Intelligent infusion anomaly recognition and early warning method based on multi-source perception fusion

Abstract

This invention discloses an intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion, belonging to the field of intelligent monitoring technology. The method includes the following steps: continuously collecting data on drip rate, pressure, and flow rate changes during the infusion process; identifying abrupt changes in each signal based on preset change judgment rules; and marking these abrupt changes according to a unified time axis to form a time-series change record corresponding to the multi-source signals. This invention organizes drip rate, pressure, and flow rate changes along a unified time axis, identifies multi-source synchronous abrupt changes, and constructs a continuous change record, enabling early detection of anomaly initiation points. Simultaneously, by forming a change correlation record throughout the entire process and continuously matching it in subsequent data, anomaly identification shifts from post-event judgment to process identification, thereby outputting early warnings at the initial stage and improving the continuity and safety of the infusion process.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology, specifically to an intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion. Background Technology

[0002] Intelligent infusion anomaly identification and early warning based on multi-source sensing fusion refers to the process of simultaneously collecting information from multiple different sensing sources during infusion, such as pressure changes in the infusion tubing, fluid flow rate, drip rate rhythm, equipment operating status, and possible environmental or human feedback signals. This independent data is then integrated and correlated to form a holistic understanding of the current infusion status. Based on this, the relationship between various data changes over time is continuously compared to identify abnormal characteristics inconsistent with the normal infusion process, such as abnormal flow rate fluctuations, tubing blockage trends, and impending fluid depletion. Furthermore, based on the changing trends of the abnormalities, early warning information is issued, allowing nursing staff to intervene before the problem fully manifests, thereby improving the safety and continuity of the infusion process.

[0003] The existing technology has the following shortcomings: In existing technologies, multi-source sensing data processing for infusion procedures often employs threshold judgment and short-time filtering. When multiple sensing signals synchronously change and quickly return to normal within a very short period, the system treats this change as transient noise and suppresses it. However, because these synchronous changes are short-lived and sudden in amplitude, they are easily misjudged as acquisition disturbances and directly filtered out, thus ignoring the true origin of the anomaly. Furthermore, as time progresses, this initial anomaly may gradually evolve and expand. Because the system failed to effectively identify it in the early stages, it may miss the optimal intervention opportunity, leading to delayed anomaly identification and increasing operational risks during the infusion process.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion, so as to solve the problems in the background art mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion, comprising the following steps: The system continuously collects data on changes in drip rate, pressure, and flow rate during infusion. Based on preset change judgment rules, it identifies abrupt changes in each signal and marks them according to a unified time axis to form a time sequence change record corresponding to multiple source signals. Based on the time sequence change record, the corresponding analysis of the mutation points in different signals is carried out within a preset time range to determine the multi-source synchronous mutation segments that appear in the same time range, and the corresponding synchronous change record is formed in the time sequence change record. Around the synchronous change record, the preset range is extended forward and backward along the time sequence change record, and the change data within the extended range is sequentially correlated to construct a continuous change record that includes the state before the mutation and the change process after the mutation. Based on continuous change records, the change process after the mutation is continuously tracked, and the tracking results are correlated with the state before the mutation to form a change correlation record that runs through the mutation initiation to the change regression process; Based on the change association records, a matching analysis is performed on the subsequent updated time sequence change records. When the corresponding synchronous mutation fragment reappears, the anomaly identification is completed based on the change association records, and an early warning information is output.

[0007] Preferably, the drip rate, pressure, and flow rate changes are continuously recorded and expressed on a unified time axis, constructing a time-series change record that includes markers of abrupt change points. The steps are as follows: The entire infusion process is divided into continuous time periods. The changes in drip rate, pressure and flow rate within each time period are collected and recorded in chronological order. Each time point is assigned a unique time identifier, forming a multi-source change sequence indexed by the time identifier. By combining multi-source change sequences, the values ​​at adjacent time points are compared point by point, and the time points that meet the change judgment criteria are identified as mutation points. The mutation points are then marked at the corresponding time marker positions to obtain the set of mutation points in each change sequence. The set of mutation points is compiled and mapped to a unified timeline, and the mutation situation at each time marker position is centrally recorded to form a unified time marker set; By integrating multi-source variation sequences with a unified set of time markers, the numerical values ​​of drip rate changes, pressure changes, and flow rate changes, as well as abrupt change markers, are recorded at each time marker location and arranged in chronological order to obtain a time-series variation record.

[0008] Preferably, the mutation point identification process uses the same change judgment criteria, compares the changes in drip rate, pressure and flow rate point by point, and maintains a one-to-one correspondence between mutation points and time markers. At the same time, the positions of each time marker in the unified time marker set record the occurrence of multi-source mutations, thereby ensuring that the multi-source changes and mutation markers in the time sequence change record correspond synchronously.

[0009] Preferably, multiple mutation points are simultaneously identified and recorded as synchronous changes within a unified timeframe, as follows: The system sequentially scans the time sequence change records and constructs a preset time range containing consecutive time markers before and after each time marker position. It also extracts the corresponding drop rate change abrupt change points, pressure change abrupt change points, and flow rate change abrupt change points to form a time segment set. A step-by-step comparison is performed on the time segment set to determine whether the same preset time range contains abrupt changes in drip rate, pressure, and flow rate. Multi-source synchronous abrupt change candidate segments are obtained and the time marker position of each abrupt change point is recorded. The candidate fragments of multi-source synchronous mutation are collected and their time-stamped ranges are organized. Candidate fragments with overlapping or closely connected relationships are integrated into continuous time intervals to form multi-source synchronous mutation fragments while preserving the original arrangement order of mutation points. By combining multi-source synchronous mutation fragments for labeling, markers are set at the start time and end time markers, and synchronous change markers are uniformly marked within the corresponding time intervals to form a synchronous change record.

[0010] Preferably, the time marker positions within the corresponding time interval of the multi-source synchronous mutation fragment are associated with the distribution order of the drop rate change mutation point, pressure change mutation point, and flow rate change mutation point, and the synchronous change marker is associated with the corresponding time range information and the arrangement relationship of the mutation point time marker positions.

[0011] Preferably, multi-source change data are extended and correlated within the range before and after the synchronous change record to form a continuous change record, and the steps are as follows: Extract the start and end time markers corresponding to the synchronous change records, extend the continuous time markers forward and backward along the time sequence change records respectively, combine them to form an extended time interval, and obtain the drip rate change value, pressure change value, and flow rate change value. Organize the change data corresponding to the extended time interval, combine the drop rate change value, pressure change value and flow rate change value corresponding to each time mark position and arrange them in chronological order to construct a continuous change sequence containing the time interval of synchronous change records; Connect adjacent time marker positions in a continuous change sequence, sequentially connect the change data of each time marker position to form a continuous change chain that runs through the preceding and following time positions, and locate the corresponding time interval of the synchronous change record; By integrating the continuous change chain and marking the start and end time markers of the synchronous change records, a continuous change record is formed that includes the state before the mutation and the change process after the mutation.

[0012] Preferably, the time interval corresponding to the synchronous change record is located in the middle of the continuous change chain, and the time marker position before the synchronous change record corresponds to the state change data before the mutation, and the time marker position after the synchronous change record corresponds to the change process data after the mutation, forming a continuous change structure that runs through the time sequence.

[0013] Preferably, the multi-source change processes are correlated chronologically to form a change correlation record, including the following steps: Extract the end time marker position corresponding to the synchronous change record in the continuous change record, read the continuous time marker position along the time sequence change record and obtain the drip rate change value, pressure change value and flow rate change value, and construct the change sequence after the sudden change. Organize the time marker positions in the change sequence after the mutation, combine and record the corresponding drop rate change value, pressure change value and flow rate change value, and arrange them in time order to form a change expression with continuous time markers. Extract the consecutive time markers before the corresponding start time marker position of the synchronous change record and obtain the corresponding change data. Sequentially associate the state range before the mutation with the change sequence after the mutation and establish a continuous connection relationship. By integrating the pre-mutation state range, the time interval of synchronous change records, and the post-mutation change sequence, and arranging them in chronological order, and marking the start and end time markers of the synchronous change records, a change association record is formed that runs through the process from the onset of the mutation to the decline of the change.

[0014] Preferably, the time marker positions in the post-mutation change sequence are arranged in a continuous order, and the corresponding values ​​of drip rate change, pressure change, and flow rate change for each time marker position are kept consistent and connected with the corresponding time marker positions in the pre-mutation state range, thereby ensuring that there is a continuous transmission relationship between the time marker positions in the change association record.

[0015] Preferably, the change-related records are used to match time-series change records and complete anomaly identification, with the following steps: Extract the time marker positions from the change correlation records and arrange them in chronological order to form a reference change sequence. At the same time, mark the starting position of the mutation and the time interval of the synchronous change record, and record the drop rate change value, pressure change value and flow rate change value. Add the change data corresponding to the newly added time marker position in the time sequence change record, extract the continuous time marker positions forward along the time sequence change record and construct the time segment to be matched, and maintain the time structure consistent with the reference change sequence. By identifying the time marker positions of the time segment to be matched and the reference change sequence, the continuous time interval that is consistent with the time interval of the synchronous change record is identified and determined as the corresponding synchronous mutation segment; The system marks the start time of the corresponding synchronous mutation fragment, records the change data, generates early warning information, and continuously records the change process along the time sequence.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention organizes drip rate, pressure, and flow rate changes along a unified timeline and identifies multi-source synchronous mutation fragments in the time-series change records. This effectively preserves short-term synchronous changes that are easily overlooked, preventing mutation information from being directly suppressed and thus forming a complete record at the initial stage of the change. Based on this, by constructing continuous change records, the pre-mutation state and the post-mutation change process are sequentially connected, enabling the change process to have continuous expression capabilities. This allows for early capture of abnormal starting points and improves the ability to identify abnormal stages in the infusion process.

[0017] This invention constructs a change correlation record that spans the entire process from the onset of a mutation to its regression, and continuously matches it with subsequent time-series change records. This allows historical change processes to participate in the judgment of current changes. When a corresponding synchronous mutation fragment reappears, anomaly identification can be completed and a warning message can be output, thus transforming anomaly identification from a post-event judgment to a process identification. This method can provide prompts when changes are still in their initial stages, enabling timely monitoring of abnormal developments during infusion, thereby improving the continuity and safety of the overall process. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This invention relates to an intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion. Detailed Implementation

[0020] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0021] This invention provides, for example Figure 1 The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion shown includes the following steps: The system continuously collects data on changes in drip rate, pressure, and flow rate during infusion. Based on preset change judgment rules, it identifies abrupt changes in each signal and marks them according to a unified time axis to form a time sequence change record corresponding to multiple source signals. During infusion, to ensure that changes in drip rate, pressure, and flow rate can be expressed continuously and in a correlated manner, multi-source changes are processed stepwise, organized on a unified time axis, and abrupt changes are clearly identified and marked, thus forming a time-series change record that can be used for subsequent analysis. The specific implementation process is as follows: The entire infusion process is divided into continuous time periods. Within each time period, the changes in drip rate, pressure, and flow rate are collected point by point. The corresponding drip rate, pressure, and flow rate values ​​at each time point are recorded in chronological order, so that the changes in drip rate, pressure, and flow rate form a continuous sequence of changes over time. During the recording process, a unique time identifier is assigned to each time point, so that the drip rate, pressure, and flow rate change sequences are all arranged correspondingly using the same time identifier as an index. This ensures that the corresponding values ​​of drip rate, pressure, and flow rate changes can be obtained simultaneously at any time position. In this way, the three types of changes form a strict correspondence in the time dimension, providing a consistent data foundation for subsequent processing on a unified time axis. Based on the established drip rate change sequence, pressure change sequence, and flow rate change sequence, the numerical changes at adjacent time points in each sequence are compared point by point. The values ​​at the current time point are compared with those at the previous time point to determine whether the change range between the two reaches the pre-set change judgment criteria. When the value at a certain time point exceeds the change judgment criteria relative to the previous time point, the time point is recorded as a change point and marked in the corresponding drip rate change sequence, pressure change sequence, or flow rate change sequence. In this process, the same point-by-point comparison operation is performed on drip rate change, pressure change, and flow rate change to ensure that the three types of changes are processed in a consistent manner during the change point identification process. This results in the sets of change points in drip rate change, pressure change, and flow rate change, respectively, and ensures that all change points have clear time markers. The sets of abrupt change points in drip rate changes, pressure changes, and flow rate changes are summarized and processed according to their respective time markers. The three types of abrupt change points are uniformly mapped onto the same time axis. All abrupt change points are arranged point by point on the unified time axis so that each time position can reflect whether there is an abrupt change in drip rate, pressure, and flow rate at that time point. During the processing, multiple abrupt change points appearing under the same time marker are recorded centrally so that the time marker is simultaneously associated with the occurrence of abrupt changes in drip rate, pressure, and flow rate. In this way, the abrupt change information that was originally scattered in different change sequences is integrated into a unified set of time markers and forms a continuously distributed sequence of abrupt change markers on the unified time axis. The drip rate change sequence, pressure change sequence, and flow rate change sequence are integrated with the abrupt change marker sequence on a unified time axis. At each time marker position, the drip rate value, pressure value, flow rate value, and corresponding abrupt change marker are combined and recorded, and arranged in the order of the time markers from front to back. This ensures that the information at all time positions is organized in a unified order, thus forming a complete time-series change record. In this time-series change record, each time position not only contains continuous numerical information of drip rate change, pressure change, and flow rate change, but also includes marker information indicating whether an abrupt change occurred at the corresponding time point. In this way, a unified record structure that can simultaneously reflect the continuous process of multi-source changes and the distribution of abrupt changes is constructed, providing a clear and complete data expression foundation for subsequent identification of synchronous changes and construction of continuous change relationships.

[0022] Based on the time sequence change record, the corresponding analysis of the mutation points in different signals is carried out within a preset time range to determine the multi-source synchronous mutation segments that appear in the same time range, and the corresponding synchronous change record is formed in the time sequence change record. Given the existing time-series change records, in order to extract change segments with synchronous characteristics from abrupt changes in drip rate, pressure, and flow rate, the distribution of these abrupt changes within a preset time range is meticulously organized and processed layer by layer. This fully unfolds the correspondence between different signals in the time dimension, thereby forming a synchronous change record that reflects the characteristics of multi-source synchronous changes. The specific implementation process is as follows: For each time marker position in the time sequence change record, a sequential scan is performed. During the scan, the current time marker is used as the center point, and a fixed number of consecutive time markers are selected forward and the same number of consecutive time markers are selected backward. The center time marker and the selected time markers together constitute a complete preset time range. Within the preset time range, the information on abrupt changes corresponding to drip rate changes, pressure changes, and flow rate changes is extracted one by one. The information is then rearranged according to the order of the time markers, so that each preset time range forms a set of time segments containing abrupt changes in drip rate, pressure, and flow rate. During the rearrangement process, the specific time position of each abrupt change in the time sequence change record is retained, so that the set of time segments not only reflects the existence of abrupt changes, but also fully reflects the distribution order of abrupt changes within the preset time range. For the established set of time segments, the occurrence of drop rate change abrupt changes, pressure change abrupt changes, and flow rate change abrupt changes within the same preset time range is compared item by item. During the comparison process, it is first confirmed whether the set of time segments contains drop rate change abrupt changes, then whether it contains pressure change abrupt changes, and then whether it contains flow rate change abrupt changes. When drop rate change abrupt changes, pressure change abrupt changes, and flow rate change abrupt changes exist simultaneously within the same preset time range, the set of time segments is identified as a candidate segment for multi-source synchronous abrupt changes. The time marker positions corresponding to the drop rate change abrupt changes, pressure change abrupt changes, and flow rate change abrupt changes are recorded in the candidate segment. At the same time, these abrupt changes are arranged according to the order of the time markers, so that the distribution of abrupt changes within the candidate segment is expressed in continuous time sequence, thereby providing a clear time correspondence for the subsequent organization of multi-source synchronous abrupt change segments. All identified multi-source synchronous mutation candidate fragments are processed centrally. Each candidate fragment is arranged according to its corresponding time marker range, and candidate fragments with overlapping or closely connected time marker ranges are merged. During the merging process, the earlier start position of the time marker is selected as the new start time marker, and the later end position of the time marker is selected as the new end time marker. The time ranges covered by multiple candidate fragments are integrated into a continuous time interval. Within this continuous time interval, the distribution of drop rate change mutation points, pressure change mutation points, and flow rate change mutation points is reorganized so that the merged fragment completely contains all mutation points in the original candidate fragments and maintains the original arrangement order of mutation points in the time sequence change record. This results in multi-source synchronous mutation fragments with continuous time ranges, so that each multi-source synchronous mutation fragment can reflect the concentrated mutation process of drop rate change, pressure change, and flow rate change within the same time range. For the compiled multi-source synchronous mutation fragments, corresponding marking processing is performed in the time sequence change records. Specifically, a start mark is set at the start time mark of each multi-source synchronous mutation fragment, and an end mark is set at the end time mark of the fragment. Synchronous change marks are uniformly marked at all time mark positions between the start and end marks, so that all records of drip rate changes, pressure changes, and flow rate changes within the time interval are associated with the same synchronous change mark. At the same time, the time range information corresponding to the synchronous change mark and the distribution of mutation points within the fragment are stored in the time sequence change records, thus forming a synchronous change record that corresponds one-to-one with the multi-source synchronous mutation fragments. This allows subsequent processing to directly identify the characteristics of multi-source synchronous mutations and conduct further continuous change correlation analysis based on the synchronous change record.

[0023] Around the synchronous change record, the preset range is extended forward and backward along the time sequence change record, and the change data within the extended range is sequentially correlated to construct a continuous change record that includes the state before the mutation and the change process after the mutation. Given the established requirement for constructing a continuous change record, to ensure that the sequential relationship between synchronous change records and chronological change records can be fully unfolded, and to create a continuous connection between the pre-mutation state and the post-mutation change process, the synchronous change records are bidirectionally expanded and the change data within the expanded range is organized layer by layer. This constructs a continuous change record with a complete time chain. The specific implementation process is as follows: For each synchronous change record that has been marked in the time sequence change record, the start time marker position and end time marker position corresponding to the synchronous change record are extracted one by one. Based on this, starting from the start time marker position, consecutive time marker positions are selected sequentially forward along the time sequence change record. At the same time, starting from the end time marker position, consecutive time marker positions are selected sequentially backward along the time sequence change record. During the forward selection process, historical data in the time sequence change record is traced back point by point according to a fixed time step. During the backward selection process, subsequent data in the time sequence change record is obtained point by point according to the same time step. The sequence of time marker positions selected forward, the time interval covered by the synchronous change record, and the sequence of time marker positions selected backward are combined to form a complete extended time interval. Within this extended time interval, the drip rate change value, pressure change value, and flow rate change value corresponding to each time marker position are extracted one by one, while keeping the original arrangement order of these values ​​in the time sequence change record unchanged, thereby obtaining an extended range change data set surrounding the synchronous change record. Based on the extracted set of change data within the extended time interval, each time marker position is organized point by point. The drop rate change value, pressure change value, and flow rate change value corresponding to the same time marker position are combined and recorded, and arranged sequentially according to the time marker order. This makes all time marker positions within the extended time interval form a continuous change sequence. In this continuous change sequence, the time interval covered by the synchronous change record is clearly marked as the middle position. This middle position plays the role of connecting the change data before and after the change in the entire continuous change sequence. At the same time, it is ensured that all time marker positions within the extended time interval are connected in chronological order. Thus, the change data before the change, the change data corresponding to the synchronous change record, and the change data after the change are presented in the same sequence, so that the changes at different stages are fully expressed in the same continuous sequence. Based on the established continuous change sequence, the change relationship between adjacent time marker positions within the extended time interval is connected segment by segment. The change data corresponding to each time marker position is sequentially connected with the previous and next time marker positions, so that the changes in drip rate, pressure, and flow rate form a continuous transmission relationship throughout the extended time interval. In this process, the change data of each time marker position is kept in a corresponding relationship with its adjacent time marker positions, so that any time point in the change sequence is in the continuous change chain. At the same time, the time interval covered by the synchronous change record is located as a key node in the continuous change chain. This key node plays the role of connecting the state before the mutation and the change process after the mutation in the entire continuous change chain, thus forming a continuous change expression structure that runs through the entire extended time interval. The constructed continuous change expression structure is systematically organized. All time marker positions within the extended time interval are uniformly numbered according to chronological order. At each numbered position, the drop rate change, pressure change, and flow rate change values ​​are recorded. Simultaneously, the start and end time marker positions corresponding to synchronous change records are marked in the numbering sequence, giving each synchronous change record a clear positional range within the continuous change expression structure. By organizing the numbering sequence, the change data before the mutation, the change data corresponding to synchronous change records, and the change data after the mutation are formed in a continuous connection within the same structure. This yields a continuous change record encompassing the pre-mutation state and the post-mutation change process, providing a complete and continuous temporal expression foundation for subsequent correlation analysis of the change process.

[0024] Based on continuous change records, the change process after the mutation is continuously tracked, and the tracking results are correlated with the state before the mutation to form a change correlation record that runs through the mutation initiation to the change regression process; Given that a continuous change record has been formed and fully expresses the pre-mutation state and post-mutation change process, in order to continuously track the post-mutation change process and establish a segmented correspondence with the pre-mutation state, the time markers in the continuous change record are systematically expanded and linked point by point around the change process, thereby forming a change association record that runs through the initiation of the mutation to the regression process. The specific implementation process is as follows: Around the end time marker position corresponding to the synchronous change record in the continuous change record, the end time marker position is selected as the starting position of the change process after the mutation. Starting from the starting position, the time marker positions in the continuous change record are read point by point in chronological order. At each time marker position, the drip rate change value, pressure change value, and flow rate change value are extracted respectively. At the same time, each time marker position is sequentially connected with the previous time marker position, so that the change process after the mutation forms a continuous change sequence with point-to-point connection in the time dimension. During the reading process, the original arrangement order of the time marker positions is maintained, so that each time marker position in the change sequence after the mutation can reflect the drip rate change, pressure change, and flow rate change at the corresponding time point. Thus, the change sequence after the mutation extending from the end time of the synchronous change record is obtained. For the established post-mutation change sequence, each time marker position in the sequence is organized point by point. The drop rate change value, pressure change value, and flow rate change value corresponding to the time marker position are combined and recorded. During the organization process, each time marker position is assigned a sequential number so that all time marker positions are arranged in the order of the numbers. At the same time, the time marker positions are kept completely consistent with the original time markers in the continuous change record during the numbering process. This allows each time point in the post-mutation change sequence to establish a direct correspondence with the corresponding position in the original continuous change record through the number, thereby forming a post-mutation change process expression with a clear time sequence identifier. The continuous time marker positions preceding the starting time marker position of the synchronous change record in the continuous change record are selected as the pre-mutation state range. Within this range, the drip rate change value, pressure change value, and flow rate change value are extracted point by point and arranged in chronological order. Each time marker position in the pre-mutation state is matched with the time marker position in the post-mutation change sequence segment by segment. In the matching process, a connection relationship is established between the last time marker position in the pre-mutation state range and the first time marker position in the post-mutation change sequence. Then, each time marker position in the pre-mutation state is sequentially associated with each time marker position in the post-mutation change sequence along the chronological order, so that the pre-mutation state and the post-mutation change process form a continuous connection in chronological order, while maintaining the consistent expression of drip rate change, pressure change, and flow rate change at the corresponding time marker positions, thereby constructing a complete change transmission relationship. The pre-mutation state range, the time interval corresponding to the synchronous change records, and the post-mutation change sequence are uniformly integrated. The three parts are arranged in the order of time markers. During the arrangement, the corresponding drop rate change value, pressure change value, and flow rate change value are retained for each time marker position. The start and end positions of the synchronous change records are clearly marked in the time marker sequence. At the same time, the time positions of the gradual decline of change are marked point by point in the post-mutation change sequence. This makes the entire change process form a continuous chain in the time dimension from the start of the mutation to the decline of the change. Through the above integration, a change correlation record is formed that runs through the process from the start of the mutation to the decline of the change. This change correlation record can fully express the continuous relationship of drop rate change, pressure change, and flow rate change in the entire change process, providing a continuous and clear time correlation basis for subsequent anomaly identification based on the change correlation record.

[0025] Based on the change association records, a matching analysis is performed on the subsequent updated time sequence change records. When the corresponding synchronous mutation fragment reappears, the anomaly identification is completed based on the change association records and an early warning message is output. Given that a change association record has been formed and fully expresses the entire process from the onset of the mutation to its fallback, in order to ensure that this change association record continues to play a role in subsequent updated time-series change records, the change process is extracted in a structured manner and compared segment by segment in subsequent data. This allows for the identification and output of early warning information at the initial stage of the change when the corresponding synchronous mutation segment reappears. The specific implementation process is as follows: For the existing change correlation records, all time marker positions contained therein are extracted sequentially. The mutation start position, the time interval corresponding to the synchronous change record, and the time marker positions corresponding to the change fall phase are arranged in chronological order to form a continuous reference change sequence. In this reference change sequence, the drop rate change value, pressure change value, and flow rate change value are recorded for each time marker position, and the order of these values ​​in the change correlation records is kept completely consistent. At the same time, the mutation start position is marked in the reference change sequence, the time interval corresponding to the synchronous change record is marked with a range, and the change process is marked point by point for each time marker position in the change fall phase. This allows the reference change sequence to fully present the entire process structure of a change from the mutation start to the change fall phase, thus forming a standard reference sequence for subsequent matching analysis. Based on the continuously updated time sequence change record, when a new time marker is added, the corresponding drop rate change value, pressure change value, and flow rate change value are appended to the end of the time sequence change record. The newly added time marker is used as the current analysis position. Starting from this position, consecutive historical time marker positions are selected point by point along the time sequence change record. The number of time markers selected is consistent with the number of time markers contained in the reference change sequence. During the selection process, the drop rate change value, pressure change value, and flow rate change value at the corresponding time marker position are extracted one by one and rearranged according to the time sequence. This ensures that the time segment to be matched is consistent with the reference change sequence in terms of time length and structure, thus providing a unified time structure basis for subsequent point-by-point matching. The time segment to be matched is compared point by point with the reference change sequence according to the time marker position. During the comparison process, the time sequence is used as the correspondence basis. Each time marker position in the time segment to be matched is matched one-to-one with the time marker position with the same number in the reference change sequence. This ensures that the drop rate change value, pressure change value, and flow rate change value are corresponding to the time marker position. During the comparison process, the focus is on whether there is a synchronous mutation segment in the time segment to be matched that is consistent with the time interval corresponding to the synchronous change record in the reference change sequence. When a continuous time interval in the time segment to be matched is consistent with the synchronous change record interval in the reference change sequence in terms of mutation distribution, the continuous time interval is identified as the corresponding synchronous mutation segment that reappears. The specific time marker range of the segment in the time sequence change record is recorded. At the same time, the continuity relationship between the segment and the time marker positions before and after is maintained, so that the synchronous mutation segment can be embedded in the continuous change process of the current time sequence change record. After confirming the reappearance of the corresponding synchronous mutation segment, the starting time marker of the synchronous mutation segment is used as the trigger position for anomaly identification. At this trigger position, the change process information of the corresponding position in the reference change sequence is combined to mark the changes in the current time sequence change record. During the marking process, the specific performance of the drop rate change value, pressure change value, and flow rate change value at the trigger position is recorded, and the mark is attached to the time marker position as a warning information. At the same time, the progress of the change process continues to be recorded along the time sequence at subsequent time marker positions, so that the warning information runs through the entire change process from the beginning of the mutation. This completes the anomaly identification and warning output based on the change association record, so that subsequent changes are identified in the initial stage and form a continuous tracking record.

[0026] This invention organizes drip rate, pressure, and flow rate changes along a unified timeline and identifies multi-source synchronous mutation fragments in the time-series change records. This effectively preserves short-term synchronous changes that are easily overlooked, preventing mutation information from being directly suppressed and thus forming a complete record at the initial stage of the change. Based on this, by constructing continuous change records, the pre-mutation state and the post-mutation change process are sequentially connected, enabling the change process to have continuous expression capabilities. This allows for early capture of abnormal starting points and improves the ability to identify abnormal stages in the infusion process.

[0027] This invention constructs a change correlation record that spans the entire process from the onset of a mutation to its regression, and continuously matches it with subsequent time-series change records. This allows historical change processes to participate in the judgment of current changes. When a corresponding synchronous mutation fragment reappears, anomaly identification can be completed and a warning message can be output, thus transforming anomaly identification from a post-event judgment to a process identification. This method can provide prompts when changes are still in their initial stages, enabling timely monitoring of abnormal developments during infusion, thereby improving the continuity and safety of the overall process.

[0028] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for intelligent infusion anomaly identification and early warning based on multi-source sensing fusion, characterized in that, Includes the following steps: The system continuously collects data on changes in drip rate, pressure, and flow rate during infusion. Based on preset change judgment rules, it identifies abrupt changes in each signal and marks them according to a unified time axis to form a time sequence change record corresponding to multiple source signals. Based on the time sequence change record, the corresponding analysis of the mutation points in different signals is carried out within a preset time range to determine the multi-source synchronous mutation segments that appear in the same time range, and the corresponding synchronous change record is formed in the time sequence change record. Around the synchronous change record, the preset range is extended forward and backward along the time sequence change record, and the change data within the extended range is sequentially correlated to construct a continuous change record that includes the state before the mutation and the change process after the mutation. Based on continuous change records, the change process after the mutation is continuously tracked, and the tracking results are correlated with the state before the mutation to form a change correlation record that runs through the mutation initiation to the change regression process; Based on the change association records, a matching analysis is performed on the subsequent updated time sequence change records. When the corresponding synchronous mutation fragment reappears, the anomaly identification is completed based on the change association records, and an early warning information is output.

2. The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion according to claim 1, characterized in that, The following steps are taken to continuously record changes in drip rate, pressure, and flow rate and generate a unified timeline representation to construct a time-series change record: The entire infusion process is divided into continuous time periods. The changes in drip rate, pressure and flow rate within each time period are collected and recorded in chronological order. Each time point is assigned a unique time identifier, forming a multi-source change sequence indexed by the time identifier. By combining multi-source change sequences, the values ​​at adjacent time points are compared point by point, and the time points that meet the change judgment criteria are identified as mutation points. The mutation points are then marked at the corresponding time marker positions to obtain the set of mutation points in each change sequence. The set of mutation points is compiled and mapped to a unified timeline, and the mutation situation at each time marker position is centrally recorded to form a unified time marker set; By integrating multi-source variation sequences with a unified set of time markers, the numerical values ​​of drip rate changes, pressure changes, and flow rate changes, as well as abrupt change markers, are recorded at each time marker location and arranged in chronological order to obtain a time-series variation record.

3. The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion according to claim 2, characterized in that, The mutation point identification process uses the same change judgment criteria, compares the changes in drip rate, pressure and flow rate point by point, and maintains a one-to-one correspondence between mutation points and time markers. At the same time, the positions of each time marker in the unified time marker set record the occurrence of multi-source mutations, thereby ensuring that the multi-source changes and mutation markers in the time sequence change record correspond synchronously.

4. The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion according to claim 2, characterized in that, The steps for synchronously identifying and recording synchronous changes of multiple mutation points within a unified timeframe are as follows: The system sequentially scans the time sequence change records and constructs a preset time range containing consecutive time markers before and after each time marker position. It also extracts the corresponding drop rate change abrupt change points, pressure change abrupt change points, and flow rate change abrupt change points to form a time segment set. A step-by-step comparison is performed on the time segment set to determine whether the same preset time range contains abrupt changes in drip rate, pressure, and flow rate. Multi-source synchronous abrupt change candidate segments are obtained and the time marker position of each abrupt change point is recorded. The candidate fragments of multi-source synchronous mutation are collected and their time-stamped ranges are organized. Candidate fragments with overlapping or closely connected relationships are integrated into continuous time intervals to form multi-source synchronous mutation fragments while preserving the original arrangement order of mutation points. By combining multi-source synchronous mutation fragments for labeling, markers are set at the start time and end time markers, and synchronous change markers are uniformly marked within the corresponding time intervals to form a synchronous change record.

5. The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion according to claim 4, characterized in that, The time marker positions of the multi-source synchronous mutation fragments are associated with the distribution order of the drop rate change mutation points, pressure change mutation points, and flow rate change mutation points within the corresponding time interval, and the synchronous change markers are associated with the corresponding time range information and the arrangement relationship of the mutation point time marker positions.

6. The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion according to claim 4, characterized in that, The steps for extending and correlating multi-source change data within the range before and after the synchronous change record to form a continuous change record are as follows: Extract the start and end time markers corresponding to the synchronous change records, extend the continuous time markers forward and backward along the time sequence change records respectively, combine them to form an extended time interval, and obtain the drip rate change value, pressure change value, and flow rate change value. Organize the change data corresponding to the extended time interval, combine the drop rate change value, pressure change value and flow rate change value corresponding to each time mark position and arrange them in chronological order to construct a continuous change sequence containing the time interval of synchronous change records; Connect adjacent time marker positions in a continuous change sequence, sequentially connect the change data of each time marker position to form a continuous change chain that runs through the preceding and following time positions, and locate the corresponding time interval of the synchronous change record; Integrate the continuous change chain and mark the start and end time markers of the synchronous change records to form a continuous change record.

7. The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion according to claim 6, characterized in that, The time interval corresponding to the synchronous change record is located in the middle of the continuous change chain. The time marker position before the synchronous change record corresponds to the state change data before the mutation, and the time marker position after the synchronous change record corresponds to the change process data after the mutation, forming a continuous change structure that runs through the time sequence.

8. The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion according to claim 6, characterized in that, The multi-source change processes are correlated chronologically to form a change correlation record, including the following steps: Extract the end time marker position corresponding to the synchronous change record in the continuous change record, read the continuous time marker position along the time sequence change record and obtain the drip rate change value, pressure change value and flow rate change value, and construct the change sequence after the sudden change. Organize the time marker positions in the change sequence after the mutation, combine and record the corresponding drop rate change value, pressure change value and flow rate change value, and arrange them in time order to form a change expression with continuous time markers. Extract the consecutive time markers before the corresponding start time marker position of the synchronous change record and obtain the corresponding change data. Sequentially associate the state range before the mutation with the change sequence after the mutation and establish a continuous connection relationship. By integrating the pre-mutation state range, the time interval of synchronous change records, and the post-mutation change sequence, and arranging them in chronological order, and marking the start and end time markers of the synchronous change records, a change association record is formed that runs through the process from the onset of the mutation to the decline of the change.

9. The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion according to claim 8, characterized in that, The time marker positions in the post-mutation change sequence are arranged in a continuous order, and the corresponding values ​​of drip rate change, pressure change and flow rate change for each time marker position are consistent. They are also connected one-to-one with the corresponding time marker positions in the pre-mutation state range, thus ensuring a continuous transmission relationship between the time marker positions in the change association record.

10. The intelligent infusion anomaly identification and early warning method based on multi-source sensing fusion according to claim 8, characterized in that, The change-related records are matched with the chronological change records to complete the anomaly identification process, as follows: Extract the time marker positions from the change correlation records and arrange them in chronological order to form a reference change sequence. At the same time, mark the starting position of the mutation and the time interval of the synchronous change record, and record the drop rate change value, pressure change value and flow rate change value. Add the change data corresponding to the newly added time marker position in the time sequence change record, extract the continuous time marker positions forward along the time sequence change record and construct the time segment to be matched, and maintain the time structure consistent with the reference change sequence. By identifying the time marker positions of the time segment to be matched and the reference change sequence, the continuous time interval that is consistent with the time interval of the synchronous change record is identified and determined as the corresponding synchronous mutation segment; The system marks the start time of the corresponding synchronous mutation fragment, records the change data, generates early warning information, and continuously records the change process along the time sequence.

Images