A method and system for postoperative delirium risk assessment based on electroencephalogram biomarkers

Abstract

The application discloses a postoperative delirium risk assessment method and system based on an electroencephalogram biomarker, acquires an electroencephalogram complexity sequence, a hemodynamics sequence and an inflammation index sequence in a general anesthesia operation process, divides a sign duration section based on an electroencephalogram complexity change direction and constructs a neural inhibition stage boundary; identifies a perfusion reverse turning structure in the stage and checks the stage boundary; further identifies an inflammation same-direction continuation section crossing adjacent stage boundaries, constructs a three-domain coupling state node; detects a nested structure in a continuous stage to determine a composite risk segment, and corrects the position of the stage boundary, and outputs a risk trajectory. By establishing a three-domain coupling structure with consistent topology across stages, the early identification stability and discrimination accuracy of the postoperative delirium risk are improved.

Description

Technical Field

[0001] This invention relates to the field of clinical anesthesia monitoring and decision support technology, specifically to a method and system for assessing postoperative delirium risk based on electroencephalogram (EEG) biomarkers. Background Technology

[0002] Postoperative delirium (POD) is a common perioperative neurological complication in elderly patients, and its occurrence is closely related to the degree of intraoperative brain function inhibition, fluctuations in cerebral perfusion, and the level of inflammatory response. In recent years, risk assessment methods based on electroencephalogram (EEG) biomarkers have been increasingly applied in perioperative monitoring. Existing techniques typically extract statistics within a fixed time window based on features such as EEG spectral characteristics, complexity indices, or the Burst Suppression ratio, and then input these statistics into a machine learning model along with hemodynamic parameters and inflammatory marker values ​​to output the probability of postoperative delirium risk. However, these methods have the following main shortcomings:

[0003] 1. Existing methods typically extract features using sliding time windows, but they do not establish the same-direction change segments and stage boundary structures of EEG complexity on the time axis, making it difficult to distinguish between sustained inhibition and short-term fluctuations, resulting in a lack of stage semantics in risk identification.

[0004] 2. Hemodynamic fluctuations often turn within the EEG inhibition phase, but current technology only treats them as independent numerical features. It does not identify whether the perfusion reversal structure is completely enveloped by a single EEG phase, nor does it verify whether the phase boundary crosses the perfusion transition segment. This may lead to a misalignment between phase division and perfusion fluctuations.

[0005] 3. Inflammatory markers usually show a lagged, continuous upward trend, but existing methods mostly model them using instantaneous values ​​or statistical mean values. They fail to identify whether inflammatory changes continue across the boundaries of EEG stages and cannot characterize the cross-stage coupling relationship between neural inhibition, perfusion fluctuations and inflammatory responses.

[0006] 4. EEG, hemodynamics and inflammatory markers change at different scales. Existing methods usually align them by using a uniform time window or resampling, which can easily disrupt the original time structure, causing key turning points to be truncated or smoothed, affecting the accuracy of risk segment identification.

[0007] In view of this, the present invention provides a method and system for assessing the risk of postoperative delirium based on electroencephalogram (EEG) biomarkers, thereby solving the above-mentioned problems. Summary of the Invention

[0008] The purpose of this invention is to provide a method and system for assessing postoperative delirium risk based on electroencephalogram (EEG) biomarkers, which solves the technical problem that the postoperative delirium risk assessment method based on EEG biomarkers cannot identify the complex risk trajectory formed by the superposition of continuous neural inhibition and transient perfusion fluctuations across time scales, resulting in delayed or misjudged intraoperative risk warnings.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] In a first aspect, the present invention provides a method for assessing the risk of postoperative delirium based on electroencephalographic biomarkers, comprising the following steps:

[0011] Obtain EEG complexity sequences, hemodynamic sequences, and inflammatory marker sequences during general anesthesia surgery, and arrange them in chronological order;

[0012] The direction of value change of adjacent sampling points in the EEG complexity sequence is compared point by point. The sampling point segments with consistent direction and continuous time are divided into symbol duration segments, and the position of direction reversal between adjacent symbol duration segments is determined as the boundary of the initial neural inhibition stage.

[0013] Within the time interval defined by the boundaries of two adjacent initial neural inhibition phases, identify perfusion reversal structures in the hemodynamic sequence that have reversed direction and whose start and end times are located within the same symbol duration. The boundaries of the initial neural inhibition phases that do not fall within the time interval of any perfusion reversal structure are determined as the verified neural inhibition phase boundaries.

[0014] Based on the verified neural inhibition stage boundary identification inflammatory index sequence, a three-domain coupled state node is constructed when the perfusion reverse inflection structure is located inside a single stage and the unidirectional continuation segment crosses the stage boundary.

[0015] The system detects and identifies time segments in which nested structures persist in two or more consecutive phases as composite risk segments. When the phase boundary time position falls within the composite risk segment, the verified neural inhibition phase boundary is moved and the state node is reconstructed to output the postoperative delirium risk trajectory.

[0016] As a preferred embodiment of the present invention, the division of the symbol duration includes:

[0017] The EEG complexity sampling points are numbered sequentially by time as the kth sampling point and the (k+1)th sampling point. The complexity values ​​of the (k+1)th sampling point and the kth sampling point are compared, and the comparison result is recorded as the directional attribute of the (k+1)th sampling point.

[0018] When multiple consecutive sampling points have the same directional properties and are continuous in time, the set of sampling points is determined as a symbol duration segment, and the start and end times of the symbol duration segment are recorded.

[0019] As a preferred embodiment of the present invention, the determination of the boundary of the initial neural inhibition phase further includes:

[0020] At the boundary of each candidate neural inhibition phase, detect whether there is temporal overlap or sampling discontinuity between adjacent symbol durations;

[0021] When sampling interruptions exist, the boundary of the candidate neural inhibition phase is moved to the nearest valid sampling point time position.

[0022] As a preferred embodiment of the present invention, the identification of the infusion reverse turning structure includes:

[0023] Extracting continuous segments of hemodynamic change with consistent direction of change from within a single initial neural inhibition phase;

[0024] When a reversal occurs between adjacent continuous change segments and the start and end times of the reversal segment are both within the time range of the same symbol duration, the reversal segment is identified as a reversal structure in the infusion process, and its start and end times are recorded.

[0025] As a preferred embodiment of the present invention, the determination of the boundary of the verified neural inhibition stage includes:

[0026] The temporal position of the boundary of each initial neural inhibition phase is compared with the start and end times of all perfusion reversal inflection structures;

[0027] When the time position falls within the time segment of any perfusion reverse inflection structure, the initial neural inhibition stage boundary is removed from the verified neural inhibition stage boundary set.

[0028] As a preferred embodiment of the present invention, the identification of the same-direction continuation segment includes:

[0029] Extract segments of inflammatory marker sequences that exhibit consistent trends and continuous temporal variation.

[0030] When the start time of the change segment is later than the end time of a certain perfusion reverse inflection structure, and its time segment crosses the boundary of the adjacent verified neural inhibition stage, and its change direction is consistent with the corresponding symbol duration segment, the change segment is determined as a co-directional continuation segment.

[0031] As a preferred embodiment of the present invention, the three-domain coupled state node includes:

[0032] The neural inhibition stage number, the start and end times of the perfusion reverse inflection structure, and the start and end times of the same-direction continuation segment are used as the main index to group and sequentially connect the state nodes.

[0033] As a preferred embodiment of the present invention, the determination of the composite risk segment includes:

[0034] When there are three-domain coupled state nodes in two or more consecutive verified neural inhibition phases, satisfying that the perfusion reverse inflection structure is located inside the phase and the same-direction continuation segment crosses the phase boundary, the time interval between the start boundary time and the end boundary time of the consecutive phases is determined as a composite risk segment.

[0035] As a preferred embodiment of the present invention, the movement of the verified boundary of the neural inhibition phase includes:

[0036] When the temporal position of the boundary of the neural inhibition phase falls between the start and end times of the composite risk segment,

[0037] The boundary of the neural inhibition phase is moved to the start or end time position of the composite risk segment, and the perfusion reverse inflection structure identification and the same-direction continuation segment identification are re-executed based on the moved phase boundary.

[0038] In a second aspect, the present invention provides a postoperative delirium risk assessment system based on electroencephalogram (EEG) biomarkers, which, based on the implementation of the first aspect, includes:

[0039] The data acquisition module is used to acquire EEG complexity sequences, hemodynamic sequences, and inflammatory marker sequences during general anesthesia surgery, and retain the original timestamps of each sequence;

[0040] The phase division module is used to compare the direction of value change between adjacent sampling points based on the EEG complexity sequence to form a set of directional attributes, and divide the sampling point segments with consistent directional attributes and continuous time into symbol duration segments. The initial neural inhibition phase boundary is determined based on the directional reversal position between adjacent symbol duration segments.

[0041] The perfusion verification module is used to identify perfusion reversal structures in the hemodynamic sequence within the time interval defined between two adjacent initial neural inhibition phase boundaries, where the direction is reversed and the start and end times are both within the same symbol duration. The initial neural inhibition phase boundaries that do not fall within the time interval of any perfusion reversal structure are determined as the verified neural inhibition phase boundaries.

[0042] The inflammation coupling module is used to identify the same-direction continuation segment in the inflammatory index sequence that crosses the boundary of adjacent neural inhibition stages and whose change direction is consistent with the corresponding symbol duration segment based on the verified neural inhibition stage boundary. It also constructs a three-domain coupling state node when the perfusion reverse turning structure is located inside a single neural inhibition stage and the same-direction continuation segment crosses the boundary of that stage.

[0043] The risk identification and correction module is used to detect the time segments in which nested structures persist in two or more consecutive neural inhibition phases in the three-domain coupled state node and identify composite risk segments. The start and end times of the composite risk segments are mapped to the original time axes of the hemodynamic sequence and the inflammatory marker sequence. When there is a positional conflict between the composite risk segment and the verified neural inhibition phase boundary, the phase boundary is adjusted and the three-domain coupled state node is reconstructed to output the postoperative delirium risk trajectory.

[0044] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0045] This invention extracts symbol duration segments from complex EEG sequences and establishes neural inhibition stage boundaries by reversing their directions. This transforms the EEG inhibition process from a continuous fluctuating signal into a stage structure with a defined time interval, thus providing a stable time container for nested determination of cross-modal signals. Furthermore, by identifying perfusion reversal inflection structures within a single stage and using these structures to perform time verification of stage boundaries, it can eliminate stage division points that traverse hemodynamic inflection processes, ensuring structural consistency between neural inhibition stages and perfusion fluctuations. Building upon this, by identifying inflammatory continuation segments that cross adjacent stage boundaries, it constructs a structure encompassing both "intra-stage perfusion reversal" and "cross-stage inflammatory continuation." The three-domain coupled state nodes explicitly express the cross-timescale coupling relationship between neural inhibition, perfusion changes, and inflammatory responses. When the coupled structure persists in a continuous phase, it is identified as a composite risk segment. Through boundary conflict judgment and position correction mechanisms, the temporal topological relationship between the phase boundary and the risk segment is kept consistent, thereby avoiding structural fragmentation caused by boundary cutting. Since the phase division and the coupled structure maintain topological consistency, the three-domain abnormality pattern can be fully expressed, thereby improving the continuity and discrimination stability of the risk trajectory. This makes the assessment results of postoperative delirium risk closer to the real physiological evolution process, improves the accuracy of early identification, and reduces the probability of false positives. Attached Figure Description

[0046] 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.

[0047] Figure 1 This is a flowchart of the postoperative delirium risk assessment method based on electroencephalogram biomarkers of the present invention;

[0048] Figure 2 This is a schematic diagram of the three-domain physiological signal coupling of the present invention;

[0049] Figure 3 This is a schematic diagram illustrating the principle of boundary correction at different stages of the present invention.

[0050] Figure 4 This is a framework diagram of the postoperative delirium risk assessment system based on electroencephalogram biomarkers of the present invention. Detailed Implementation

[0051] 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. The drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0052] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more exemplary embodiments. Numerous specific details are provided in the following description to give a full understanding of exemplary embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure may be practiced with one or more specific details omitted, or methods, components, steps, etc. In other instances, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0053] Example 1

[0054] like Figure 1 As shown, this invention provides a method for assessing the risk of postoperative delirium based on electroencephalographic biomarkers, comprising the following steps:

[0055] S101: Obtain EEG complexity sequence, hemodynamic sequence, and inflammatory marker sequence during general anesthesia surgery;

[0056] Specifically, during general anesthesia surgery, EEG complexity sampling points are obtained through anesthesia monitoring equipment, with each sampling point including a timestamp and the corresponding complexity value; hemodynamic sampling points are obtained through blood pressure monitoring modules or cardiac output monitoring modules, with each sampling point including a timestamp and the corresponding hemodynamic value; and inflammatory marker sampling points are obtained through laboratory testing or bedside testing equipment, with each inflammatory marker sampling point including a timestamp and the corresponding inflammatory marker value.

[0057] EEG complexity sampling points, hemodynamic sampling points, and inflammatory marker sampling points were sorted according to their timestamps, and their original timestamps were retained. No uniform resampling processing was performed on sampling points of different types. The sorted EEG complexity sampling points were used as the baseline data source for symbol duration extraction, and after sorting, EEG complexity sequences, hemodynamic sequences, and inflammatory marker sequences were obtained respectively.

[0058] S102: Extract the direction of change of the EEG complexity sequence and divide it into multiple symbolic duration segments. The symbolic duration segments are used to characterize the unidirectional change of EEG complexity over time. Construct the boundary of the initial neural inhibition stage based on the reverse position of the change of the symbolic duration segments.

[0059] Specifically, the division of the symbol duration includes:

[0060] The EEG complexity sampling points sorted by time are the kth sampling point and the (k+1)th sampling point. The complexity value of the (k+1)th sampling point is compared with that of the kth sampling point to determine the direction of change between them. The direction of change is recorded as the direction attribute of the (k+1)th sampling point, and so on, to obtain the set of direction attributes corresponding to all EEG complexity sampling points. The set of direction attributes is used for subsequent continuous segment splicing.

[0061] The EEG complexity sampling points are scanned sequentially along the time axis. When the directional attributes of adjacent sampling points are consistent and the time is continuous, the corresponding sampling points are grouped into the same symbol duration segment. When the directional attributes change, the current symbol duration segment is terminated and a new symbol duration segment is started. The start time and end time of each symbol duration segment are recorded.

[0062] Extract the direction of change and start and end times of adjacent symbol durations. When the direction of change of two adjacent symbol durations is opposite and the end time of the previous symbol duration is directly adjacent to the start time of the next symbol duration on the time axis, the switching position between the two symbol durations is determined as the boundary of the candidate neural inhibition stage. Arrange all candidate neural inhibition stage boundaries in chronological order to form the initial neural inhibition stage boundary set.

[0063] In other words, the duration of adjacent symbols, their respective directions of change, and start and end times are extracted. Adjacent durations are examined; if their directions of change are opposite and their end and start times are directly adjacent on the time axis, the switching position between the two segments is determined as the candidate neural inhibition stage boundary. All candidate neural inhibition stage boundaries are arranged in chronological order to obtain the initial set of neural inhibition stage boundaries. The initial neural inhibition stage boundaries are used to mark the switching positions between adjacent segments moving in the same direction, while the candidate neural inhibition stage boundaries form the basis for subsequent stage determinations: only after the boundaries are determined can the corresponding time intervals for each stage be formed, which are then used for intra-stage identification and envelope judgment of perfusion transition structures.

[0064] To further clarify, the determination of the boundary of the initial neural inhibition phase also includes:

[0065] At the boundary of each candidate neural inhibition stage, the start and end times of the symbol duration segments on both sides of the boundary are extracted to determine whether there is temporal overlap between the two symbol duration segments on the time axis; at the same time, the timestamps of the EEG complexity sampling points near the boundary are extracted to determine whether there are sampling missing segments between adjacent sampling points.

[0066] When there is no temporal overlap between adjacent symbol durations and no missing sampling segments, the corresponding candidate neural inhibition stage boundary is retained; when there is a missing sampling segment at the candidate neural inhibition stage boundary, the time position of the candidate neural inhibition stage boundary is moved to the timestamp of the effective EEG complexity sampling point closest to the candidate neural inhibition stage boundary; after the above processing, the initial neural inhibition stage boundary set is obtained.

[0067] By constructing symbolic duration segments, the second-level EEG complexity fluctuations are transformed into time intervals of unidirectional change; the stage boundaries are determined by reversing the direction of position, forming neural inhibition stages with clear time ranges; boundary stabilization processing avoids interference from sampling anomalies on stage division. The initial set of neural inhibition stage boundaries provides a time constraint basis for subsequent identification of perfusion reversal structures and construction of three-domain coupled state nodes.

[0068] S103: Within the boundary of the initial neural inhibition phase, determine whether the hemodynamic sampling point forms a perfusion reverse inflection structure enveloped by the same symbol duration, and map the start and end positions of the perfusion reverse inflection structure to the boundary of the initial neural inhibition phase to identify whether the boundary of the initial neural inhibition phase crosses the perfusion reverse inflection structure; determine the initial neural inhibition phase boundary that does not cross the perfusion reverse inflection structure as the verified neural inhibition phase boundary.

[0069] It should be noted that the initial neural inhibition stage boundaries are arranged in chronological order. The time interval between two adjacent initial neural inhibition stage boundaries is defined as a neural inhibition stage interval. Within a single neural inhibition stage interval, two adjacent hemodynamic sampling points are designated as the m-th sampling point and the (m+1)-th sampling point. The hemodynamic values ​​of the (m+1)-th sampling point and the m-th sampling point are compared to determine the direction of change between them. This direction of change is recorded as the directional attribute of the (m+1)-th hemodynamic sampling point. The hemodynamic sequence is then traversed sequentially to obtain the set of directional attributes corresponding to each hemodynamic sampling point.

[0070] Hemodynamic sampling points are scanned sequentially along the time axis. When adjacent sampling points have the same directional attributes and are continuous in time, the corresponding sampling points are grouped into a continuous change segment. When the directional attributes change, the current continuous change segment is terminated and a new continuous change segment is started. The start time and end time of each continuous change segment are recorded.

[0071] When a continuous change segment undergoes a change in direction attribute within the time range defined by the same symbol duration segment, and the start and end times corresponding to the change in direction attribute are both located within the symbol duration segment, the time segment is defined as a perfusion reversal transition structure; for each perfusion reversal transition structure, its start time, end time, and the interval number of the neural inhibition stage to which it belongs are recorded.

[0072] Extract the start and end times of each perfusion reversal structure and compare them with the boundary times in the initial neural inhibition phase boundary set. When the time position of a certain initial neural inhibition phase boundary is between the start and end times of a certain perfusion reversal structure, the initial neural inhibition phase boundary is marked as crossing the perfusion reversal structure. When the time position of a certain initial neural inhibition phase boundary does not fall within the time range of any perfusion reversal structure, the initial neural inhibition phase boundary is marked as not crossing.

[0073] All initial neural inhibition stage boundaries that do not cross perfusion reversal structures are retained, and all initial neural inhibition stage boundaries that cross perfusion reversal structures are deleted, forming a validated set of neural inhibition stage boundaries.

[0074] S104: Based on the verified neural inhibition stage boundary, identify whether the inflammatory index sampling points form a continuous segment in the same direction that crosses the boundary of adjacent neural inhibition stages on the time axis, and construct a three-domain coupled state node according to the nested positional relationship between the verified neural inhibition stage boundary, the perfusion reverse turning structure and the continuous segment in the same direction.

[0075] Based on the set of verified neural inhibition stage boundaries arranged in chronological order, the time interval between two adjacent verified neural inhibition stage boundaries is defined as a verified neural inhibition stage interval. Let the time-ordered inflammatory marker sampling points be the nth inflammatory marker sampling point and the (n+1)th inflammatory marker sampling point; compare the magnitude of the inflammatory marker value at the (n+1)th inflammatory marker sampling point with the inflammatory marker value at the nth inflammatory marker sampling point to determine the direction of change between them; record this direction of change as the directional attribute of the (n+1)th inflammatory marker sampling point; sequentially traverse the inflammatory marker sequence to obtain the set of directional attributes corresponding to each inflammatory marker sampling point.

[0076] The inflammatory marker sampling points are scanned sequentially along the time axis. When the directional attributes of adjacent sampling points are consistent and the time is continuous, the corresponding sampling points are grouped into a continuous change segment. When the directional attributes change, the current continuous change segment is terminated and a new continuous change segment is started. The start time and end time of each continuous change segment are recorded.

[0077] For each segment of continuous change in inflammatory markers, perform the following judgments sequentially:

[0078] (1) Determine whether the start time of the continuously changing section is later than the end time of a certain injection reverse turning structure;

[0079] (2) Determine whether the time range of the continuously changing segment crosses the boundary between two adjacent verified neural inhibition phases;

[0080] (3) Determine whether the directional attribute of the continuously changing segment is consistent with the changing direction of the corresponding symbol duration segment within the time range.

[0081] When a continuously changing segment simultaneously meets the above three conditions, the continuously changing segment is identified as a continuous segment in the same direction; the start time, end time, and the number of the neural inhibition stage it crosses are recorded.

[0082] For each perfusion reversal structure, determine whether its start and end times are both within a single verified neural inhibition phase interval; simultaneously, determine whether there is a co-current continuation segment that crosses the boundary of the verified neural inhibition phase and extends into an adjacent phase; when the above conditions are met simultaneously, record the interval number of the verified neural inhibition phase, the time range of the corresponding perfusion reversal structure, and the time range of the corresponding co-current continuation segment, forming a three-domain coupled state node. This three-domain coupled state node simultaneously includes phase boundary constraints, perfusion reversal constraints, and inflammation cross-phase continuation constraints on the time axis, providing a structural basis for subsequent identification of complex risk fragments.

[0083] S105: Establish a connection relationship between the three-domain coupled state nodes with the verified neural inhibition stage boundary as the main index, and identify structural combinations that simultaneously satisfy the condition that the perfusion reverse inflection structure is located inside the neural inhibition stage and the unidirectional continuation segment crosses the adjacent neural inhibition stage. Determine the structural combination as a composite risk segment, wherein the composite risk segment presents a topological positional relationship on the time axis that the EEG inhibition stage contains a perfusion reverse inflection structure and the inflammatory unidirectional continuation segment crosses the adjacent EEG inhibition stage.

[0084] The three-domain coupled state nodes include: the verified interval number of the neural inhibition stage, the start and end times of the corresponding perfusion reverse inflection structure, and the start and end times of the corresponding same-direction continuation segment.

[0085] The three-domain coupled state nodes are grouped according to the verified neural inhibition stage interval number. Three-domain coupled state nodes with the same verified neural inhibition stage interval number are grouped into the same stage node group. The sequential connection relationship of each stage node group is established according to the time order of the verified neural inhibition stage interval number to form a stage node sequence structure.

[0086] In the stage node sequence structure, traverse two or more adjacent verified neural inhibition stage intervals sequentially and perform the following judgment:

[0087] (1) Determine whether there are three-domain coupled state nodes in the current stage node group, whose start and end times of the corresponding infusion reverse turning structure are both located within the stage interval, and whose time range of the corresponding same-direction continuation segment spans the stage interval and its adjacent stage interval.

[0088] (2) Determine whether there are any three-domain coupled state nodes in the subsequent stage node group adjacent to the stage interval that satisfy the above conditions.

[0089] When two or more consecutive verified neural inhibition phase intervals contain three-domain coupled state nodes that meet the above conditions, the start boundary time of the first phase interval of the consecutive phase interval is determined as the candidate start time, and the end boundary time of the last phase interval of the consecutive phase interval is determined as the candidate end time. The continuous time segment between the candidate start time and the candidate end time is determined as a composite risk segment, and the start time, end time, and set of verified neural inhibition phase interval numbers involved in each composite risk segment are recorded.

[0090] By detecting the continuous existence of nested structures in the stage master index structure, coupled anomalous structures that persist across stages can be identified. The composite risk segments exhibit the following structural features on the time axis: within the EEG inhibition stage, there are perfusion reversal inflection structures and unidirectional continuation segments spanning adjacent EEG inhibition stages; they recur in continuous stages, reflecting a cross-timescale neural inhibition-perfusion-inflammation coupling pattern, which is used for subsequent risk trajectory construction.

[0091] S106: The distribution position of the composite risk fragment on the verified boundary of the neural inhibition stage is reverse-mapped to the time mapping relationship between the hemodynamic sampling point and the inflammatory index sampling point. If there is a topological inconsistency between the composite risk fragment and the verified boundary of the neural inhibition stage, the verified boundary of the neural inhibition stage is readjusted and the three-domain coupled state node is reconstructed. The postoperative delirium risk trajectory is output based on the adjusted node sequence.

[0092] Each composite risk segment includes: a start time, an end time, and a set of verified neural inhibition phase numbers involved. The start and end times of the composite risk segment are mapped to the timestamps of the hemodynamic and inflammatory marker sequences to confirm the original temporal position of the time segment within these sequences. A unified temporal reference relationship is established between the composite risk segment and the three types of sampling points through a mapping relationship.

[0093] The integrity determination of the composite risk segment includes:

[0094] (1) Determine whether the time segment of the composite risk segment completely covers the start and end times of the corresponding infusion reverse turning structure;

[0095] (2) Determine whether the composite risk segment contains a complete segment spanning the same direction of continuation;

[0096] When a composite risk segment meets all of the above conditions, the composite risk segment is determined as a complete structural unit.

[0097] Based on the complete set of structural units, the verified set of neural inhibition stage boundaries is checked one by one. For each complete structural unit, the following processing is performed:

[0098] Determine whether there is a verified neural inhibition phase boundary whose time position is between the start and end times of the complete structural unit;

[0099] When the above boundary exists, the stage boundary is moved to the start or end time position of the complete structural unit;

[0100] When the boundary of the same stage is located inside multiple complete structural units, the position adjustment is performed sequentially according to the time order of the complete structural units.

[0101] After completing the above adjustments, the adjusted set of neural inhibition phase boundaries is obtained.

[0102] Based on the adjusted set of neural inhibition stage boundaries, the processing rules of steps S103 and S104 are re-executed, the neural inhibition stage intervals are re-divided, the same-direction continuation segments are re-identified, and the set of three-domain coupled state nodes is reconstructed. The updated set of three-domain coupled state nodes is arranged in chronological order, and the time segments corresponding to the composite risk segments are marked on the time axis to form a postoperative delirium risk trajectory.

[0103] Example 2

[0104] like Figure 4 As shown, this embodiment provides a postoperative delirium risk assessment system based on EEG biomarkers, based on embodiment 1. The system includes a data acquisition module, a stage division module, a perfusion verification module, an inflammation coupling module, and a risk identification and correction module. The above modules establish data association through a unified time axis and form a progressive and reversible processing link in sequence.

[0105] The data acquisition module is used to acquire EEG complexity sequences, hemodynamic sequences, and inflammatory marker sequences during general anesthesia surgery, and retain the original timestamps of each sequence;

[0106] The phase division module is used to compare the direction of value change between adjacent sampling points based on the EEG complexity sequence to form a set of directional attributes, and divide the sampling point segments with consistent directional attributes and continuous time into symbol duration segments. The initial neural inhibition phase boundary is determined based on the directional reversal position between adjacent symbol duration segments.

[0107] The perfusion verification module is used to identify perfusion reversal structures in the hemodynamic sequence within the time interval defined between two adjacent initial neural inhibition phase boundaries, where the direction is reversed and the start and end times are both within the same symbol duration. The initial neural inhibition phase boundaries that do not fall within the time interval of any perfusion reversal structure are determined as the verified neural inhibition phase boundaries.

[0108] The inflammation coupling module is used to identify the same-direction continuation segment in the inflammatory index sequence that crosses the boundary of adjacent neural inhibition stages and whose change direction is consistent with the corresponding symbol duration segment based on the verified neural inhibition stage boundary. It also constructs a three-domain coupling state node when the perfusion reverse turning structure is located inside a single neural inhibition stage and the same-direction continuation segment crosses the boundary of that stage.

[0109] The risk identification and correction module is used to detect the time segments in which nested structures persist in two or more consecutive neural inhibition phases in the three-domain coupled state node and identify composite risk segments. The start and end times of the composite risk segments are mapped to the original time axes of the hemodynamic sequence and the inflammatory marker sequence. When there is a positional conflict between the composite risk segment and the verified neural inhibition phase boundary, the phase boundary is adjusted and the three-domain coupled state node is reconstructed to output the postoperative delirium risk trajectory.

[0110] Preferably, the division of the symbol duration includes:

[0111] The EEG complexity sampling points are numbered sequentially by time as the kth sampling point and the (k+1)th sampling point. The complexity values ​​of the (k+1)th sampling point and the kth sampling point are compared, and the comparison result is recorded as the directional attribute of the (k+1)th sampling point.

[0112] When multiple consecutive sampling points have the same directional properties and are continuous in time, the set of sampling points is determined as a symbol duration segment, and the start and end times of the symbol duration segment are recorded.

[0113] Preferably, determining the boundary of the initial neural inhibition phase further includes:

[0114] At the boundary of each candidate neural inhibition phase, detect whether there is temporal overlap or sampling discontinuity between adjacent symbol durations;

[0115] When sampling interruptions exist, the boundary of the candidate neural inhibition phase is moved to the nearest valid sampling point time position.

[0116] Preferably, the identification of the infusion reverse turning structure includes:

[0117] Extracting continuous segments of hemodynamic change with consistent direction of change from within a single initial neural inhibition phase;

[0118] When a reversal occurs between adjacent continuous change segments and the start and end times of the reversal segment are both within the time range of the same symbol duration, the reversal segment is identified as a reversal structure in the infusion process, and its start and end times are recorded.

[0119] Preferably, the determination of the verified boundary of the neural inhibition phase includes:

[0120] The temporal position of the boundary of each initial neural inhibition phase is compared with the start and end times of all perfusion reversal inflection structures;

[0121] When the time position falls within the time segment of any perfusion reverse inflection structure, the initial neural inhibition stage boundary is removed from the verified neural inhibition stage boundary set.

[0122] Preferably, the identification of the same-direction continuation segment includes:

[0123] Extract segments of inflammatory marker sequences that exhibit consistent trends and continuous temporal variation.

[0124] When the start time of the change segment is later than the end time of a certain perfusion reverse inflection structure, and its time segment crosses the boundary of the adjacent verified neural inhibition stage, and its change direction is consistent with the corresponding symbol duration segment, the change segment is determined as a co-directional continuation segment.

[0125] Preferably, the three-domain coupled state node includes:

[0126] The neural inhibition stage number, the start and end times of the perfusion reverse inflection structure, and the start and end times of the same-direction continuation segment are used as the main index to group and sequentially connect the state nodes.

[0127] Preferably, the determination of the composite risk segment includes:

[0128] When there are three-domain coupled state nodes in two or more consecutive verified neural inhibition phases, satisfying that the perfusion reverse inflection structure is located inside the phase and the same-direction continuation segment crosses the phase boundary, the time interval between the start boundary time and the end boundary time of the consecutive phases is determined as a composite risk segment.

[0129] Preferably, the shift of the verified boundary of the neural inhibition phase includes:

[0130] When the temporal position of the boundary of the neural inhibition phase falls between the start and end times of the composite risk segment,

[0131] The boundary of the neural inhibition phase is moved to the start or end time position of the composite risk segment, and the perfusion reverse inflection structure identification and the same-direction continuation segment identification are re-executed based on the moved phase boundary.

[0132] Example 3

[0133] This embodiment is based on Embodiment 1, and takes the physiological monitoring data of a 75-year-old patient during the induction and maintenance phases of general anesthesia surgery as an example;

[0134] First, obtain the EEG complexity sequence, hemodynamic sequence, and inflammatory marker sequence during the general anesthesia procedure, such as... Figure 2 As shown; where:

[0135] The permutation entropy algorithm is used for real-time calculation, and the EEG complexity sequence Ct is acquired at a sampling frequency of 10Hz. The values ​​of adjacent EEG complexity sampling points are compared point by point to extract the directional attribute;

[0136] Symbolic duration segment: The first 4 points 45→42→38→35 show a continuous decrease with a direction attribute of "-", which is identified as symbolic duration segment D1 (representing deepening neural inhibition); Points 5-6 35→36→38 show a continuous increase with a direction attribute of "+", which is identified as symbolic duration segment D2 (representing short-term relief of inhibition); Points 7-9 38→34→30→28 show another continuous decrease with a direction attribute of "-", which is identified as symbolic duration segment D3.

[0137] Boundary localization: Initial neural inhibition stage boundaries B1 and B2 are established at the transition point between D1 and D2 (between points 4 and 5) and the transition point between D2 and D3 (between points 6 and 7), respectively, forming three neural inhibition stage intervals.

[0138] Hemodynamic sequence MAP was acquired using an invasive arterial blood pressure monitoring module at a sampling frequency of 1 Hz. Within the interval defined by the initial boundary, hemodynamic features were simultaneously retrieved. Within the time envelope corresponding to D1, the MAP sequence exhibited fluctuations of [85, 82, 75, 78, 80]. Comparing the directions of adjacent sampling points: 85→82→75 was a decrease; 75→78→80 was an increase. The lowest point (75 mmHg) of this "V-shaped" reverse inflection structure and its recovery phase were identified as being completely within the sign duration D1. Upon comparison, the initial boundary B1 was found to be located after the time endpoint of this inflection structure, without any "boundary penetration" phenomenon. B1 was determined to be a physiologically logical steady-state boundary and was identified as the verified boundary of the neural inhibition stage. Three neural inhibition stage intervals were formed between the initial boundary B1 and the initial boundary B2: stage A (D1), stage B (D2), and stage C (D3).

[0139] The IL-6 inflammatory marker sequence was acquired using a point-of-care testing (POCT) device. Cross-dimensional feature coupling was performed using the IL-6 inflammatory sequence. Near the boundary B1, the IL-6 sequence showed a continuous upward trend. The algorithm detected that this upward trend started in stage D1, crossed boundary B1 on the time axis, and continued to B2, classifying it as a "continuous segment in the same direction." At this point, the system identified a nested relationship satisfying both the "D1 stage envelope perfusion reversal structure" and "IL-6 rise crossing B1," constructing a three-domain coupled state node. When identifying composite risk segments, it was found that due to the physiological lag in the inflammatory response of elderly patients, the initially defined boundary B2 precisely cut off the peak region of the inflammatory rise segment, resulting in impaired connectivity of the three-domain state tensor at B2 and a "topological inconsistency" conflict.

[0140] like Figure 3 As shown, the left side represents the original stage division structure; the right side represents the composite risk segment time interval (Ts—Te). The initial boundary B2 is located inside the inflammation rising peak, i.e., B2 ∈ [Ts, Te]. The position conflict determination checks whether the stage boundary is located inside the risk segment. If a conflict exists, boundary correction is initiated. During correction, after acquiring the heterogeneous data, its original timestamps are preserved, and a unified time reference axis is established, but resampling is not performed. To ensure the integrity of the risk events, the algorithm initiates an adaptive adjustment procedure, shifting boundary B2 two sampling points in the positive direction of the time axis. Based on the corrected boundary, the three-domain coupled state nodes are reconstructed; the updated risk trajectory is output. After adjustment, the expression of the inflammation continuation segment in the same direction within stage D3 tends to be complete, and the energy measure of the three-domain coupled nodes is significantly improved.

[0141] Before correction, due to the segmentation by B2, the fragment was judged to be fragmented and fluctuating, with a POD risk probability assessment value of 0.45. After correction, the feature integrity was restored, and the system captured a coherent "inhibition-perfusion-inflammation" abnormal trajectory, accurately correcting the POD risk probability to 0.88. Finally, the updated risk trajectory was output, and the composite risk fragment was prominently marked on the monitoring interface, successfully triggering a clinical alert.

[0142] 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 assessing the risk of postoperative delirium based on electroencephalographic biomarkers, characterized in that, Includes the following steps: Obtain EEG complexity sequences, hemodynamic sequences, and inflammatory marker sequences during general anesthesia surgery, and arrange them in chronological order; The direction of value change of adjacent sampling points in the EEG complexity sequence is compared point by point. The sampling point segments with consistent direction and continuous time are divided into symbol duration segments, and the position of direction reversal between adjacent symbol duration segments is determined as the boundary of the initial neural inhibition stage. Within the time interval defined by the boundaries of two adjacent initial neural inhibition phases, identify perfusion reversal structures in the hemodynamic sequence that have reversed direction and whose start and end times are located within the same symbol duration. The boundaries of the initial neural inhibition phases that do not fall within the time interval of any perfusion reversal structure are determined as the verified neural inhibition phase boundaries. Based on the verified neural inhibition stage boundary identification inflammatory marker sequence, a co-directional continuation segment that crosses the boundary of an adjacent stage and whose direction of change is consistent with the corresponding symbol duration segment is constructed when the perfusion reverse inflection structure is located within a single stage and the co-directional continuation segment crosses the boundary of that stage; the three-domain coupled state node includes: The neural inhibition stage number, the start and end times of the perfusion reverse inflection structure, and the start and end times of the same-direction continuation segment are used as the main index to group and sequentially connect the state nodes. The system detects and identifies time segments in which nested structures persist in two or more consecutive phases as composite risk segments. When the phase boundary time position falls within the composite risk segment, the verified neural inhibition phase boundary is moved and the state node is reconstructed to output the postoperative delirium risk trajectory.

2. The method for assessing postoperative delirium risk based on electroencephalographic biomarkers according to claim 1, characterized in that, The division of the symbol duration includes: The EEG complexity sampling points are numbered sequentially by time as the kth sampling point and the (k+1)th sampling point. The complexity values ​​of the (k+1)th sampling point and the kth sampling point are compared, and the comparison result is recorded as the directional attribute of the (k+1)th sampling point. When multiple consecutive sampling points have the same directional properties and are continuous in time, the set of sampling points is determined as a symbol duration segment, and the start and end times of the symbol duration segment are recorded.

3. The method for assessing postoperative delirium risk based on electroencephalographic biomarkers according to claim 2, characterized in that, The determination of the boundary of the initial neural inhibition phase also includes: At the boundary of each candidate neural inhibition phase, detect whether there is temporal overlap or sampling discontinuity between adjacent symbol durations; When sampling interruptions exist, the boundary of the candidate neural inhibition phase is moved to the nearest valid sampling point time position.

4. The method for assessing postoperative delirium risk based on electroencephalographic biomarkers according to claim 1, characterized in that, The identification of the infusion reverse inversion structure includes: Extracting continuous segments of hemodynamic change with consistent direction of change from within a single initial neural inhibition phase; When a reversal occurs between adjacent continuous change segments and the start and end times of the reversal segment are both within the time range of the same symbol duration, the reversal segment is identified as a reversal structure in the infusion process, and its start and end times are recorded.

5. The method for assessing postoperative delirium risk based on electroencephalographic biomarkers according to claim 4, characterized in that, The determination of the verified boundaries of the neural inhibition phase includes: The temporal position of the boundary of each initial neural inhibition phase is compared with the start and end times of all perfusion reversal inflection structures; When the time position falls within the time segment of any perfusion reverse inflection structure, the initial neural inhibition stage boundary is removed from the verified neural inhibition stage boundary set.

6. The method for assessing postoperative delirium risk based on electroencephalographic biomarkers according to claim 1, characterized in that, The identification of the same-direction continuation segment includes: Extract segments of inflammatory marker sequences that exhibit consistent trends and continuous temporal variation. When the start time of the change segment is later than the end time of a certain perfusion reverse inflection structure, and its time segment crosses the boundary of the adjacent verified neural inhibition stage, and its change direction is consistent with the corresponding symbol duration segment, the change segment is determined as a co-directional continuation segment.

7. The method for assessing postoperative delirium risk based on electroencephalographic biomarkers according to claim 6, characterized in that, The determination of the composite risk segment includes: When there are three-domain coupled state nodes in two or more consecutive verified neural inhibition phases, where the perfusion reverse inflection structure is located inside the phase and the same-direction continuation segment crosses the phase boundary, the time interval between the start boundary time and the end boundary time of the consecutive phases is determined as the composite risk segment.

8. The method for assessing postoperative delirium risk based on electroencephalographic biomarkers according to claim 7, characterized in that, The shift in the verified boundary of the neural inhibition phase includes: When the temporal position of the boundary of the neural inhibition phase falls between the start and end times of the composite risk segment, The boundary of the neural inhibition phase is moved to the start or end time position of the composite risk segment, and the perfusion reverse inflection structure identification and the same-direction continuation segment identification are re-executed based on the moved phase boundary.

9. A postoperative delirium risk assessment system based on electroencephalogram (EEG) biomarkers, comprising the implementation of a postoperative delirium risk assessment method based on EEG biomarkers as described in any one of claims 1-8, characterized in that, include: The data acquisition module is used to acquire EEG complexity sequences, hemodynamic sequences, and inflammatory marker sequences during general anesthesia surgery, and retain the original timestamps of each sequence; The phase division module is used to compare the direction of value change between adjacent sampling points based on the EEG complexity sequence to form a set of directional attributes, and divide the sampling point segments with consistent directional attributes and continuous time into symbol duration segments. The initial neural inhibition phase boundary is determined based on the directional reversal position between adjacent symbol duration segments. The perfusion verification module is used to identify perfusion reversal structures in the hemodynamic sequence within the time interval defined between two adjacent initial neural inhibition phase boundaries, where the direction is reversed and the start and end times are both within the same symbol duration. The initial neural inhibition phase boundaries that do not fall within the time interval of any perfusion reversal structure are determined as the verified neural inhibition phase boundaries. The inflammation coupling module is used to identify the same-direction continuation segment in the inflammatory index sequence that crosses the boundary of adjacent neural inhibition stages and whose change direction is consistent with the corresponding symbol duration segment based on the verified neural inhibition stage boundary. It also constructs a three-domain coupling state node when the perfusion reverse turning structure is located inside a single neural inhibition stage and the same-direction continuation segment crosses the boundary of that stage. The risk identification and correction module is used to detect the time segments in which nested structures persist in two or more consecutive neural inhibition phases in the three-domain coupled state node and identify composite risk segments. The start and end times of the composite risk segments are mapped to the original time axes of the hemodynamic sequence and the inflammatory marker sequence. When there is a positional conflict between the composite risk segment and the verified neural inhibition phase boundary, the phase boundary is adjusted and the three-domain coupled state node is reconstructed to output the postoperative delirium risk trajectory.

Images