field indicates the timestamp when the ward broadcast record was received after it was sent; and the spatial location field indicates the location of the broadcast target of the ward broadcast record.

[0106] The event type field, device number field, and spatial location field are concatenated into a broadcast vector, and a hash function is used to calculate the broadcast fingerprint value of the broadcast vector.

[0107] In this embodiment, the fields of event type, target triggering device and physical space from which the ward broadcast record originates are converted into a unified format broadcast vector. The broadcast fingerprint value of the corresponding ward broadcast record is generated by using a hash function, which serves as the data basis for subsequent judgment.

[0108] Based on the broadcast timestamp field, all ward broadcast records are divided into consecutive broadcast batches, and ward broadcast records with the same broadcast fingerprint value in each broadcast batch are clustered into similar type broadcast clusters.

[0109] Based on the broadcast cycle in the simulated medical scenario, the ward broadcast information corresponding to the broadcast timestamp field within each broadcast cycle is divided into a broadcast batch, resulting in several consecutive broadcast batches. In each broadcast batch, ward broadcast records with the same broadcast fingerprint value are extracted to form a similar type of broadcast cluster, indicating that these broadcast records are easily considered as duplicate broadcast content.

[0110] The ward broadcast records in each similar type broadcast cluster are grouped based on the spatial location field to obtain spatial group sub-clusters. If the number of spatial group sub-clusters is higher than the preset number threshold, the corresponding similar type broadcast cluster is determined to be a multi-source broadcast cluster.

[0111] Within each similar type of broadcast cluster, the broadcasts are grouped by the spatial location field in the ward broadcast records. If broadcast records from different beds or different wards appear, it means that although these broadcasts have the same content, they should be triggered independently from different physical areas. The preset quantity threshold is used to indicate how many spatial sources are allowed to be regarded as the same event within a cluster. It is set based on historical broadcast suppression experience. If the threshold is exceeded, the broadcast cluster is considered to be a broadcast cluster with multiple sources uploading at the same time.

[0112] Each spatial sub-cluster is sorted according to the broadcast timestamp field to obtain a broadcast subsequence. The ward broadcast information recorded in the same spatial sub-cluster is arranged in order of broadcast timestamp to form a broadcast subsequence that reflects the evolution of broadcast behavior.

[0113] Perform independent broadcast filtering on broadcast subsequences and output independent trigger tags;

[0114] By using independent trigger tags to mark ward broadcast information, suppressed and corrected broadcast information is obtained. For ward broadcast information with independent trigger tags, the system's communication terminal restores the information to a forwardable state. This tag allows the system to identify that the ward broadcast information is meaningful, avoiding subsequent suppression and ensuring that the independent alarms of each bed or ward are not incorrectly marked as redundant information or noise.

[0115] Methods for conducting independent broadcast filtering include:

[0116] A sliding broadcast window is constructed to traverse the broadcast subsequence, the number of ward broadcast messages in each sliding broadcast window is counted, and the difference in the number of ward broadcast messages between adjacent sliding broadcast windows is calculated to obtain the change in broadcast density.

[0117] The specific duration of the sliding broadcast window is set based on historical independent broadcast screening experience. The change in broadcast density is obtained by calculating the difference in the number of ward broadcast messages in adjacent sliding broadcast windows, which reflects the change in broadcast activity within the time period of adjacent windows.

[0118] If the change in broadcast density is higher than the preset density fluctuation threshold, and the direction of broadcast density change is the same in two or more consecutive sliding broadcast windows, then the segment formed by the consecutive sliding broadcast windows is determined to be a broadcast behavior mutation segment; the ward broadcast information in the broadcast behavior mutation segment is taken as an independent broadcast candidate record.

[0119] The direction of broadcast density change refers to whether the broadcast density change is increasing or decreasing within a continuous window. Based on historical independent broadcast screening experience, a density fluctuation threshold is set. If the broadcast density change is higher than the preset density fluctuation threshold and the broadcast density change direction is the same in the continuous windows, the segment formed by the corresponding continuous sliding broadcast windows is marked as a broadcast behavior mutation segment, and the ward broadcast information in it is used as an independent broadcast candidate record for subsequent processing.

[0120] In the independent broadcast candidate records, ward broadcast information with different adjacent device number fields and spatial location fields is marked as an independent broadcast record, and an independent trigger field is constructed as an independent trigger label. In the independent broadcast candidate records, adjacent continuous comparison is performed. If the device number field and spatial location field of adjacent ward broadcast information are not repeated, the corresponding medical device source is considered independent, the ward broadcast information is marked as an independent broadcast record, and an independent trigger label is constructed using encoding to identify real multi-source emergency alarms and prevent communication terminals from suppressing or merging them.

[0121] Methods of device communication include:

[0122] The system matches the transmitted data stream with the preset communication template, matches the corresponding communication protocol for different upload path types, and performs multi-channel device communication scheduling; the transmitted data stream is then distributed according to the corresponding communication sub-channels allocated by the schedule.

[0123] The communication template is a communication structure defined based on communication specifications for different device communication scenarios. It includes information such as upload path type, communication protocol specifications, data transmission priority, and service type identifier. In the transmitted data stream, each record carries upload path type field, event chain identifier field, and data tag field (such as fluctuation filtering tag and independent trigger tag). By calling template matching, the fields are compared and logically mapped, thereby matching the data record with the appropriate communication protocol and allocating it to the corresponding communication service channel to realize multi-channel device communication scheduling, such as ordinary device synchronization channel, independent event transmission channel, and emergency broadcast priority distribution channel. Finally, each data is encapsulated, packaged, scheduled, and distributed to the medical device side according to the target communication sub-channel, realizing device communication behavior that includes information such as spatial ownership and event status.

[0124] This embodiment obtains a multi-source medical device dataset by collecting simulated operating data from medical equipment and performing data cleaning. Based on this dataset, data latency estimation and compensation, event chain construction, signal anomaly filtering, and broadcast oversuppression correction are performed to form a unified communication and data integration platform for IoT devices in smart wards. Compared with existing experience, by constructing a changing distribution vector for similarity screening and dynamically selecting estimation results, pseudo-time-series errors in multi-source data caused by latency differences in heterogeneous communication paths are reduced. By constructing an event chain path structure based on spatial range and time window, the causal relationship of concurrent data streams is avoided from being disrupted. By calculating anomaly scores and judging path and location changes, misjudgments of signal fluctuations caused by multipath effects in the ward environment are identified. By performing cluster analysis and identifying abrupt broadcast behavior segments, independently suppressed broadcast records are recovered, ensuring that multi-source broadcast alarm information is not erroneously discarded. This improves the accuracy and reliability of IoT device communication in smart wards.

[0125] Example 2

[0126] Please see Figure 2 As shown, parts not described in detail in this embodiment are described in Embodiment 1. A unified communication and data integration method for IoT devices in smart wards is provided, including:

[0127] S1. Collect simulated operation data of medical equipment and perform data cleaning to obtain a multi-source medical equipment dataset;

[0128] S2. Based on the multi-source medical device dataset, perform data latency estimation, and based on the estimation results, perform latency compensation on the multi-source medical device dataset to obtain the time-series corrected dataset;

[0129] S3. Construct a related event chain for the time-series correction dataset, output the event chain identifier field, and label the time-series correction dataset to obtain the related labeled dataset;

[0130] S4. Perform transmission signal anomaly filtering on the associated labeled dataset and output a fluctuation-eliminated dataset;

[0131] S5. Collect ward broadcast information, perform oversuppression correction on the ward broadcast information, and obtain suppressed and corrected broadcast information;

[0132] S6. Integrate the fluctuation elimination dataset and suppression correction broadcast information into a transmission data stream, and use this transmission data stream for device communication.

[0133] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. 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.

[0134] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0135] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A unified communication and data integration platform for IoT devices in smart wards, characterized in that, include: The data acquisition module collects simulated operating data from medical equipment and performs data cleaning to obtain a multi-source medical equipment dataset. The delay compensation module estimates data delay based on a multi-source medical device dataset, and then compensates for the delay based on the estimation results to obtain a time-corrected dataset. The correlation correction module constructs a correlation event chain for the time-series correction dataset, outputs the event chain identifier field, and labels the time-series correction dataset to obtain the correlation labeled dataset. The anomaly filtering module performs transmission signal anomaly filtering on the associated labeled dataset, including: Extract the signal strength value and corresponding medical device number of each data record in the associated labeled dataset. Sort the signal strength values ​​belonging to the same medical device number according to the path structure of the event chain to obtain the signal strength time series. The difference between the maximum and minimum signal strength values ​​belonging to each period in the signal strength time series is calculated to obtain the period intensity range; based on the period intensity range, abnormal periods in the signal strength time series are identified, and abnormal signal fluctuation segments are screened out; Extract the communication path field and device location field of the abnormal signal fluctuation segment and the adjacent period before and after it to form the communication path sequence and the location change sequence, respectively; identify the communication ID of the adjacent period in the communication path sequence, and mark it as a path change if there is a change; identify the location information of the adjacent period in the location change sequence, and mark it as a location change if there is a change. Identify the signal strength direction of the abnormal signal fluctuation segment, and determine whether the signal strength direction is accompanied by a path change or a location change. If neither occurs, the abnormal signal fluctuation segment is determined to be a false record segment, and a fluctuation filtering label is added. At the same time, the device location field in the false record segment is kept to be the value of the previous cycle. The variability-reduced dataset is obtained by adding labels to the associated labeled dataset. The methods for identifying abnormal cycles include: Calculate the standard deviation of signal strength in the time series of signal strength, construct an anomaly scoring function based on the standard deviation of signal strength and the periodic intensity range, calculate the anomaly score using the anomaly scoring function, and determine the time segment with the anomaly score higher than the preset scoring threshold as the signal fluctuation anomaly segment. The suppression adjustment module collects ward broadcast information, performs oversuppression correction on the ward broadcast information, and obtains suppressed and corrected broadcast information; The unified transmission module integrates the fluctuation cancellation dataset and suppression correction broadcast information into a transmission data stream, and uses this transmission data stream for device communication; the modules are connected to each other via wired and / or wireless means.

2. The unified communication and data integration platform for IoT devices in smart wards according to claim 1, characterized in that, The methods for performing data latency estimation include: The multi-source medical device dataset is grouped based on the serial number of each medical device to obtain a subset of device data for each medical device. Identify the upload path type corresponding to each device data subset, and calculate the time difference between the collection timestamp and the reception timestamp for each upload path in a single period to obtain a time difference sequence; For each upload path, the time difference sequence of consecutive periods is traversed, and the mean time difference, the maximum time difference, and the variance of the time difference are calculated to construct the delay features of the corresponding upload path. Calculate the first derivative of the delay feature of continuous periods and construct the derivative sequence. Statistically calculate the proportion of the number of signs of each component in the delay feature of each period in the derivative sequence to obtain the change distribution vector. Calculate the similarity between all change distribution vectors and filter the change distribution vectors with similarity higher than a preset similarity threshold to obtain a similar vector set. The mean time difference of each period corresponding to the change distribution vector in the similar vector set is sorted, and the ratio of the difference between the median value and the mean value is calculated. If the ratio is less than the preset deviation threshold, the mean value is used as the estimation result; otherwise, the median value is used as the estimation result.

3. The unified communication and data integration platform for IoT devices in smart wards according to claim 2, characterized in that, The methods for performing delay compensation include: Based on the estimation results, the receiving timestamps of the corresponding upload paths in the multi-source medical device dataset are adjusted to obtain the compensated receiving timestamps; The receiving time series of each device's data subset is constructed based on the compensated receiving timestamp; the time interval between adjacent compensated receiving timestamps is calculated to obtain the actual interval sequence; A sliding window is constructed to traverse the actual interval sequence, the interval fluctuation value of each sliding window is calculated, and the sliding window with an interval fluctuation value higher than a preset fluctuation threshold is marked as an unstable window. Linear interpolation is performed on the timestamps corresponding to the device data subset containing the unstable window to obtain the corrected data subset; all corrected data subsets are integrated to obtain the time-series corrected dataset.

4. The unified communication and data integration platform for IoT devices in smart wards according to claim 3, characterized in that, The methods for constructing the chain of related events include: Set the time window and spatial range for association recognition, extract data records from the time-series corrected dataset that conform to the time window and spatial range for association recognition, and identify the event labels and spatial identifiers corresponding to continuous data records; Data records whose parameter changes are higher than the corresponding parameter thresholds for the selected event tags are used as candidate trigger record sets, and the candidate trigger record sets are divided into spatial groups based on spatial identifiers; Extract continuous data records of spatial groups within the subsequent association identification time window to obtain a response record set; construct a path structure for each spatial group and its corresponding response record set based on the preset medical device type combination order and operation logic order, and add an event chain number to the path structure to construct an event chain identifier field.

5. The unified communication and data integration platform for IoT devices in smart wards according to claim 4, characterized in that, The spatial identifier includes: the medical device number and the matching bed number and ward number.

6. The unified communication and data integration platform for IoT devices in smart wards according to claim 5, characterized in that, The methods for performing oversuppression correction include: Extract the event type field, device number field, broadcast timestamp field, and spatial location field from each ward broadcast record; The event type field, device number field, and spatial location field are concatenated into a broadcast vector, and the broadcast fingerprint value of the broadcast vector is calculated using a hash function; Based on the broadcast timestamp field, all ward broadcast records are divided into consecutive broadcast batches, and ward broadcast records with the same broadcast fingerprint value in each broadcast batch are clustered into similar type broadcast clusters. The ward broadcast records in each similar type broadcast cluster are grouped based on the spatial location field to obtain spatial group sub-clusters. If the number of spatial group sub-clusters is higher than the preset number threshold, the corresponding similar type broadcast cluster is determined to be a multi-source broadcast cluster. Sort each spatial sub-cluster according to the broadcast timestamp field to obtain the broadcast subsequence; perform independent broadcast filtering on the broadcast subsequence and output independent trigger tags; By using independent trigger tags to mark ward broadcast information, suppressed and corrected broadcast information can be obtained.

7. The unified communication and data integration platform for IoT devices in smart wards according to claim 6, characterized in that, The methods for performing independent broadcast filtering include: Construct a sliding broadcast window to traverse the broadcast subsequence, count the number of ward broadcast messages in each sliding broadcast window, and calculate the difference in the number of ward broadcast messages between adjacent sliding broadcast windows to obtain the broadcast density change. If the change in broadcast density is higher than the preset density fluctuation threshold, and the direction of broadcast density change is the same in two or more consecutive sliding broadcast windows, then the segment formed by the consecutive sliding broadcast windows is determined to be a broadcast behavior mutation segment; the ward broadcast information in the broadcast behavior mutation segment is taken as an independent broadcast candidate record. Ward broadcast information in independent broadcast candidate records where adjacent device number fields and spatial location fields are different is marked as an independent broadcast record, and an independent trigger field is constructed as an independent trigger label.

8. The unified communication and data integration platform for IoT devices in smart wards according to claim 7, characterized in that, The methods for performing device communication include: The system matches the transmitted data stream with the preset communication template, matches the corresponding communication protocol for different upload path types, and performs multi-channel device communication scheduling; the transmitted data stream is then distributed according to the corresponding communication sub-channels allocated by the schedule.

Images