An evaluation and early warning method for postoperative nursing scheme of free skin flap based on multi-modal monitoring data fusion

Abstract

The application discloses a kind of free skin flap postoperative nursing scheme evaluation and early warning method based on multi-modal monitoring data fusion, it is related to intelligent medical care technical field, for solving the problem that existing monitoring cannot distinguish vascular pathological crisis and nursing environment interference.First, a multi-dimensional time-varying state matrix containing physiological drive and nursing environment characteristics is constructed, realizing the time sequence alignment of heterogeneous data;Subsequently, the pressure-flow conduction characteristics are analyzed, and the blood flow conduction efficiency index reflecting the inherent patency of blood vessels is extracted;At the same time, a physical gain mapping is established based on nursing parameters, and the theoretical target perfusion trajectory under the current nursing state is deduced;Finally, the dynamic residual vector is calculated and combined with the conduction efficiency index for joint determination, accurately identifying vascular structural occlusion and functional inhibition of nursing program, achieving accurate tracing of postoperative crisis and dynamic evaluation of nursing program.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical monitoring technology, specifically to a method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion. Background Technology

[0002] Free flap transplantation is a core technique in modern microsurgery for repairing severe tissue defects and reconstructing organs. Its success rate highly depends on the stability of postoperative microcirculatory blood supply. During the early postoperative window of high incidence of vascular crisis, the flap tissue is extremely sensitive to ischemia and hypoxia. Even minor anastomotic thrombosis or obstruction of venous return, if not detected in time, can lead to irreversible necrosis of the flap. Therefore, continuous monitoring of the flap's color, temperature, and capillary refill response postoperatively is crucial for ensuring quality nursing care and patient recovery, and is of paramount importance in clinical microsurgical nursing.

[0003] Existing flap monitoring methods largely rely on absolute threshold alarms for single physical indicators, such as simply monitoring skin temperature or blood flow values. This monitoring model has substantial flaws in clinical application, leading to potential misjudgments. Because the microcirculation perfusion of a flap depends not only on the patency of local blood vessels but is also significantly affected by external factors such as fluctuations in the patient's systemic blood pressure, changes in ambient temperature, and the angle of the affected limb, current methods cannot identify these multi-source interferences. They often misreport physiological slowing of blood flow caused by decreased blood pressure or cold environments as a vascular crisis, leading to frequent and ineffective investigations by healthcare professionals. More seriously, existing monitoring equipment cannot assess the appropriateness of current nursing interventions. When blood flow decreases due to improper positioning compressing blood vessels, the equipment can only issue a low perfusion alarm but cannot distinguish whether this is due to functional inhibition caused by improper care or structural embolism within the blood vessel, resulting in unclear clinical decision-making and delays in optimal intervention. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion, thus solving the problems mentioned in the background.

[0005] To achieve the above objectives, this invention provides the following technical solution: a method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion, comprising the following steps: S1. Simultaneously acquiring the whole-body mean arterial pressure sequence and the flap local microcirculation velocity sequence to form an endogenous physiological load set; simultaneously acquiring the affected limb posture and tilt sequence and the local environmental temperature and humidity sequence to form an exogenous nursing intervention set; selecting the flow velocity of the symmetrical part of the healthy side as a benchmark; performing differential normalization and temporal alignment on the endogenous physiological load set to construct a multidimensional time-varying state matrix containing physiological driving characteristics and nursing environment characteristics; S2. Performing cross-correlation calculations on the multidimensional time-varying state matrix; extracting the phase lag time and cross-correlation coefficient of the whole-body mean arterial pressure sequence fluctuation transmitted to the flap local microcirculation velocity sequence; performing phase compensation based on the phase lag time; calculating the flow velocity response amplitude ratio under unit pressure change; and combining... S3. Establish a mapping relationship between the exogenous nursing intervention set and the flow velocity gain, map the limb posture tilt sequence to the gravitational potential energy compensation coefficient, and map the local environmental temperature and humidity sequence to the vascular tension compensation coefficient. Modulate the whole-body mean arterial pressure sequence through the gravitational potential energy compensation coefficient and the vascular tension compensation coefficient to calculate the target perfusion trajectory under the current nursing state. S4. Calculate the numerical difference between the local microcirculation velocity sequence of the flap and the target perfusion trajectory to obtain the dynamic residual vector. Jointly determine the dynamic residual vector and the blood flow conduction efficiency index: if the dynamic residual vector is lower than the preset negative threshold and the blood flow conduction efficiency index is lower than the preset occlusion threshold, output the vascular structural occlusion signal; if the dynamic residual vector is lower than the preset negative threshold and the blood flow conduction efficiency index is located in the preset connected interval, output the functional inhibition signal of the nursing plan.

[0006] Furthermore, the specific process of simultaneously acquiring the whole-body mean arterial pressure sequence and the local microcirculation velocity sequence of the flap to form an endogenous physiological load set, and simultaneously acquiring the affected limb posture and tilt angle sequence and the local environmental temperature and humidity sequence to form an exogenous nursing intervention set is as follows: the mean systolic and diastolic pressure integrals of the continuous arterial pressure waveform are extracted through cardiac cycle gating sampling to construct the whole-body mean arterial pressure sequence; the spatiotemporal contrast analysis of the free flap area and the symmetrical area on the healthy side is performed using a laser speckle blood flow imaging system to quantify and generate the local microcirculation velocity sequence of the flap; the spatial Euler angle of the affected limb relative to the vertical line of gravity is calculated using an inertial measurement unit, and the pitch angle component is extracted as the affected limb posture and tilt angle sequence; the thermal radiation flux of the flap surface and the surrounding environment is captured by a non-contact thermal imaging array to invert and generate the local environmental temperature and humidity sequence.

[0007] Furthermore, using the flow velocity of the symmetrical part of the healthy side as a benchmark, the process of constructing a multidimensional time-varying state matrix containing physiological driving characteristics and nursing environment characteristics by differential normalization and temporal alignment of the endogenous physiological load set is as follows: The flow velocity of the symmetrical part of the healthy side is used as the individual benchmark background flow velocity. Background difference operation is performed on the local microcirculation flow velocity sequence of the flap to remove individual basal metabolic drift and obtain the net flow velocity sequence. The frequency resampling of the limb posture tilt sequence and the local environmental temperature and humidity sequence is performed using a multi-rate signal synchronization algorithm. The temporal resolution is unified to the sampling frequency of the whole body mean arterial pressure sequence. Using the time axis of the whole body mean arterial pressure sequence as a benchmark, the whole body mean arterial pressure sequence, the net flow velocity sequence and the resampled exogenous nursing intervention set are spliced ​​in the feature dimension to generate a multidimensional time-varying state matrix.

[0008] Furthermore, the specific process of performing cross-correlation operations on the multidimensional time-varying state matrix to extract the phase lag time and cross-correlation coefficient of the transmission of the whole-body mean arterial pressure sequence fluctuation to the local microcirculation velocity sequence of the flap is as follows: the whole-body mean arterial pressure sequence segment and the local microcirculation velocity sequence segment of the flap are extracted from the multidimensional time-varying state matrix by sliding time window, and the cross-correlation function of the two sequence segments is constructed. The global maximum point of the cross-correlation function is retrieved within the preset physiological conduction delay domain. The time delay corresponding to the global maximum point is locked as the phase lag time, and the normalized amplitude of the global maximum point is locked as the cross-correlation coefficient.

[0009] Furthermore, based on phase lag time, phase compensation is performed, and the velocity response amplitude ratio under unit pressure change is calculated. Combining the cross-correlation coefficient and the velocity response amplitude ratio, the specific process of obtaining the blood flow conduction efficiency index is as follows: A phase lag time shift is applied to the whole-body mean arterial pressure sequence to eliminate the time-domain delay of pressure wave transmission to peripheral blood vessels, thereby achieving phase synchronization of pressure and flow velocity waveforms; the effective value ratio of the phase-synchronized whole-body mean arterial pressure sequence and the local microcirculation flow velocity sequence of the flap is calculated as the velocity response amplitude ratio, the cross-correlation coefficient is used as the waveform similarity weight, and the velocity response amplitude ratio is used as the conduction gain weight. The blood flow conduction efficiency index, which characterizes the patency of the vascular bed, is obtained by fusion calculation.

[0010] Furthermore, the mapping relationship between the exogenous nursing intervention set and the flow rate gain was established. The specific process of mapping the limb posture tilt sequence to the gravitational potential energy compensation coefficient and the local environmental temperature and humidity sequence to the vascular tension compensation coefficient is as follows: The vertical component in the limb posture tilt sequence was analyzed, and the hydrostatic pressure gradient change caused by the change in body position was calculated by combining the principles of hydrostatics. This was then normalized and mapped to the gravitational potential energy compensation coefficient. A nonlinear response curve of local environmental temperature and humidity and vascular smooth muscle tension was constructed. The real-time acquired local environmental temperature and humidity sequence was projected onto this response curve, and a dimensionless factor characterizing vasomotor activity was output as the vascular tension compensation coefficient.

[0011] Furthermore, the specific process of calculating the target perfusion trajectory under the current nursing state by modulating the whole-body mean arterial pressure sequence using the gravitational potential energy compensation coefficient and the vascular tension compensation coefficient is as follows: The whole-body mean arterial pressure sequence is amplitude corrected using the gravitational potential energy compensation coefficient to restore the effective perfusion pressure after eliminating gravitational potential energy interference; the vascular tension compensation coefficient is used as a dynamic gain control variable to modulate the effective perfusion pressure and simulate the theoretical flow velocity response under the current vascular tension; the average flow velocity of the symmetrical part on the healthy side is used as the reference conductivity, and the modulated theoretical flow velocity response is mapped to the flow velocity domain to generate the target perfusion trajectory.

[0012] Furthermore, the numerical difference between the local microcirculation velocity sequence of the flap and the target perfusion trajectory is calculated to obtain the dynamic residual vector. The logical process of jointly determining the dynamic residual vector and the blood flow conduction efficiency index is as follows: extract the instantaneous deviation amplitude of the local microcirculation velocity sequence of the flap relative to the target perfusion trajectory, and the gradient direction of the deviation amplitude changing over time; construct a state feature vector composed of the deviation amplitude and the gradient direction as the dynamic residual vector; establish a two-dimensional state classification plane containing structural blockage area and functional inhibition area, and project the dynamic residual vector and the blood flow conduction efficiency index as coordinate parameters onto the classification plane for region matching.

[0013] Furthermore, if the dynamic residual vector is lower than a preset negative threshold and the blood flow conduction efficiency index is lower than a preset blocking threshold, a vascular structural occlusion signal is output; if the dynamic residual vector is lower than a preset negative threshold and the blood flow conduction efficiency index is located within a preset connected interval, a nursing plan functional inhibition signal is output. The specific process is as follows: the instantaneous amplitude of the dynamic residual vector is detected. When the instantaneous amplitude is less than a preset negative threshold and the corresponding blood flow conduction efficiency index is lower than a preset blocking threshold, a vascular structural occlusion signal is generated; when the instantaneous amplitude is less than a preset negative threshold, but the corresponding blood flow conduction efficiency index is higher than a preset blocking threshold and is within a preset connected interval, a nursing plan functional inhibition signal is generated.

[0014] The present invention has the following beneficial effects:

[0015] (1) A method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion. This method constructs a multidimensional time-varying state matrix that includes physiological driving characteristics and nursing environment characteristics, and uses cross-correlation calculations to extract the phase lag time and cross-correlation coefficient of the transmission of systemic mean arterial pressure fluctuations to the local microcirculation velocity of the flap. This effectively solves the problem that single-indicator monitoring is easily interfered with by systemic physiological fluctuations. This method calculates the velocity response amplitude ratio under unit pressure change and obtains the blood flow conduction efficiency index by combining the cross-correlation coefficient, thus realizing the separation of the inherent mechanical patency characteristics of the vascular bed from complex physiological signals. This means that the system can identify the passive response capability of flap vessels when systemic blood pressure fluctuates, thereby accurately eliminating false alarms caused by changes in the patient's baseline blood pressure, and ensuring the specificity and accuracy of the assessment of the inherent patency status of the vessels.

[0016] (2) A method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion. By establishing a mapping relationship between exogenous nursing intervention sets and flow velocity gain, the method quantifies the limb position and tilt angle and local environmental temperature and humidity into specific physical compensation coefficients, and deduces the target perfusion trajectory under the current nursing state, thus achieving dynamic evaluation of the effectiveness of the nursing plan. This method can accurately distinguish between vascular structural occlusion and functional inhibition of the nursing plan by jointly determining the dynamic residual vector and the blood flow conduction efficiency index. When the blood flow conduction efficiency is maintained at a high level but the actual flow velocity is lower than the target trajectory, the system can clearly determine that it is inhibited by nursing environmental factors, thereby guiding medical staff to optimize the body position or adjust the room temperature, rather than blindly performing surgical exploration, significantly improving the precision of postoperative care and the efficiency of clinical decision-making.

[0017] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0018] Figure 1 This is a flowchart of a method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion, as described in this invention. Detailed Implementation

[0019] This application provides an evaluation and early warning method for postoperative care of free flaps based on multimodal monitoring data fusion. This method addresses the problem that existing monitoring technologies cannot distinguish between vascular pathological crises and nursing environment-induced hypoperfusion, leading to high false alarm rates and an inability to guide nursing decisions.

[0020] The overall concept of the solution in this application embodiment is as follows:

[0021] First, the system simultaneously collects the patient's endogenous physiological data and exogenous nursing environment data. Using the data from the healthy side as a benchmark, the data is normalized and time-series aligned to construct a multidimensional matrix reflecting the overall system status. Next, by analyzing the process of systemic blood pressure fluctuations being transmitted to the local flap flow velocity, the system extracts a blood flow conduction efficiency index reflecting the physical patency of blood vessels to determine if structural blockage exists. Simultaneously, based on the current limb tilt angle and ambient temperature and humidity, the system calculates the theoretically target perfusion trajectory the flap should achieve under the given nursing conditions using physical mapping relationships. Finally, by comparing the difference between real-time flow velocity and the target trajectory, and combining this with the blood flow conduction efficiency index, the system performs logical classification: if vascular conduction efficiency is significantly reduced, it is determined to be a structural problem such as vascular embolism; if vascular conduction efficiency is normal but flow velocity is below target, it is determined to be functional inhibition caused by improper nursing measures such as body position or temperature, thereby achieving precise graded early warning and nursing plan evaluation.

[0022] Please see Figure 1 This invention provides a technical solution: a method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion, comprising the following steps: S1. Simultaneously acquiring the whole-body mean arterial pressure sequence and the local microcirculation velocity sequence of the flap to form an endogenous physiological load set; simultaneously acquiring the position and tilt sequence of the affected limb and the local environmental temperature and humidity sequence to form an exogenous nursing intervention set; selecting the flow velocity of the symmetrical part of the healthy side as a benchmark; performing differential normalization and temporal alignment on the endogenous physiological load set to construct a multidimensional time-varying state matrix containing physiological driving characteristics and nursing environment characteristics; S2. Performing cross-correlation calculation on the multidimensional time-varying state matrix; extracting the phase lag time and cross-correlation coefficient of the whole-body mean arterial pressure sequence fluctuation transmitted to the local microcirculation velocity sequence of the flap; performing phase compensation based on the phase lag time; calculating the velocity response amplitude ratio under unit pressure change; and combining the cross-correlation coefficient... S3. Establish a mapping relationship between the exogenous nursing intervention set and the flow velocity gain, map the limb posture tilt sequence to the gravitational potential energy compensation coefficient, and map the local environmental temperature and humidity sequence to the vascular tension compensation coefficient. Modulate the whole-body mean arterial pressure sequence through the gravitational potential energy compensation coefficient and the vascular tension compensation coefficient to calculate the target perfusion trajectory under the current nursing state. S4. Calculate the numerical difference between the local microcirculation flow velocity sequence of the flap and the target perfusion trajectory to obtain the dynamic residual vector. Jointly determine the dynamic residual vector and the blood flow conduction efficiency index: if the dynamic residual vector is lower than the preset negative threshold and the blood flow conduction efficiency index is lower than the preset occlusion threshold, output the vascular structural occlusion signal; if the dynamic residual vector is lower than the preset negative threshold and the blood flow conduction efficiency index is located in the preset connected interval, output the functional inhibition signal of the nursing plan.

[0023] In this implementation plan, the main function of step S1 is to construct a standardized multidimensional data foundation, aiming to eliminate individual differences and achieve spatiotemporal alignment of heterogeneous data. In this step, the endogenous physiological load set refers to the internal dynamic data driving the flap's blood circulation, namely the patient's systemic blood pressure and local blood flow; the exogenous nursing intervention set refers to external nursing environment parameters affecting blood flow, namely the placement angle of the affected limb and ambient temperature. By selecting the flow velocity of the symmetrical site on the healthy side as a benchmark for differential normalization, a relative coordinate system is essentially established to eliminate background noise such as the patient's own fever and fluctuations in basal metabolic rate. Constructing a multidimensional time-varying state matrix maps the three types of data—physiological driving force, microcirculatory response, and external environmental disturbances—with different frequencies and dimensions onto the same time axis, providing accurate data input for subsequent analysis of the coupling relationships between various factors. The core of step S2 lies in quantifying the mechanical patency of the blood vessels themselves, that is, assessing the vascular bed's response to systemic blood pressure fluctuations. The phase lag time extracted through cross-correlation calculations in this step physically represents the time delay in the transmission of the pulse wave from the heart to the skin flap. Phase compensation is used to eliminate this time difference during calculation, ensuring that the pressure peak and flow velocity peak are aligned. The blood flow conduction efficiency index is a key output of this step. It is a physical quantity describing the vascular impedance characteristics, obtained by calculating the ratio of the flow velocity response amplitude caused by a unit pressure change. A high index indicates good vascular elasticity and unobstructed pathways, with blood flow fluctuating with blood pressure; a low index indicates physical obstruction of the blood vessel. The technical role of this step is to separate features related only to the patency of vascular structure from complex blood flow signals, eliminating misjudgments of low flow velocity caused solely by hypotension. Step S3 aims to establish a theoretical expectation model based on the current nursing scenario, that is, to calculate the blood flow rate under the current body position and temperature, assuming no vascular obstruction. The gravitational potential energy compensation coefficient quantifies the effect of hydrostatic pressure on blood flow by converting the elevation angle of the affected limb; the vascular tension compensation coefficient quantifies the effect of ambient temperature on the diameter of blood vessels by converting the contraction or relaxation of vascular smooth muscle. The target perfusion trajectory is a theoretical curve generated by modulating systemic blood pressure using the above two coefficients. It represents the ideal blood flow state of a skin flap without pathological obstruction under the current specific nursing plan. The technical value of this step lies in providing a dynamically changing pass / fail line, so that the assessment standard is no longer a rigid, fixed threshold, but a dynamic standard that adapts to changes in the nursing plan. Step S4 is the final intelligent decision-making stage, used to accurately distinguish between pathological vascular crises and functional nursing malpractice. The dynamic residual vector is the difference between the real-time measured value and the theoretical target value calculated in S3, representing the degree of blood flow abnormality.The system combines this discrepancy with the blood flow conduction efficiency index obtained in S2 for joint judgment: when the residual is large and the conduction efficiency is extremely low, it indicates that the blood vessel is not responding to pressure and the physical pathway is interrupted, defined as a signal of structural vascular occlusion, suggesting the need for surgical exploration; when the residual is large but the conduction efficiency is still good, it indicates that the physical pathway of the blood vessel is open, but is inhibited by external factors such as excessively high body position or excessively low room temperature, defined as a signal of functional inhibition of the nursing plan, suggesting that only adjustments to nursing measures are needed. This step solves the pain point of traditional monitoring equipment that only alarms but cannot identify the cause, realizing graded early warning.

[0024] Specifically, the process of simultaneously acquiring the whole-body mean arterial pressure sequence and the local microcirculation velocity sequence of the flap to form an endogenous physiological load set, and simultaneously acquiring the affected limb posture and tilt angle sequence and the local environmental temperature and humidity sequence to form an exogenous nursing intervention set is as follows: The mean systolic and diastolic pressure integrals of the continuous arterial pressure waveform are extracted through cardiac cycle gating sampling to construct the whole-body mean arterial pressure sequence; a laser speckle flow imaging system is used to perform spatiotemporal contrast analysis on the free flap area and the symmetrical area on the healthy side to quantify and generate the local microcirculation velocity sequence of the flap; the spatial Euler angles of the affected limb relative to the vertical line of gravity are calculated using an inertial measurement unit, and the pitch angle component is extracted as the affected limb posture and tilt angle sequence; a non-contact thermal imaging array is used to capture the thermal radiation flux of the flap surface and the surrounding environment to invert and generate the local environmental temperature and humidity sequence.

[0025] In this implementation plan, the core of the acquisition process for the endogenous physiological load set and the exogenous nursing intervention set lies in ensuring the accurate quantification of each physical quantity in both time and space dimensions. When acquiring the systemic mean arterial pressure, to eliminate random errors caused by sampling at a single time point, the system employs the cardiac cycle integration method. This method automatically identifies characteristic points of the arterial pressure waveform to define a complete pumping cycle, and then performs integral averaging of the pressure waveform energy within the cycle. The specific calculation formula for the systemic mean arterial pressure in this step is as follows: ;in, Calculated value of mean arterial pressure throughout the body during the kth cardiac cycle; The complete duration of the Kth cardiac cycle; : A function of instantaneous continuous arterial pressure waveform acquired during the current cardiac cycle; Zero-point drift correction factor for the pressure sensor. This calculation process effectively smooths instantaneous blood pressure fluctuations and extracts the true blood perfusion driving potential energy. For microcirculation velocity, the system quantifies red blood cell migration rate using the spatiotemporal statistical characteristics of laser speckle; for pose tilt angle, Euler angles are calculated using the projection relationship of the gravitational acceleration component sensed by the inertial measurement unit in three-dimensional space; for environmental parameters, the thermodynamic temperature of the target surface is retrieved based on thermal radiation flux, thus completing the low-level capture of multimodal raw data.

[0026] Specifically, the process of constructing a multidimensional time-varying state matrix containing physiological driving features and nursing environment features is as follows: using the flow velocity of the symmetrical part of the healthy side as the benchmark, differential normalization and temporal alignment are performed on the endogenous physiological load set. The flow velocity of the symmetrical part of the healthy side is used as the individual benchmark background flow velocity. Background difference operation is performed on the local microcirculation flow velocity sequence of the flap to remove individual basal metabolic drift and obtain the net flow velocity sequence. The frequency resampling of the limb posture tilt sequence and the local environmental temperature and humidity sequence is performed using a multi-rate signal synchronization algorithm. The temporal resolution is unified to the sampling frequency of the whole body mean arterial pressure sequence. Using the time axis of the whole body mean arterial pressure sequence as the benchmark, the whole body mean arterial pressure sequence, the net flow velocity sequence and the resampled exogenous nursing intervention set are spliced ​​in the feature dimension to generate a multidimensional time-varying state matrix.

[0027] In this implementation scheme, the key focus in constructing the multidimensional time-varying state matrix is ​​eliminating individual physiological background noise and addressing the frequency misalignment problem of multi-source heterogeneous data. In the differential normalization step, the system does not directly use the absolute flow velocity value of the flap, but instead introduces data from the healthy side as a dynamic reference benchmark. By constructing a differential model, interference from systemic vasoconstriction caused by patient fever, pain, or mental stress is eliminated, thereby separating the net flow velocity characteristics that only reflect the local microcirculation state of the flap. The calculation formula for this net flow velocity sequence is as follows: ;in, : Normalized net flow velocity value of the skin flap at the nth sampling point; : The original measured flow velocity sequence values ​​of the flap area; Background reference velocity sequence values ​​for the symmetrical region on the healthy side; Bilateral limb blood flow balance coefficient, which is the ratio of the mean flow velocity of the affected side to the mean flow velocity of the unaffected side under the preoperative resting state; : The basal metabolic flow rate variance constant used for normalization. After obtaining the pure flow rate features, to address the issue of low sampling frequencies for pose tilt angle and ambient temperature and humidity, the system employs a multiphase polynomial interpolation filtering algorithm to resample the data, ensuring its time resolution strictly matches the high-frequency blood pressure data. The final multidimensional time-varying state matrix structure is as follows: ;in, The fused multidimensional time-varying state feature vector at time t; The pitch angle of the affected limb after frequency resampling and phase alignment; : The composite characteristic value of local ambient temperature and humidity after frequency resampling and phase alignment; Vector transpose operator. Through the above processing, isomorphic fusion of physiological driving features and nursing environment features in the feature space is achieved.

[0028] Specifically, the process of performing cross-correlation operations on the multidimensional time-varying state matrix to extract the phase lag time and cross-correlation coefficient of the transmission of the whole-body mean arterial pressure sequence fluctuation to the local microcirculation velocity sequence of the flap is as follows: the whole-body mean arterial pressure sequence segment and the local microcirculation velocity sequence segment of the flap are extracted from the multidimensional time-varying state matrix by sliding time window, and the cross-correlation function of the two sequence segments is constructed. The global maximum point of the cross-correlation function is retrieved within the preset physiological conduction delay domain. The time delay corresponding to the global maximum point is locked as the phase lag time, and the normalized amplitude of the global maximum point is locked as the cross-correlation coefficient.

[0029] In this implementation scheme, the core of the dynamic analysis of the transmission of systemic mean arterial pressure fluctuations to the local microcirculation velocity of the skin flap lies in identifying the coupling relationship between the two in the time domain. Instead of performing global calculations on all-time data, the system employs an overlapping sliding window technique to extract signal segments with short-term stationary characteristics to capture the dynamic evolution of vascular tension over time. When calculating the cross-correlation coefficient and phase lag time, the system searches for the moments with the highest similarity between two sequences within a preset physiological conduction delay domain (typically covering the time range of arterial wave transmission to the extremities). The discretization calculation formula for the normalized cross-correlation function in this step is as follows: ;in, The cross-correlation function value when the time delay is d; The total number of sampling points within the sliding time window; : The data value at the m-th point of the whole-body mean arterial pressure sequence segment extracted within the current time window; This represents the arithmetic mean of the current systemic mean arterial pressure sequence segments; : The data value at the (m+d)th point of the local microcirculation velocity sequence segment of the skin flap captured within the current time window; The arithmetic mean of the current local microcirculation velocity sequence fragments on the skin flap. By traversing all d values ​​within the physiologically permissible range, the system locks onto... The moment when the global maximum value is reached is defined as the phase lag time, and this maximum value is the cross-correlation coefficient.

[0030] Specifically, the process of obtaining the blood flow conduction efficiency index by performing phase compensation based on phase lag time, calculating the velocity response amplitude ratio under unit pressure change, and combining the cross-correlation coefficient with the velocity response amplitude ratio is as follows: Applying a phase lag time shift to the whole-body mean arterial pressure sequence to eliminate the time-domain delay of pressure wave transmission to peripheral blood vessels, thereby achieving phase synchronization of pressure and flow velocity waveforms; calculating the effective value ratio of the whole-body mean arterial pressure sequence and the local microcirculation velocity sequence of the flap after phase synchronization as the velocity response amplitude ratio, using the cross-correlation coefficient as the waveform similarity weight, and the velocity response amplitude ratio as the conduction gain weight, and then calculating the blood flow conduction efficiency index that characterizes the patency of the vascular bed.

[0031] In this implementation scheme, after establishing the time delay relationship between pressure and flow velocity, the system mathematically eliminates the time lag caused by the vascular transmission path through phase compensation technology. This synchronizes the input signal (blood pressure) and the output signal (flow velocity) in waveform phase, thereby enabling accurate assessment of the transient response capability of the vascular bed to pressure fluctuations. This process essentially calculates the mechanical admittance characteristics of the vascular system. The system first calculates the ratio of the effective value of the synchronized flow velocity fluctuation to the effective value of the pressure fluctuation as a physical gain. Then, it integrates waveform similarity weights to construct a comprehensive index reflecting the inherent patency of the blood vessels. The fusion calculation model for the blood flow conduction efficiency index is as follows: ;in, Blood flow conduction efficiency index, used to characterize the physical patency and compliance of the vascular bed; The global maximum value of the cross-correlation coefficient extracted in the preceding steps; The waveform similarity weight index, typically set between 0.5 and 1.0, is used to adjust the degree of influence of signal correlation on the final index. Numerical values ​​of the flap microcirculation velocity sequence fragment aligned with the pressure sequence after phase compensation translation; The conduction gain weighting index, obtained through principal component analysis based on extensive historical clinical data, is used to balance the contribution of amplitude ratio in the assessment. This formula considers not only whether the flow velocity changes with blood pressure (correlation) but also how much flow velocity a unit of blood pressure can drive (gain), thus accurately eliminating false hypoperfusion. For example, during nighttime monitoring, patients enter deep sleep, and the mean arterial pressure decreases physiologically. At this time, the microcirculation velocity in the flap area decreases accordingly. Traditional monitoring equipment, relying solely on the absolute value of the flow velocity being below a threshold, is prone to issuing false ischemia alarms, leading to frequent ward rounds by medical staff and disturbing the patient's rest. The system in this embodiment first performs time-series alignment and differential normalization on the collected multi-source data to eliminate basal metabolic differences. Then, through cross-correlation calculations and phase compensation analysis, it captures that the changes in flap flow velocity and systemic blood pressure are highly consistent in time and waveform, and the calculated blood flow conduction efficiency index remains at a high level. This indicates that the vascular bed has a good mechanical response to blood pressure, and there is no physical obstruction inside the blood vessel. Based on this, the system determined that the current decrease in flow rate was a normal physiological phenomenon caused by fluctuations in systemic blood pressure, and automatically filtered out the alarm, effectively reducing the clinical false alarm rate and alleviating the burden on medical staff.

[0032] Specifically, the process of establishing a mapping relationship between the exogenous nursing intervention set and the flow rate gain, mapping the limb posture tilt sequence to the gravitational potential energy compensation coefficient, and mapping the local environmental temperature and humidity sequence to the vascular tension compensation coefficient is as follows: Analyze the vertical component in the limb posture tilt sequence, calculate the hydrostatic pressure gradient change caused by the change in body position using the principles of hydrostatics, and normalize and map it to the gravitational potential energy compensation coefficient; construct a nonlinear response curve of local environmental temperature and humidity and vascular smooth muscle tension, project the real-time acquired local environmental temperature and humidity sequence onto this response curve, and output a dimensionless factor characterizing vasomotor activity as the vascular tension compensation coefficient.

[0033] In this implementation plan, the core logic for establishing the mapping relationship between the exogenous nursing intervention set and the flow rate gain lies in transforming physical environmental parameters into quantitative influencing factors on hemodynamics. For the affected limb posture tilt sequence, the system first uses trigonometric functions to analyze the projection component of the gravity vector along the vascular axis, and then calculates the hydrostatic pressure difference caused by limb elevation or drooping based on hydrostatic principles. To convert this pressure difference into a coefficient that can participate in dimensionless calculations, the system introduces a standard reference blood pressure for normalization. The formula for calculating the gravitational potential energy compensation coefficient is as follows: ;in, The gravitational potential energy compensation coefficient at time t is used to reflect the attenuation or enhancement effect of body position on local perfusion pressure. Blood fluid density constant; Gravitational acceleration constant; The effective fluid tube length of the blood vessels in the affected limb from the level of the heart to the tip of the flap is set by preoperative imaging measurements. The pose and tilt angle of the affected limb relative to the horizontal plane, acquired in real time; The standard reference mean arterial pressure when the patient is in a supine position is used to normalize and calibrate the hydrostatic pressure gradient. For the local environmental temperature and humidity sequence, the system constructs a Sigmoid response model based on the nonlinear sensitivity of vascular smooth muscle to temperature changes. This model simulates the vasoconstriction limit of blood vessels at low temperatures and the vasodilation saturation state at high temperatures, mapping environmental thermodynamic parameters into dimensionless tension regulation factors. The calculation model for the vascular tension compensation coefficient is as follows: ;in, The vascular tension compensation coefficient at time t characterizes the vasomotor state of the blood vessel diameter. and These are the upper and lower bounds of the conductance gain of blood vessels under extreme vasodilation and extreme vasoconstriction states, respectively, determined by analyzing the extreme correlation between temperature and flow rate in historical clinical data. The slope factor of blood vessel sensitivity to temperature changes; Real-time acquisition of local ambient thermodynamic temperature; The half-saturation inflection point temperature of the vasomotor response is usually set as the comfortable temperature threshold of the human body surface.

[0034] Specifically, the process of modulating the whole-body mean arterial pressure sequence using the gravitational potential energy compensation coefficient and the vascular tension compensation coefficient to calculate the target perfusion trajectory under the current nursing state is as follows: The whole-body mean arterial pressure sequence is amplitude corrected using the gravitational potential energy compensation coefficient to restore the effective perfusion pressure after eliminating gravitational potential energy interference; the vascular tension compensation coefficient is used as a dynamic gain control variable to modulate the effective perfusion pressure and simulate the theoretical flow velocity response under the current vascular tension; the average flow velocity of the symmetrical part on the healthy side is used as the reference conductivity, and the modulated theoretical flow velocity response is mapped to the flow velocity domain to generate the target perfusion trajectory.

[0035] In this implementation scheme, when calculating the target perfusion trajectory under the current nursing condition, the system aims to reconstruct a theoretical flow velocity curve driven only by the current nursing environment and systemic blood pressure, assuming ideal vascular patency. This process first modulates the systemic mean arterial pressure using a gravitational potential energy compensation coefficient to recover the actual effective perfusion pressure reaching the flap tip. Then, a vascular tension compensation coefficient is superimposed to simulate the amplification or flow restriction effect of changes in vessel diameter on flow velocity. Finally, the blood flow admittance of the unaffected limb is introduced as a reference to map the pressure domain signal back to the flow velocity domain. The dynamic evolution formula of the target perfusion trajectory is as follows: ;in, : Calculate the theoretical value of the target perfusion trajectory at time t; Mean microcirculation velocity at rest in the symmetrical region of the healthy side; The mean arterial pressure of the whole body during the sampling period at the symmetrical site on the healthy side; the ratio of the two constitutes the baseline vascular conductivity of the individual patient. Real-time acquired mean arterial pressure sequence values ​​throughout the body; and These are the gravitational potential energy compensation coefficient and the vascular tension compensation coefficient, respectively, calculated in the previous steps. Through layer-by-layer correction of the physical model, this formula generates a dynamically adaptive early warning baseline that adapts to the nursing environment.

[0036] Specifically, the numerical difference between the local microcirculation velocity sequence of the flap and the target perfusion trajectory is calculated to obtain the dynamic residual vector. The logical process of jointly determining the dynamic residual vector and the blood flow conduction efficiency index is as follows: extract the instantaneous deviation amplitude of the local microcirculation velocity sequence of the flap relative to the target perfusion trajectory, and the gradient direction of the deviation amplitude changing over time; construct a state feature vector composed of the deviation amplitude and the gradient direction as the dynamic residual vector; establish a two-dimensional state classification plane containing structural blockade area and functional inhibition area, and project the dynamic residual vector and the blood flow conduction efficiency index as coordinate parameters onto the classification plane for region matching.

[0037] In this implementation scheme, when calculating the numerical difference between the local microcirculation velocity sequence of the flap and the target perfusion trajectory and constructing the dynamic residual vector, the system aims to quantify the degree and trend of the actual blood flow state deviating from the theoretical optimal state. First, the system performs low-pass filtering on the real-time acquired local microcirculation velocity of the flap to suppress measurement noise. Then, it performs time-domain subtraction with the target perfusion trajectory generated in the previous step to obtain the instantaneous deviation amplitude. To capture the rate of disease progression, the system further differentiates the deviation amplitude over time to extract the first-order gradient feature. The dynamic residual vector, composed of these two components, not only reflects the current degree of ischemia but also predicts the trend of ischemia worsening or easing. The construction model of the dynamic residual vector is described as follows: ;in, The dynamic residual vector at time t is used as the input feature vector for the subsequent state classifier; Instantaneous deviation amplitude; the larger the negative value, the more severe the current blood flow deficit. Real-time acquired and denoised local microcirculation velocity sequence values ​​of the skin flap; The theoretical target perfusion trajectory value generated by the preceding steps; The first derivative of the deviation amplitude over time is used to characterize the rate of disease deterioration. Subsequently, the system establishes a two-dimensional feature space with the dynamic residual vector amplitude and blood flow conduction efficiency index as axes, and projects the current state into this space to prepare for subsequent region matching.

[0038] Specifically, if the dynamic residual vector is lower than a preset negative threshold and the blood flow conduction efficiency index is lower than a preset blocking threshold, a vascular structural occlusion signal is output; if the dynamic residual vector is lower than a preset negative threshold and the blood flow conduction efficiency index is located within a preset connected interval, a nursing plan functional inhibition signal is output. The specific process is as follows: the instantaneous amplitude of the dynamic residual vector is detected. When the instantaneous amplitude is less than a preset negative threshold and the corresponding blood flow conduction efficiency index is lower than a preset blocking threshold, a vascular structural occlusion signal is generated; when the instantaneous amplitude is less than a preset negative threshold, but the corresponding blood flow conduction efficiency index is higher than a preset blocking threshold and is within a preset connected interval, a nursing plan functional inhibition signal is generated.

[0039] In this implementation plan, the core logic of the joint determination process for structural occlusion signals of output vessels and functional inhibition signals of nursing plans lies in using the blood flow conduction efficiency index as the gold standard for differential diagnosis. The system pre-sets judgment boundaries based on statistically significant differences, including a negative deviation threshold to define blood flow abnormalities and an obstruction threshold to define vascular patency. When the system detects that the actual flow rate is significantly lower than the theoretical target, indicating a severe perfusion deficit, the system further examines the vessel's ability to conduct blood pressure. If the conduction ability is extremely low, it means that the physical pathway of the vessel is interrupted, indicating an organic lesion; if the conduction ability is still acceptable, it indicates that the vessel itself has good elasticity, and the low flow rate is caused by inhibition from the external nursing environment (such as postural compression, hypothermia). The logical threshold function for this joint determination is as follows: ;in, The output warning signal status code at time t: 1 represents a structural vascular occlusion signal, requiring immediate surgical exploration; 2 represents a functional inhibition signal in the nursing plan, requiring optimization of nursing measures; 0 represents normal or non-specific fluctuations. The preset negative deviation threshold is set to -20% to -30% of the mean of the target perfusion trajectory, and is used to filter out normal physiological fluctuations. The preset blocking threshold is set at 10% to 15% of the average blood flow conduction efficiency index of the healthy limb. A value below this threshold indicates that the vascular bed has lost its mechanical conduction properties. Real-time blood flow conduction efficiency index obtained in the preceding steps; The preset upper limit of the connectivity interval is typically set at 80% to 90% of the average blood flow conduction efficiency index of the healthy limb. Through this logic, the system achieves precise source identification of crisis types. For example, during subsequent monitoring, due to adjustments in the ward's air conditioning direction, the local ambient temperature of the affected limb gradually decreased, causing cold-induced vasoconstriction and a further decrease in flow velocity. Simultaneously, the patient's turning over slightly altered the elevation angle of the affected limb. At this point, using the established physical mapping relationship, the system analyzed the specific impact of the current low temperature and body position on blood flow, deducing the target perfusion trajectory the flap should achieve under the current nursing environment if the blood vessels were patent. Comparing the measured flow velocity with this target trajectory revealed a significant numerical deviation, indicating that the dynamic residual vector exceeded the negative threshold, suggesting insufficient perfusion. Crucially, the system again jointly verified the blood flow conduction efficiency index, finding that it remained within the connectivity interval (i.e., the blood vessels themselves were still patent, only their diameter narrowed due to cold). Based on this, no vascular embolism alarm was issued; instead, a functional inhibition signal from the nursing plan was output. The system alerted medical staff that the blood vessels were not blocked, and the low flow rate was due to "being too cold" or "being in the wrong position." Following the system's instructions, the medical staff simply warmed the affected limb and slightly adjusted its position; the skin flap flow rate immediately returned to the target trajectory range.

[0040] In summary, this application has at least the following effects:

[0041] A method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion is proposed. This method constructs a multidimensional time-varying state matrix incorporating physiological driving characteristics and nursing environment features, and extracts the blood flow conduction efficiency index using cross-correlation operations. This allows for the precise extraction of the inherent mechanical patency characteristics of the vascular bed from complex physiological fluctuations, effectively avoiding false alarms caused by systemic blood pressure fluctuations. Simultaneously, by establishing a physical mapping relationship between the exogenous nursing intervention set and flow velocity gain, the target perfusion trajectory under the current nursing state is deduced. Combined with dynamic residual vectors and the blood flow conduction efficiency index for joint judgment, this method achieves, for the first time, accurate differentiation between vascular structural occlusion and functional inhibition of the nursing plan. This enables timely detection of pathological crises while objectively assessing the effectiveness of nursing measures such as body position and temperature, and guiding optimization, significantly improving the level of refined postoperative care and the accuracy of clinical decision-making.

[0042] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0043] This invention is described with reference to flowchart illustrations and / or block diagrams of systems, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0044] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0045] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0046] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0047] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion, characterized in that, Includes the following steps: S1. The mean arterial pressure sequence of the whole body and the local microcirculation velocity sequence of the skin flap were collected simultaneously to form an endogenous physiological load set. The position and tilt angle sequence of the affected limb and the local temperature and humidity sequence were collected simultaneously to form an exogenous nursing intervention set. The flow velocity of the symmetrical part of the healthy side was selected as the benchmark. The endogenous physiological load set was differentially normalized and time-series aligned to construct a multidimensional time-varying state matrix containing physiological driving characteristics and nursing environment characteristics. S2. Perform cross-correlation calculation on the multidimensional time-varying state matrix, extract the phase lag time and cross-correlation coefficient of the systemic mean arterial pressure sequence wave conduction to the local microcirculation velocity sequence of the flap, perform phase compensation based on the phase lag time, calculate the velocity response amplitude ratio under unit pressure change, and obtain the blood flow conduction efficiency index by combining the cross-correlation coefficient and the velocity response amplitude ratio. S3. Establish the mapping relationship between the exogenous nursing intervention set and the flow rate gain, map the limb position tilt angle sequence to the gravitational potential energy compensation coefficient, map the local environmental temperature and humidity sequence to the vascular tension compensation coefficient, modulate the whole body mean arterial pressure sequence through the gravitational potential energy compensation coefficient and the vascular tension compensation coefficient, and calculate the target perfusion trajectory under the current nursing state. S4. Calculate the numerical difference between the local microcirculation velocity sequence of the flap and the target perfusion trajectory to obtain the dynamic residual vector. Combine the dynamic residual vector with the blood flow conduction efficiency index for judgment: if the dynamic residual vector is lower than the preset negative threshold and the blood flow conduction efficiency index is lower than the preset occlusion threshold, output the vascular structural occlusion signal; if the dynamic residual vector is lower than the preset negative threshold and the blood flow conduction efficiency index is located in the preset connected interval, output the nursing plan functional inhibition signal.

2. The method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion as described in claim 1, characterized in that: The specific process of simultaneously acquiring systemic mean arterial pressure sequences and local microcirculation velocity sequences of the skin flap to form an endogenous physiological load set, and simultaneously acquiring affected limb position and tilt angle sequences and local environmental temperature and humidity sequences to form an exogenous nursing intervention set is as follows: The mean systolic and diastolic blood pressure values ​​of continuous arterial pressure waveforms were extracted by cardiac cycle gating sampling to construct a systemic mean arterial pressure sequence; Spatiotemporal contrast analysis of the free flap region and the symmetrical region on the healthy side was performed using a laser speckle blood flow imaging system to quantify and generate a local microcirculation velocity sequence of the flap. The spatial Euler angles of the affected limb relative to the vertical line of gravity are calculated by the inertial measurement unit, and the pitch angle component is extracted as the pose tilt sequence of the affected limb. The thermal radiation flux of the skin flap surface and the surrounding environment is captured by a non-contact thermal imaging array, and the local environmental temperature and humidity sequence is generated by inversion.

3. The method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion as described in claim 2, characterized in that: The specific process of constructing a multidimensional time-varying state matrix that includes physiological driving characteristics and nursing environment characteristics by selecting the flow velocity of the symmetrical part of the healthy side as the benchmark, performing differential normalization and time-series alignment on the endogenous physiological load set, is as follows: Using the flow velocity of the symmetrical part of the healthy side as the individual baseline background flow velocity, background difference operation was performed on the local microcirculation flow velocity sequence of the flap to remove individual basal metabolic drift and obtain the net flow velocity sequence. A multi-rate signal synchronization algorithm was used to resample the limb posture tilt sequence and the local environmental temperature and humidity sequence, and the time resolution was unified to the sampling frequency of the whole body mean arterial pressure sequence. Based on the time axis of the whole body mean arterial pressure sequence, the whole body mean arterial pressure sequence, the net flow velocity sequence and the resampled exogenous nursing intervention set were spliced ​​in the feature dimension to generate a multidimensional time-varying state matrix.

4. The method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion as described in claim 1, characterized in that: The specific process of performing cross-correlation calculations on the multidimensional time-varying state matrix to extract the phase lag time and cross-correlation coefficient of the systemic mean arterial pressure sequence wave propagation to the local microcirculation velocity sequence of the flap is as follows: By using a sliding time window to extract systemic mean arterial pressure sequence fragments and flap local microcirculation velocity sequence fragments from the multidimensional time-varying state matrix, a cross-correlation function of the two sequence fragments is constructed, and the global maximum point of the cross-correlation function is retrieved within a preset physiological conduction delay domain. The time delay corresponding to the global maximum point is locked as the phase lag time, and the normalized amplitude of the global maximum point is locked as the cross-correlation coefficient.

5. The method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion as described in claim 4, characterized in that: The specific process for obtaining the blood flow conduction efficiency index is as follows: Phase compensation is performed based on the phase lag time, the velocity response amplitude ratio under unit pressure change is calculated, and the cross-correlation coefficient is combined with the velocity response amplitude ratio. Applying a phase lag time shift to the whole-body mean arterial pressure sequence eliminates the time-domain delay of pressure wave propagation to peripheral blood vessels, thereby achieving phase synchronization between pressure and flow velocity waveforms; The effective value ratio of the whole-body mean arterial pressure sequence and the local microcirculation velocity sequence of the flap after phase synchronization is calculated as the velocity response amplitude ratio. The cross-correlation coefficient is used as the waveform similarity weight, and the velocity response amplitude ratio is used as the conduction gain weight. The blood flow conduction efficiency index, which characterizes the patency of the vascular bed, is obtained by fusion calculation.

6. The method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion as described in claim 1, characterized in that: The specific process of establishing the mapping relationship between the exogenous nursing intervention set and the flow rate gain, mapping the affected limb posture tilt angle sequence to the gravitational potential energy compensation coefficient, and mapping the local environmental temperature and humidity sequence to the vascular tension compensation coefficient is as follows: The vertical component in the position tilt sequence of the affected limb is analyzed, and the hydrostatic pressure gradient change caused by the change in body position is calculated by combining the principle of hydrostatics. The change is then normalized and mapped to the gravitational potential energy compensation coefficient. A nonlinear response curve of local environmental temperature and humidity versus vascular smooth muscle tension is constructed. The real-time acquired local environmental temperature and humidity sequence is projected onto this response curve, and a dimensionless factor characterizing vasomotor activity is output as the vascular tension compensation coefficient.

7. The method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion as described in claim 6, characterized in that: The specific process of modulating the whole-body mean arterial pressure sequence using the gravitational potential energy compensation coefficient and the vascular tension compensation coefficient to calculate the target perfusion trajectory under the current nursing condition is as follows: The amplitude of the whole-body mean arterial pressure sequence is corrected by the gravitational potential energy compensation coefficient to restore the effective perfusion pressure after eliminating gravitational potential energy interference; The vascular tension compensation coefficient is used as a dynamic gain control variable to modulate the effective perfusion pressure and simulate the theoretical flow velocity response under the current vascular tension. Using the average flow velocity of the symmetrical part of the healthy side as the reference conductivity, the modulated theoretical flow velocity response is mapped to the flow velocity domain to generate the target infusion trajectory.

8. The method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion as described in claim 1, characterized in that: The numerical difference between the local microcirculation velocity sequence of the flap and the target perfusion trajectory is calculated to obtain the dynamic residual vector. The logical process of jointly determining the dynamic residual vector with the blood flow conduction efficiency index is as follows: Extract the instantaneous deviation amplitude of the local microcirculation velocity sequence of the flap relative to the target perfusion trajectory, and the gradient direction of the deviation amplitude over time; Construct a state feature vector consisting of the deviation magnitude and gradient direction as the dynamic residual vector; A two-dimensional state classification plane containing structural blockade and functional inhibition zones is established. The dynamic residual vector and blood flow conduction efficiency index are projected onto this classification plane as coordinate parameters for region matching.

9. The method for evaluating and providing early warning of postoperative care for free flaps based on multimodal monitoring data fusion as described in claim 8, characterized in that: If the dynamic residual vector is lower than a preset negative threshold and the blood flow conduction efficiency index is lower than a preset occlusion threshold, a structural occlusion signal is output; if the dynamic residual vector is lower than a preset negative threshold and the blood flow conduction efficiency index is within a preset connectivity interval, a functional inhibition signal of the nursing plan is output. The specific process is as follows: The instantaneous amplitude of the dynamic residual vector is detected. When the instantaneous amplitude is less than a preset negative threshold and the corresponding blood flow conduction efficiency index is lower than a preset occlusion threshold, a vascular structural occlusion signal is generated. When the instantaneous amplitude is less than the preset negative threshold, but the corresponding blood flow conduction efficiency index is higher than the preset blocking threshold and is within the preset connectivity interval, a functional inhibition signal of the nursing plan is generated.

Images