CN-121987160-A - Multi-parameter monitoring intelligent early warning method and system for cerebral infarction after vascular intervention operation

CN121987160ACN 121987160 ACN121987160 ACN 121987160ACN-121987160-A

Abstract

The invention relates to the technical field of medical monitoring, and provides a multi-parameter monitoring blood vessel interventional post-operation brain stem intelligent early warning method which comprises the steps of modeling a brain stem evolution path through a preset post-operation brain stem evolution state model to generate a corresponding post-operation brain stem evolution hypothesis, analyzing the current brain stem evolution state according to the post-operation brain stem evolution hypothesis to generate corresponding acquisition control parameters, acquiring multi-mode physiological signals through the acquisition control parameters, preprocessing data, constructing a multi-mode feature vector through a feature splicing mode, inputting a brain stem evolution consistency assessment model, outputting brain stem risk grades and triggering corresponding grading early warning information. By introducing a multi-mode physiological signal acquisition mechanism, a monitored object of a patient after vascular intervention operation is expanded from a traditional single or small number of vital signs to multi-parameter information comprising electrocardio, electroencephalogram, hemodynamics, movement functions and the like, and continuous evaluation and early warning of cerebral infarction risks are realized by combining a postoperative cerebral infarction evolution state model.

Inventors

LI DIE
ZHONG WEIJIE
SUN ZHAOLIANG
ZHOU LINGLING

Assignees

上海交通大学医学院附属第九人民医院

Dates

Publication Date: 20260508
Application Date: 20260130

Claims (10)

1. An intelligent early warning method for a cerebral infarction after a vascular intervention operation with multi-parameter monitoring is characterized by comprising the following steps: Step S1, modeling a cerebral infarction evolution path through a preset postoperative cerebral infarction evolution state model based on operation information, an intraoperative state and basic pathological characteristics of a patient, and generating a corresponding postoperative cerebral infarction evolution hypothesis; S2, analyzing the current cerebral infarction evolution state according to the postoperative cerebral infarction evolution hypothesis, generating corresponding acquisition control parameters, and controlling the intelligent monitoring chest plaster, the intelligent monitoring wrist strap, the head nerve function monitoring equipment and the limb movement monitoring equipment worn by the patient to acquire multi-mode physiological signals and perform data preprocessing through the acquisition control parameters, wherein the acquisition control parameters at least comprise acquisition start-stop parameters, acquisition frequency parameters and data weight parameters; Step S3, extracting key features from the preprocessed multi-mode physiological signals, and constructing the key features into multi-mode feature vectors in a feature splicing mode; S4, inputting the multi-modal feature vector into a brain stem evolution consistency assessment model, and calculating a consistency score between the current monitoring feature and the current brain stem evolution state based on a Bayesian likelihood function, wherein the consistency score is used for judging whether the current physiological change of the patient accords with the brain stem evolution trend; And S5, confirming or correcting the evolution state of the cerebral infarction according to the comparison result of the consistency score and the preset condition, outputting the risk level of the cerebral infarction based on the time sequence prediction model when the consistency score accords with the preset condition, and triggering corresponding hierarchical early warning information.
2. The intelligent pre-warning method for cerebral infarction after vascular intervention with multi-parameter monitoring as set forth in claim 1, wherein step S1 includes: Step S11, obtaining the operation information, the operation state and the initial key feature after basic pathological features of the patient, carrying out feature coding and numerical processing, and constructing a basic feature vector for modeling cerebral infarction evolution, wherein the operation information at least comprises operation type, intervention part and operation duration, the operation state at least comprises blood pressure fluctuation condition, blood flow blocking condition and operation concurrent event identification, and the basic pathological features at least comprise past cerebrovascular disease history, hypertension or diabetes history; Step S12, inputting the basic feature vector into a hidden Markov model based on state transition probability, and calculating the state transition probability among different brain stem evolution states through model parameters or initialization parameters obtained through historical sample training; and S13, selecting the cerebral infarction evolution state sequence with occurrence probability meeting the preset confidence coefficient condition from the candidate cerebral infarction evolution state sequence as the postoperative cerebral infarction evolution hypothesis.
3. The intelligent pre-warning method for cerebral infarction after vascular intervention with multi-parameter monitoring as set forth in claim 2, wherein step S12 includes: Calculating initial state probability distribution of the patient in a preset brain stem evolution state set based on the basic feature vector, wherein the initial state probability distribution is the possibility that the patient is in each brain stem evolution state at the initial moment after operation; Constructing a brain stem evolution state transition probability model according to the state transition probability between any two adjacent states in the brain stem evolution state set by a state transition matrix in the hidden Markov model; generating a plurality of candidate brain stem evolution state sequences by adopting a sequence deduction algorithm based on the initial state probability distribution and the brain stem evolution state transition probability model; Calculating corresponding joint occurrence probabilities for each candidate brain stem evolution state sequence, and taking the joint occurrence probabilities as the occurrence probabilities of the candidate brain stem evolution state sequences.
4. The intelligent pre-warning method for cerebral infarction after vascular intervention with multi-parameter monitoring as set forth in claim 1, wherein step S2 includes: Analyzing the current cerebral infarction evolution state of the patient based on the postoperative cerebral infarction evolution hypothesis, determining the type of physiological signals to be acquired according to the cerebral infarction evolution state, and distributing corresponding acquisition control parameters for each physiological signal; The acquisition control parameters are used for acquiring the multi-mode physiological signals by controlling an intelligent monitoring chest patch, an intelligent monitoring wrist strap, a head nerve function monitoring device and a limb movement monitoring device, wherein the intelligent monitoring chest patch acquires electrocardiosignals, blood oxygen saturation and blood flow dynamics parameters, the intelligent monitoring wrist strap acquires continuous blood pressure, pulse pressure difference and blood flow perfusion, the head nerve function monitoring device acquires electroencephalogram signals, eye movement data and sound analysis data, and the limb movement monitoring device acquires movement function data, joint angles, tactile feedback and limb movement modes; preprocessing the acquired multi-mode physiological signals, and specifically comprises denoising, standardization, time alignment and abnormal data processing.
5. The intelligent pre-warning method for cerebral infarction after vascular intervention based on multi-parameter monitoring as set forth in claim 1, wherein the step S4 includes: inputting the preprocessed multi-modal feature vector into the cerebral infarction evolution consistency assessment model; the brain stem evolution consistency evaluation model calculates a Bayesian likelihood function value corresponding to each physiological feature according to the current multi-modal feature vector and the model parameters obtained by previous training, wherein the Bayesian likelihood function value is used for quantifying the consistency of each feature and the current brain stem evolution state; and carrying out weighted summarization on the Bayesian likelihood function values of each physiological characteristic to generate a comprehensive consistency score, wherein the consistency score is used for judging whether the current physiological change of the patient accords with the evolution trend of the brain peduncles.
6. The intelligent early warning method for cerebral infarction after vascular intervention with multi-parameter monitoring according to claim 5, wherein in step S5, the evolution state of cerebral infarction is confirmed or corrected according to the comparison result of the consistency score and a preset condition, and when the consistency score meets the preset condition, a cerebral infarction risk level is output based on a time sequence prediction model, and corresponding hierarchical early warning information is triggered, comprising: Judging whether the consistency score is higher than a preset threshold value or not; when the consistency score is lower than the preset threshold, updating the cerebral infarction evolution state according to the subsequently acquired multi-mode physiological signals, and recalculating the consistency score for judging whether the consistency score is higher than the preset threshold; When the consistency score is higher than or equal to the preset threshold, the current physiological change accords with the evolution trend of the cerebral infarction, and the corresponding cerebral infarction risk grade and the corresponding grading early warning information are output based on a time sequence prediction model.
7. The intelligent pre-warning method for cerebral infarction after a multi-parameter monitoring vascular intervention operation according to claim 6, wherein updating the cerebral infarction evolution state according to the subsequently acquired multi-modal physiological signals, recalculating the consistency score for judging whether the consistency score is higher than the preset threshold value comprises inputting the acquired multi-modal feature vector into the hidden markov model in combination with the initial key feature to update the corresponding state transition probability, acquiring the corresponding cerebral infarction evolution state, and judging the consistency score.
8. The intelligent pre-warning method for cerebral infarction after vascular intervention based on multi-parameter monitoring as set forth in claim 6, wherein outputting the corresponding cerebral infarction risk level and the corresponding hierarchical pre-warning information based on a time-series prediction model includes: transmitting the multi-modal feature vector as input into a pre-trained long-short-time memory network model, performing serialization analysis, calculating and predicting the cerebral infarction risk level of the patient, and performing risk assessment based on the multi-modal physiological signal; and automatically generating corresponding grading early warning information according to the cerebral infarction risk grade and the risk assessment result output by the long-short-term memory network model.
9. The intelligent pre-warning method for cerebral infarction after vascular intervention with multi-parameter monitoring as set forth in claim 6, wherein the key characteristic data comprises, Based on the heart rate value, heart rate variability parameter, ST segment offset and QT interval parameter extracted from the electrocardiosignal acquired by the intelligent detection chest patch, the blood oxygen saturation mean value of the blood oxygen saturation and the blood oxygen fluctuation amplitude parameter, and the pulse waveform characteristic parameter and the peripheral blood flow perfusion index of the hemodynamic parameter; based on the systolic pressure value, the diastolic pressure value and the pulse pressure difference of the continuous blood pressure acquired by the intelligent monitoring wrist strap, peripheral blood flow perfusion intensity parameters and blood flow perfusion change rate parameters of the blood flow perfusion; Based on the preset frequency band brain electricity power value and brain electricity rhythm distribution parameter acquired by the head nerve function monitoring equipment, the gaze direction offset and gaze duration parameter extracted by the eye movement data, and the voice sounding duration parameter and voice break time parameter of the voice analysis data; Based on the limb movement amplitude parameter and the movement frequency parameter of the movement function data acquired by the limb movement monitoring equipment, the joint movement angle value and the angle change rate parameter of the joint angle, the pressure response value of the tactile feedback and the tactile stimulus response duration parameter, and the limb movement time sequence distribution parameter of the limb movement mode.
10. A multiparameter monitored intelligent early warning system for a post-interventional vascular procedure cerebral infarction, which adopts the multiparameter monitored intelligent early warning method for the post-interventional vascular procedure cerebral infarction according to any one of claims 1 to 9, and is characterized by comprising the following steps: The cerebral infarction evolution modeling module is used for modeling a cerebral infarction evolution path through a preset postoperative cerebral infarction evolution state model based on operation information, an intraoperative state and basic pathological characteristics of a patient, and generating a corresponding postoperative cerebral infarction evolution hypothesis; The data acquisition module is used for analyzing the current cerebral infarction evolution state according to the postoperative cerebral infarction evolution hypothesis, generating corresponding acquisition control parameters, and controlling the intelligent monitoring chest plaster, the intelligent monitoring wrist strap, the head nerve function monitoring equipment and the limb movement monitoring equipment worn by the patient to acquire multi-mode physiological signals and perform data preprocessing through the acquisition control parameters, wherein the acquisition control parameters at least comprise acquisition start-stop parameters, acquisition frequency parameters and data weight parameters; The feature construction module is used for extracting key features from the preprocessed multi-mode physiological signals and constructing the key features into multi-mode feature vectors in a feature splicing mode; The evaluation module is used for inputting the multi-modal feature vector into a brain stem evolution consistency evaluation model, calculating a consistency score between the current monitoring feature and the current brain stem evolution state based on a Bayesian likelihood function, and judging whether the current physiological change of the patient accords with the brain stem evolution trend or not; And the prediction module is used for confirming or correcting the evolution state of the cerebral infarction according to the comparison result of the consistency score and the preset condition, outputting the risk level of the cerebral infarction based on the time sequence prediction model when the consistency score accords with the preset condition, and triggering corresponding grading early warning information.

Description

Multi-parameter monitoring intelligent early warning method and system for cerebral infarction after vascular intervention operation Technical Field The invention relates to the technical field of medical monitoring, in particular to an intelligent early warning method and system for cerebral infarction after vascular intervention by multi-parameter monitoring. Background The vascular intervention treatment is used as an important treatment means for cerebrovascular diseases, and is widely applied to the clinical treatment of acute cerebral apoplexy, intracranial aneurysms, arteriovenous malformations and other diseases. The treatment mode has the advantages of small wound, quick recovery and the like, but in the early stage of operation, serious complications such as cerebral infarction and the like can still occur to patients, the complications tend to have hidden diseases and rapid progress, and if the complications cannot be found and intervened in time, the nerve function damage is easily caused, and the prognosis of the patients is influenced. At present, a monitoring mode after vascular intervention mainly depends on manual ward-round, conventional vital sign monitoring and staged nerve function evaluation, and is assisted with imaging examination if necessary. Most of the monitoring modes are implemented intermittently, monitoring continuity is insufficient, and slight changes of the postoperative physiological state of a patient are difficult to capture in time. Meanwhile, the existing monitoring parameters are mainly concentrated on basic vital signs such as blood pressure, heart rate, blood oxygen saturation and the like, and comprehensive monitoring of indexes closely related to cerebral infarction such as brain electrical activity, brain perfusion state, limb movement function and the like is lacked. In addition, the existing monitoring equipment mostly adopts a fixed threshold alarming mode, lacks a risk assessment and grading early warning mechanism based on multi-parameter fusion analysis, and is difficult to adapt to individual differences and dynamic changes of illness states of patients. Although the wearable monitoring technology is applied to partial medical scenes, the existing equipment is multifunctional, a systematic monitoring and evaluating scheme for the risk of cerebral infarction after vascular intervention operation is not formed yet, and the actual requirements of continuous monitoring and early warning of the risk of cerebral infarction in early stage after operation are difficult to meet. Disclosure of Invention Aiming at the technical problems that the prior vascular interventional operation monitoring process depends on manual observation, has limited monitoring parameter dimension, insufficient risk assessment accuracy, poor early warning timeliness and the like, the invention provides a multi-parameter monitoring vascular interventional operation post-cerebral infarction intelligent early warning method and system, and the dynamic assessment and grading early warning of post-operation cerebral infarction risk are realized by continuously collecting and fusing and analyzing multi-source physiological parameters of patients, so that the early recognition capability and clinical intervention timeliness of cerebral infarction complications are improved, and the invention can be realized by the following technical scheme: The invention provides an intelligent early warning method for cerebral infarction after vascular intervention operation with multi-parameter monitoring, which comprises the following steps: Step S1, modeling a cerebral infarction evolution path through a preset postoperative cerebral infarction evolution state model based on operation information, an intraoperative state and basic pathological characteristics of a patient, and generating a corresponding postoperative cerebral infarction evolution hypothesis; S2, analyzing the current cerebral infarction evolution state according to the postoperative cerebral infarction evolution hypothesis, generating corresponding acquisition control parameters, and controlling the intelligent monitoring chest patch, the intelligent monitoring wrist strap, the head nerve function monitoring equipment and the limb movement monitoring equipment worn by a patient to acquire multi-mode physiological signals and perform data preprocessing through the acquisition control parameters, wherein the acquisition control parameters at least comprise acquisition start-stop parameters, acquisition frequency parameters and data weight parameters; Step S3, extracting key features from the preprocessed multi-mode physiological signals, and constructing the key features into multi-mode feature vectors in a feature splicing mode; s4, inputting the multi-mode feature vector into a brain stem evolution consistency assessment model, calculating consistency scores between current monitoring features and current brain stem evolution states based