Integrated traditional chinese and western medicine clinical pathway gallstone postoperative rehabilitation evaluation and management system
By integrating multimodal data acquisition and adaptive adjustment modules into a medical edge computing gateway, the problems of nonlinear fluctuations in the physiological homeostasis of patients after cholelithiasis surgery and conflicts between traditional Chinese and Western medicine interventions were solved, achieving a smooth transition and high availability of dynamically reconstructed rehabilitation pathways.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- THE FIRST AFFILIATED HOSPITAL OF GUANGXI UNIV OF TRADITIONAL CHINESE MEDICINE (GUANGXI TRADITIONAL CHINESE MEDICINE HOSPITAL)
- Filing Date
- 2026-03-26
- Publication Date
- 2026-06-26
Smart Images
Figure CN122290883A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of intelligent rehabilitation systems and medical edge computing technology, specifically to a postoperative rehabilitation evaluation and management system for cholelithiasis using a clinical pathway combining traditional Chinese and Western medicine. Background Technology
[0002] With the increasing integration of traditional Chinese and Western medicine in the rehabilitation of cholelithiasis patients, the management of clinical pathways has become significantly more complex. This complexity brings many challenges, especially in the fusion of multi-source heterogeneous data and dynamic intervention. Currently, postoperative management is generally carried out through pre-set static clinical pathways. Medical staff rely on objective indicators of fast variables from Western medicine and characteristic data of slow variables from Traditional Chinese Medicine to guide and promote the patient's recovery process. However, traditional static clinical pathways cannot cope with the nonlinear fluctuations of individualized postoperative physiological homeostasis. When processing cross-modal data, they face the technical defect of extremely unequal sampling frequencies of Western and Traditional Chinese Medicine indicators. At the same time, the intervention logic of Western and Traditional Chinese Medicine often has directional deviations at specific physiological stages, and blind combination can easily lead to internal rejection of physiological dimensions. Moreover, frequent or concentrated reconstruction of rehabilitation pathways can easily cause intervention oscillations and drastic jumps in intervention intensity, resulting in the deterioration of vital signs in weak patients. In addition, the existing architecture relies too much on centralized cloud processing, which faces a very high risk of response latency when the network fluctuates. Therefore, how to effectively integrate multimodal heterogeneous data with unequal sampling frequencies, and achieve dynamic adaptive optimization and low-latency intervention of postoperative clinical pathways while resolving the underlying medical conflicts between traditional Chinese and Western medicine and ensuring the smoothness of control execution, has become an urgent problem to be solved in this field. Summary of the Invention
[0003] The purpose of this invention is to provide a postoperative rehabilitation evaluation and management system for cholelithiasis based on a clinical pathway combining traditional Chinese and Western medicine, and to solve the following technical problems: It avoids the shortcomings of traditional static clinical pathways, which cannot cope with the nonlinear fluctuations of patients' physiological homeostasis and the conflict caused by the direct combination of Chinese and Western medicine intervention logics. It can also achieve deep integration of interdisciplinary rehabilitation strategies at the system level, adaptive optimization, and smooth transition of execution intensity.
[0004] The objective of this invention can be achieved through the following technical solutions: The postoperative rehabilitation evaluation and management system for cholelithiasis using a clinical pathway combining traditional Chinese and Western medicine is characterized in that the system is integrated into a medical edge computing gateway and includes, in sequence, a multimodal data acquisition module for acquiring fast variable objective index data and slow variable characteristic data of a preset target object. The fast variable objective index data includes at least one of drainage fluid characteristics data and pain score data, and the slow variable characteristic data includes at least one of pulse characteristic data and tongue characteristic data. The state space modeling module is used to construct a multimodal rehabilitation state space based on the time-aligned multimodal data, and to calculate the weighted dynamic adjustment rate of the fast and slow variables fused as heterogeneous evaluation data. The path adaptive tracking module is used to calculate the offset tolerance degree of the current node relative to the preset baseline clinical path node based on the multimodal rehabilitation state space and the weight dynamic adjustment rate, and generate a dynamic reconstruction intervention strategy containing multiple execution nodes, including traditional Chinese medicine physiotherapy nodes and Western medicine rehabilitation training nodes. The conflict resolution and smoothing control module is used to extract heterogeneous logic conflict features based on the angle between the state vectors representing the intervention logic of traditional Chinese and Western medicine in the multimodal rehabilitation state space, and calculate the number of convergence steps for resolving medical conflict to update the dynamic reconstruction intervention strategy, and calculate the smoothness of rehabilitation trajectory state transition based on the updated dynamic reconstruction intervention strategy. The strategy output and management module is used to filter the updated dynamic reconstruction intervention strategy based on the smoothness of the rehabilitation trajectory state transition, and output the final intervention strategy sequence, which includes the sequence of traditional Chinese medicine physiotherapy and Western medicine rehabilitation training nodes.
[0005] As a further aspect of the present invention: the state space modeling module includes: a time alignment unit, used to resample the fast variable objective index data and the slow variable feature data according to a preset time base to generate time-aligned multimodal data; The weight calculation unit, connected to the time alignment unit, is used to extract nonlinear coupling features by calculating the mutual information entropy between the time series of fast variable objective index data and slow variable feature data in the time-aligned multimodal data, and to calculate the weight dynamic adjustment rate of the heterogeneous evaluation data fusion based on the nonlinear coupling features. The spatial mapping unit, connected to the weight calculation unit, is used to project the fast variable objective index data and slow variable feature data in the time-aligned multimodal data to a high-dimensional space using a preset mapping matrix to generate a first state vector and a second state vector. The first state vector and the second state vector are then weighted and superimposed using the weight dynamic adjustment rate of the heterogeneous evaluation data fusion to generate the current node. The multimodal rehabilitation state space is constructed based on the current node, and a multimodal state transition prediction model is pre-set in the multimodal rehabilitation state space.
[0006] As a further aspect of the present invention: the path adaptive tracing module includes: a topology distance calculation unit, used to obtain preset baseline path nodes in a pre-stored standard rehabilitation model from a preset static knowledge base, calculate the distance between the current node and the preset baseline path node in the multimodal rehabilitation state space, and output the topology offset distance; The tolerance assessment unit, connected to the topology distance calculation unit, is used to divide the preset maximum tolerance distance by the topology offset distance when the topology offset distance is not zero, so as to generate the tolerance of the clinical pathway node offset. A strategy generation unit, connected to the inclusion assessment unit, is used to compare the clinical pathway node offset inclusion with a preset inclusion threshold. The strategy generation unit is also used to trigger a strategy generation operation to generate the dynamic reconstruction intervention strategy when the clinical pathway node offset tolerance is greater than or equal to the preset tolerance threshold. The strategy generation unit is also used to generate a manual takeover alarm signal and terminate the generation of the dynamic reconstruction intervention strategy when the clinical pathway node offset tolerance is less than the preset tolerance threshold.
[0007] As a further aspect of the present invention: the conflict resolution and smoothing control module includes: The conflict detection unit is used to compare the first state vector and the second state vector in the multimodal rehabilitation state space, and output the heterogeneous logic conflict feature when the cosine of the angle between the first state vector and the second state vector is less than a preset consistency threshold. An iterative backtracking unit, connected to the conflict detection unit, is used to initiate state machine backtracking calculation in response to the heterogeneous logic conflict feature, and record the number of iterations required to make the cosine of the angle between the first state vector and the second state vector greater than or equal to the preset consistency threshold, so as to generate the convergence steps for resolving the medical conflict. The strategy correction unit, connected to the iterative backtracking unit, is used to determine whether to adjust the execution nodes in the dynamic reconstruction intervention strategy in advance or delay them based on the phase difference direction between the first state vector and the second state vector, i.e. the deflection direction representing the leading or lagging relationship of recovery progress, and to determine the adjustment magnitude based on the number of convergence steps of the medical conflict resolution, so as to generate the updated dynamic reconstruction intervention strategy.
[0008] As a further aspect of the present invention, the conflict resolution and smoothing control module further includes: a smoothness calculation unit, connected to the strategy correction unit, used to calculate the discrete difference rate of change of the state vector in the multimodal rehabilitation state space before and after the simulation execution of the updated dynamic reconstruction intervention strategy based on discrete time steps, and generate the smoothness of the rehabilitation trajectory state transition. The strategy output and management module is used to compare the smoothness of the rehabilitation trajectory state transition with a preset smoothness threshold; the strategy output and management module is also used to confirm the updated dynamic reconstruction intervention strategy as the final intervention strategy sequence when the smoothness of the rehabilitation trajectory state transition is greater than or equal to the preset smoothness threshold. The strategy output and management module is also used to send a readjustment instruction to the strategy correction unit when the smoothness of the recovery trajectory state transition is less than the preset smoothness threshold.
[0009] As a further aspect of the present invention: the multimodal data acquisition module includes: The first acquisition unit is used to acquire the property data of the drainage fluid through a preset biochemical detection device; The second acquisition unit is used to acquire the pain score data through a preset interactive terminal; The third acquisition unit is used to acquire the tongue image feature data through a preset image sensor; The fourth acquisition unit is used to acquire the pulse characteristic data through a preset pressure sensor.
[0010] As a further embodiment of the present invention: the multimodal data acquisition module, the state space modeling module, the path adaptive tracing module, the conflict resolution and smoothing control module, and the strategy output and management module are integrated into a preset medical edge computing gateway; The medical edge computing gateway interacts with a pre-set hospital information system through a preset secure communication protocol.
[0011] As a further aspect of the present invention: in the weight calculation unit, the formula for calculating the dynamic adjustment rate of the weights for the fusion of heterogeneous evaluation data is: ;in, The weight dynamic adjustment rate; This is the preset sensitivity adjustment coefficient; The mutual information entropy between the time series of fast variable objective indicator data and slow variable characteristic data; This is the preset threshold for mutual information entropy.
[0012] As a further aspect of the present invention: in the strategy correction unit, the formula for calculating the adjustment magnitude is: in, The time step for the adjustment; The number of convergence steps is used to resolve conflicts in medical theory. The system's built-in preset base time step.
[0013] As a further aspect of the present invention: when the heartbeat monitoring process of the medical edge computing gateway detects an unexpected disconnection of the secure communication link with the preset hospital information system, the medical edge computing gateway implements communication degradation isolation, temporarily stores dynamically generated intervention strategy execution logs using the built-in solid-state storage array, and directly sends control commands to the external preset device through the local physical serial port.
[0014] The beneficial effects of this invention are: 1. This invention resamples fast and slow variables with unequal sampling frequencies by using time alignment units, extracts nonlinear coupling features to construct a multimodal rehabilitation state space; and generates dynamic reconstruction intervention strategies based on the clinical path node offset inclusiveness, which solves the problem that static paths cannot cope with postoperative physiological nonlinear fluctuations, and realizes the standardized fusion of heterogeneous data and adaptive path finding. 2. This invention addresses the directional divergence between the intervention logic of traditional Chinese medicine and Western medicine. The system extracts heterogeneous logical conflict features by comparing the cosine value of the angle between the state vectors through a conflict detection unit; the iterative backtracking unit starts the state machine to calculate the convergence step number, and adjusts the execution node accordingly by advancing or delaying it; this mechanism avoids the physiological internal rejection caused by blind combination and ensures the absolute self-consistency of the interdisciplinary rehabilitation strategy. 3. To prevent excessive intervention from causing a deterioration in vital signs, the smoothness calculation unit of this invention generates the smoothness of the state transition of the rehabilitation trajectory based on the discrete difference rate of change of the state vector; when the smoothness does not meet the standard, the system forcibly sends a readjustment command; this measure realizes physical boundary control of the intensity of the implementation, avoiding violent jumps and oscillations in intensity caused by path reconstruction. 4. This invention integrates all core algorithm modules into a medical edge computing gateway, and interacts with the cloud-based hospital information system only through a secure communication protocol. This completely overcomes the risk of high response latency that can easily be caused by over-reliance on the cloud when the network is congested. It pushes computing power down to the ground and realizes highly available reconstruction and localized closed-loop management of instantaneous intervention strategies in acute and critical care scenarios. Attached Figure Description
[0015] The invention will now be further described with reference to the accompanying drawings.
[0016] Figure 1 This is a schematic diagram of the modules of the postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway provided in this application embodiment. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] Please see Figure 1The postoperative rehabilitation evaluation and management system for cholelithiasis using integrated traditional Chinese and Western medicine clinical pathways includes a multimodal data acquisition module that is connected in sequence to acquire fast variable objective index data and slow variable characteristic data of a preset target object. The fast variable objective index data includes at least one of drainage fluid characteristics data and pain score data, and the slow variable characteristic data includes at least one of pulse characteristic data and tongue characteristic data. The state space modeling module is used to construct a multimodal rehabilitation state space based on fast variable objective index data and slow variable feature data, and to calculate the weight dynamic adjustment rate of heterogeneous evaluation data fusion. The path adaptive tracking module is used to calculate the clinical path node offset tolerance based on the multimodal rehabilitation state space and weight dynamic adjustment rate, and generate a dynamic reconstruction intervention strategy with multiple execution nodes, including traditional Chinese medicine physiotherapy nodes and Western medicine rehabilitation training nodes. The conflict resolution and smoothing control module is used to extract heterogeneous logical conflict features based on the multimodal rehabilitation state space, calculate the number of convergence steps for medical conflict resolution to update the dynamic reconstruction intervention strategy, and calculate the smoothness of rehabilitation trajectory state transitions based on the updated dynamic reconstruction intervention strategy. The strategy output and management module is used to filter and update the dynamic reconstruction intervention strategy based on the smoothness of the rehabilitation trajectory state transition, and output the final intervention strategy sequence, which includes the sequence of TCM physiotherapy and Western medicine rehabilitation training nodes.
[0019] This embodiment provides a dynamic tracking and intervention mechanism for postoperative rehabilitation that integrates multimodal evaluation data. Specifically, the system acquires objective fast variable indicators and subjective or slowly evolving slow variable characteristics of patients through a multimodal data acquisition module, such as the pixel ratio of tongue coating, which characterizes damp-heat in the spleen and stomach. and the characteristic values of pulse peaks that indicate deficiency of Qi and blood ; The state space modeling module maps the two types of heterogeneous data to the multimodal rehabilitation state space based on a preset dimension, and calculates the dynamic adjustment rate of the weights when the two are fused. The path adaptive tracking module adjusts according to this rate. Based on the current state space coordinates, the allowable deviation of clinical pathway node offsets from the standard clinical pathway is calculated. ; The conflict resolution and smoothing control module extracts the heterogeneous logical conflict characteristics between the holistic conditioning logic of Traditional Chinese Medicine and the local intervention logic of Western Medicine, and calculates the convergence steps of medical conflict resolution. The intervention strategy was updated and reconstructed, and the smoothness of the rehabilitation trajectory state transitions after the simulated implementation of the strategy was calculated. ; The strategy output and management module utilizes smoothness The system performs filtering and control to output an optimal sequence composed of a mixture of traditional Chinese medicine physiotherapy nodes and Western medicine rehabilitation training nodes. If, during the multimodal data acquisition phase, slow variable feature data is missing due to equipment failure or patient positioning restrictions, the system will trigger a degraded operation mechanism, forcibly adjusting the weight dynamic adjustment rate. It tends to rely solely on the single extreme value operation of fast variables and pushes early warning of blind spots in TCM characteristics to the attending physician to request manual on-site supplementation; For example, in the rehabilitation scenario of a 65-year-old patient who underwent laparoscopic cholecystectomy 48 hours after surgery, the system not only recorded that the postoperative drainage fluid gradually became clearer and the fast variable reached the target, but also keenly captured that the patient's tongue coating was still thick and greasy and the pulse was thin and hesitant, and the slow variable was not resolved. This avoided the misjudgment of blindly arranging strenuous ambulation just because the Western medicine indicators were normal under the traditional approach. Instead, it adopted a dynamic reorganization strategy, inserting traditional Chinese medicine hot compress and Zusanli acupoint application in the early stage to regulate the middle jiao. The purpose of this mechanism is to address the pain point that traditional static clinical pathways cannot cope with the nonlinear fluctuations in the individualized postoperative physiological homeostasis of patients, thereby achieving deep integration and adaptive optimization of Chinese and Western medicine interventions at the system level.
[0020] In a preferred embodiment of the present invention, the state space modeling module includes: a time alignment unit, used to resample the fast variable objective index data and the slow variable feature data according to a preset time base to generate time-aligned multimodal data; The weight calculation unit, connected to the time alignment unit, is used to extract nonlinear coupling features in time-aligned multimodal data and calculate the weight dynamic adjustment rate of heterogeneous evaluation data fusion based on the nonlinear coupling features. The spatial mapping unit, connected to the weight calculation unit, is used to project the fast variable objective index data and slow variable feature data in the time-aligned multimodal data to a high-dimensional space using a preset mapping matrix to generate a first state vector and a second state vector. It also constructs a multimodal rehabilitation state space by combining the weight dynamic adjustment rate of the heterogeneous evaluation data fusion. The multimodal rehabilitation state space contains the current node generated by the time-aligned multimodal data mapping, and a multimodal state transition prediction model based on historical real rehabilitation data and long short-term memory network training is pre-set in the multimodal rehabilitation state space.
[0021] Objectively, the rehabilitation state space constructed by the previous embodiment faces a technical defect when directly processing raw cross-modal data: the sampling frequency of Chinese and Western medicine indicators is extremely unequal. For example, postoperative heart rate and drainage fluid monitoring may occur once per hour, while the four diagnostic features of traditional Chinese medicine may only be collected twice a day by the physician. Forced fusion will cause the spatial vector to be distorted on the time axis. Therefore, this embodiment provides a heterogeneous data isomorphism processing mechanism based on time alignment and spatial projection; specifically, the time alignment unit sets a reference time window. For example, if the sampling frequency is 4 hours, the sampling frequency will be... The fast variable and the sampling frequency are The slow variables are interpolated and aligned to generate time-aligned multimodal data; the weighting unit extracts nonlinear coupling features from the data array, specifically by calculating the mutual information entropy between the time series of fast and slow variables. This is used to quantify the degree of nonlinear correlation between the two, and the dynamic adjustment rate of the weights is calculated based on the mutual information entropy. Specifically, the system calculates the dynamic adjustment rate of the weights using the following normalized exponential mapping formula: in, This is a preset sensitivity adjustment coefficient used to control the smoothness of weight adjustment; The preset mutual information entropy threshold is used as the reference threshold; this mapping mechanism can smoothly map the quantized nonlinear mutual information entropy to the target value through an activation function. Within the interval, thus providing a computable and explicit mathematical benchmark for the subsequent weighted superposition step of spatial mapping units; The spatial mapping unit utilizes a preset An orthogonal mapping matrix is used to project scalar fast and slow variables onto each other, generating a first state vector representing the physical signs in Western medicine. and the second state vector representing the overall syndrome in Traditional Chinese Medicine and based on Weighted spatial overlay is performed to generate the current node coordinates representing the patient's real-time recovery progress. ; If the weight dynamic adjustment rate of the nonlinear coupling characteristic The activity level remains consistently below the preset minimum threshold. If the rate of change is close to 0 within three consecutive baseline time windows, it indicates that the patient's vital signs have entered a period of extreme stability or recovery stagnation. This differs from the degraded operation mechanism triggered by data loss due to equipment failure. In this case, to save computing resources, the system will temporarily freeze the iterative update of the mapping matrix in the spatial mapping unit to avoid unnecessary computing power consumption. For example, for the above-mentioned patients after laparoscopic cholecystectomy, the system performs cubic spline interpolation on the time axis to align the pain score data entered by the patient every 2 hours through the interactive terminal with the feature readings obtained by the image sensor once in the morning and once in the evening, and maps them into two high-dimensional vectors respectively. Finally, they are merged into a single coordinate node in the high-dimensional space that can accurately locate the current comprehensive state of qi, blood and inflammation. The purpose of this step is to eliminate the evolution time difference and asynchronicity of the evaluation indicators of traditional Chinese medicine and Western medicine, so as to realize the standardized quantification and comparative evaluation of heterogeneous medical features in the same mathematical space.
[0022] In a preferred embodiment of the present invention, the path adaptive tracing module includes: a topology distance calculation unit, used to obtain preset baseline path nodes in a pre-stored standard rehabilitation model from a preset static knowledge base, calculate the distance between the current node and the preset baseline path node in the multimodal rehabilitation state space, and output the topology offset distance; and a tolerance assessment unit, connected to the topology distance calculation unit, used to divide the preset maximum tolerance distance by the topology offset distance to generate the clinical path node offset tolerance. The strategy generation unit, connected to the inclusion assessment unit, is used to compare the inclusion of clinical pathway node offset with a preset inclusion threshold. The strategy generation unit is also used to trigger a strategy generation operation to generate a dynamic reconstruction intervention strategy when the inclusion of clinical pathway node offset is greater than or equal to the preset inclusion threshold. The strategy generation unit is also used to generate a manual takeover alarm signal and terminate the generation of dynamic reconstruction intervention strategy when the inclusion of clinical pathway node offset is less than the preset inclusion threshold.
[0023] Although the previous embodiment achieved high-precision modeling of the state space, in rigorous clinical scenarios, if the system frequently reconstructs the rehabilitation path for any minor fluctuations in vital signs, it will not only cause intervention oscillations and waste medical resources, but also easily disrupt the coherence of the rehabilitation plan. Therefore, this embodiment introduces a path deviation tolerance assessment and threshold triggering mechanism based on spatial topological distance. Specifically, the topology distance calculation unit extracts the preset baseline path nodes corresponding to the current recovery day from the static knowledge base. And calculate the current node of the patient in the state space. and The Euclidean distance between them is used as the topological offset distance. The inclusion assessment unit utilizes a pre-defined maximum tolerance distance for medical safety. Divide by The clinical pathway node offset tolerance was calculated. ; The strategy generation unit will With preset tolerance threshold Verification: If the current deviation is too large, i.e., the tolerance is too small and cannot be tolerated, then the algorithm's self-generation permission must be forcibly blocked; if the calculated clinical pathway node offset tolerance is... Not only less than the preset tolerance threshold It even fell below the minimum critical safety threshold set by the system. In this case, the strategy generation unit will not only generate a manual takeover alarm signal, but also link the control gateway to forcibly lock all automated physiotherapy instruments at the bedside, and directly initiate a call to the high-priority communication terminals of the department director and the responsible nurse; for example, if the aforementioned 65-year-old patient experiences severe abdominal distension due to sudden intestinal gas obstruction on the third day after surgery, causing a rapid deterioration of the slow variable, making its current node in the state space extremely far from the baseline node of routine liquid diet and getting out of bed, the system calculates the tolerance. The value was only 0.3, far below the permissible threshold of 0.85 for triggering reconstruction. Since it exceeded the safety fallback authority of the algorithm's automatic scheduling, the system immediately stopped outputting the next stage of rehabilitation plan and sounded an alarm to request the attending physician to intervene and assess whether intestinal obstruction complications had occurred. The purpose of this mechanism is to define the authority boundary of the algorithm to dynamically modify the clinical pathway through a strict mathematical model, so as to achieve an absolute balance between intelligent system scheduling and underlying medical safety and ethical constraints.
[0024] In a preferred embodiment of the present invention, the conflict resolution and smoothing control module includes: a conflict detection unit, used to compare a first state vector and a second state vector in a multimodal rehabilitation state space, and output heterogeneous logic conflict features when the cosine of the angle between the first state vector and the second state vector is less than a preset consistency threshold; The iterative backtracking unit, connected to the conflict detection unit, is used to initiate state machine backtracking calculation in response to heterogeneous logic conflict characteristics, and record the number of iterations required to make the cosine of the angle between the first state vector and the second state vector greater than or equal to a preset consistency threshold, so as to generate the convergence steps for medical conflict resolution. The strategy correction unit, connected to the iterative backtracking unit, is used to determine whether to make an advance or delay adjustment to the execution node in the dynamic reconstruction intervention strategy based on the phase difference direction between the first state vector and the second state vector, and to determine the adjustment magnitude based on the number of convergence steps for resolving medical conflicts, so as to generate an updated dynamic reconstruction intervention strategy.
[0025] After the system was approved to trigger the strategy reconstruction, it faced the most critical medical bottleneck: the intervention logic of traditional Chinese medicine and Western medicine often diverged in certain special physiological stages. For example, Western medicine test indicators suggested that the inflammation subsided and the rehabilitation exercise load could be increased, but the pulse of traditional Chinese medicine indicated that the qi and yin were both damaged and needed to be rested and recuperated. Blindly combining elements can lead to internal repulsion in the physiological dimension; therefore, this embodiment provides a mechanism for resolving medical conflicts based on vector space angles and state machine backtracking; specifically, the conflict detection unit calculates the first state vector representing the logic of Western medicine. With the second state vector representing the logic of Traditional Chinese Medicine cosine value of the included angle ;when Preset consistency threshold When the time comes, determine and output the heterogeneous logic conflict characteristics; The iterative backtracking unit then starts the state machine, fine-tuning the time step of the intervention node in the background and simulating its correction effect on the two types of state vectors. Specifically, the simulation process is as follows: a pre-built multimodal state transition prediction model based on a long short-term memory network is invoked, and the dynamically reconstructed intervention strategy sequence after fine-tuning the time steps is used as the input variable to predict the future continuous state transitions. The first and second state vector prediction sequences at discrete time steps are used to generate the future state vector sequence of the patient population by weighted fusion of the prediction sequences. ; The system records the values in the predicted future state vector sequence that make Minimum number of loop iterations required The aforementioned predictive model not only enables a forward-looking quantitative simulation of the intervention's effects but also generates a sequence of future state vectors. This also provides complete basic data support for subsequent calculations of discrete difference change rate and state transition smoothness; the strategy correction unit, based on the phase difference direction between vectors, such as the progress of TCM recovery lagging behind Western medicine indicators, decides whether to advance the physiotherapy node or postpone the rehabilitation training node, and according to... The time step for quantization adjustment; specifically, the system has a built-in preset base time step. Adjustment time step The calculation formula is: If during the state machine backtracking calculation process, the number of iterations... If the maximum computational depth limit is reached but the cosine of the angle still fails to converge to a consistency threshold, it means that an irreconcilable underlying medical conflict has occurred. The system will abandon the reconstruction of the dynamic intervention strategy and safely revert to an extremely conservative backup intervention strategy that only includes basic anti-inflammatory and fluid replacement, and record a logic non-convergence log. For example, in the early postoperative period, the system detected a serious conflict between the Western medicine immediate semi-liquid diet and high-frequency walking nodes and the traditional Chinese medicine state of unresolved damp-heat and extremely weak spleen and stomach function, with the vector angle approaching orthogonal. After the system reached convergence by backtracking through the state machine for 8 steps, the strategy correction unit decided to postpone the execution time of Western medicine's getting out of bed and walking by 12 hours, while significantly advancing the TCM heat-clearing and dampness-removing herbal packing therapy node, thus generating a set of update strategies that compromise between Chinese and Western medicine and provide a stable guarantee. The purpose of this step is to solve the problem of nonlinear coupling and mutual exclusion when combining heterogeneous intervention measures, so as to achieve absolute self-consistency of interdisciplinary rehabilitation strategies in the underlying physiological operating logic.
[0026] In a preferred embodiment of the present invention, the conflict resolution and smoothing control module further includes: a smoothness calculation unit connected to the strategy correction unit, used to calculate the discrete difference rate of change of the state vector in the multimodal rehabilitation state space before and after the execution of the updated dynamic reconstruction intervention strategy based on discrete time steps, and generate the smoothness of the rehabilitation trajectory state transition. The strategy output and management module is used to compare the smoothness of the rehabilitation trajectory state transition with a preset smoothness threshold; the strategy output and management module is also used to confirm the updated dynamic reconstruction intervention strategy as the final intervention strategy sequence when the smoothness of the rehabilitation trajectory state transition is greater than or equal to the preset smoothness threshold; the strategy output and management module is also used to send a readjustment instruction to the strategy correction unit when the smoothness of the rehabilitation trajectory state transition is less than the preset smoothness threshold.
[0027] While the previous embodiment successfully resolved the conflict between traditional Chinese and Western medicine theories at the abstract logical level, the use of temporal shifting in the adjustment could easily lead to a large stacking of intervention nodes within a local time period. For example, if the morning is all physiotherapy and the afternoon is all physical training, this would cause a technical defect at the physical execution level: a drastic jump in intervention intensity, which could easily lead to a precipitous deterioration of the vital signs of weak patients. This embodiment introduces a predictive and closed-loop feedback mechanism for the smoothness of the intensity transition of the rehabilitation trajectory. Specifically, the smoothness calculation unit utilizes discrete time steps... Calculate the second-order discrete difference rate of change of the patient's overall state vector sequence after the simulated execution of the update strategy. Based on this, a recovery trajectory state transition smoothness inversely proportional to the rate of change is generated. ; If the strategy output and management module determines that... Preset smoothness threshold If this indicates that the intervention pace is mild and controllable, then a confirmation output should be given; if If the execution intensity change is too steep, it indicates that the strategy output and management module sends readjustment instructions to the strategy correction unit more than the preset maximum iteration adjustment threshold. If the readjustment instructions are triggered four times in a row and still fail to meet the standard, the system will forcibly activate the peak-shaving and valley-filling fallback logic, hard-eliminating or significantly weakening the secondary intervention nodes with peak intensity within the local time window, such as canceling certain physical labor-intensive assessment actions, in exchange for absolute physiological smoothness at the cost of sacrificing a little recovery speed. For example, the new strategy planned by the system for this elderly patient originally required low-frequency pulse analgesia, limb linkage device training and traditional Chinese medicine acupuncture stimulation in succession on the afternoon of the fourth day after surgery due to node delay. After algorithm pre-simulation, it was found that this high-density scheduling would cause the state vector difference change rate representing the patient's physical labor consumption to soar sharply and the smoothness to plummet. The system reorganized and combined these elements, ultimately halving the equipment training time and forcibly dispersing it to the following morning, ensuring a smooth recovery trajectory. This allowed for physical boundary control over the intensity of intervention strategy implementation, preventing drastic fluctuations in the patient's vital signs due to sudden changes or excessive concentration of intervention strategies.
[0028] In a preferred embodiment of the present invention, the multimodal data acquisition module includes: a first acquisition unit for acquiring drainage fluid property data through a preset biochemical detection device; a second acquisition unit for acquiring pain score data through a preset interactive terminal; a third acquisition unit for acquiring tongue feature data through a preset image sensor; and a fourth acquisition unit for acquiring pulse feature data through a preset pressure sensor.
[0029] The aforementioned embodiments are all based on high-level algorithm deductions of multi-source data that have been fully acquired; however, in complex real ward environments, without the underlying hardware support with clear physical mounting attributes and dedicated data channels, the so-called multimodal rehabilitation state space will not be effectively applied in the actual physical environment due to the lack of underlying data support; therefore, this embodiment has made specific engineering definitions for the physical topology of data acquisition and the selection of various sensor devices. Specifically, the first acquisition unit uses biochemical testing equipment deployed on the side of the bedside drainage line to obtain continuous volume and color transmittance data; the second acquisition unit uses an interactive terminal fixed at the patient's bedside to obtain quantitative pain score data. The third acquisition unit uses an image sensor integrated on the nurse rounds cart or the top of the ward to obtain objective tongue images free from ambient light pollution; the fourth acquisition unit continuously acquires pulse waveform features with varying depths through a pressure sensor attached to the patient's wrist at the Cun, Guan, and Chi positions. If any of the acquisition units detects continuous frame loss of data packets or impedance abnormalities exceeding the preset tolerance time, the acquisition module will automatically cut off the source of the distorted data stream and enable the feature value freezing and confidence smoothing attenuation processing mechanism. That is, the spatial coordinates are maintained continuously using the most recent valid waveform data in history, but the weight ratio of the data in the subsequent fusion model is gradually reduced according to the time attenuation function. For example, during the transition from routine daytime monitoring to nighttime rest after surgery, the biochemical detection equipment converts the clarity of the peritoneal drainage fluid into an electrical signal and reports it in real time. At the same time, the pressure sensor captures the subtle physiological transition of the pulse from tight and rough to gentle without disturbing the patient's sleep. This provides objective, real and high-frequency basic data support for the subsequent adaptive tracking algorithm. The purpose of this mechanism is to solidify the information acquisition foundation of the entire intelligent rehabilitation system, so as to achieve accurate, imperceptible and standardized entry of multi-source heterogeneous medical feature data containing subjective and objective dimensions into the system.
[0030] In a preferred embodiment of the present invention, the multimodal data acquisition module, the state space modeling module, the path adaptive tracing module, the conflict resolution and smoothing control module, and the strategy output and management module are integrated into a preset medical edge computing gateway; the medical edge computing gateway interacts with a preset hospital information system through a preset secure communication protocol.
[0031] With the large-scale introduction of bedside high-frequency acquisition devices, if all such massive amounts of multimodal raw data are transmitted to the central cloud server for state space modeling and matrix deduction, the hospital's internal core network will face extremely high risk of response delay once congestion or fluctuations occur, which may delay the reconstruction of instantaneous intervention strategies for sudden postoperative lesions. This embodiment provides a distributed local hard real-time processing architecture based on cloud-edge collaboration; specifically, core computationally intensive modules with extremely high real-time requirements, such as data acquisition and cleaning, high-dimensional state modeling, adaptive trajectory finding, and conflict resolution, are all solidified and executed within a medical edge computing gateway deployed in the weak current room of an independent ward. The edge gateway only pushes the final purified intervention strategy sequence or lightweight desensitized state vector features to the remote hospital information system through preset medical security communication protocols such as HL7 and FHIR for long-term electronic medical record archiving and medical order accounting synchronization. If the heartbeat monitoring process of the medical edge computing gateway detects an unexpected disconnection of the secure communication link between it and the preset hospital information system, the gateway will immediately implement communication degradation and isolation, blocking all external pull requests. Instead, it will use the NVMe solid-state storage array built into the gateway motherboard to temporarily store all dynamically generated intervention strategy execution logs and send control commands directly to the bedside infusion pump or physiotherapy device only through the local physical serial port. After the network link is restored, it will perform breakpoint resume and HIS state machine synchronization based on timestamps. For example, when a 65-year-old patient suddenly experiences short, paroxysmal biliary spasms causing severe pain, the medical edge computing gateway on the local ward side independently completes the emergency reorganization of the state space and smoothness calculation within milliseconds. It directly issues preliminary instructions for basic analgesia and acupoint relief to the bedside automated drug delivery system and physiotherapy equipment, without waiting for the potentially fatal delays of a few seconds caused by network queuing in the preset hospital information system. The purpose of this architecture is to sink the core computing power and physically isolate unreliable external network factors, thereby achieving high availability, extremely low latency, and localized closed-loop management that complies with medical data compliance and privacy regulations in the dynamic reconstruction of clinical pathways in critical care rehabilitation scenarios.
[0032] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.
Claims
1. A postoperative rehabilitation evaluation and management system for cholelithiasis based on an integrated traditional Chinese and Western medicine clinical pathway, characterized in that, The system is integrated into a medical edge computing gateway and includes, in sequence, a multimodal data acquisition module for acquiring fast variable objective index data and slow variable feature data of a preset target object. The fast variable objective index data includes at least one of drainage fluid characteristics data and pain score data, and the slow variable feature data includes at least one of pulse characteristic data and tongue characteristic data. The state space modeling module is used to construct a multimodal rehabilitation state space based on the time-aligned multimodal data, and to calculate the weighted dynamic adjustment rate of the fast and slow variables fused as heterogeneous evaluation data. The path adaptive tracking module is used to calculate the offset tolerance degree of the current node relative to the preset baseline clinical path node based on the multimodal rehabilitation state space and the weight dynamic adjustment rate, and generate a dynamic reconstruction intervention strategy containing multiple execution nodes, including traditional Chinese medicine physiotherapy nodes and Western medicine rehabilitation training nodes. The conflict resolution and smoothing control module is used to extract heterogeneous logic conflict features based on the angle between the state vectors representing the intervention logic of traditional Chinese and Western medicine in the multimodal rehabilitation state space, and calculate the number of convergence steps for resolving medical conflict to update the dynamic reconstruction intervention strategy, and calculate the smoothness of rehabilitation trajectory state transition based on the updated dynamic reconstruction intervention strategy. The strategy output and management module is used to filter the updated dynamic reconstruction intervention strategy based on the smoothness of the rehabilitation trajectory state transition, and output the final intervention strategy sequence, which includes the sequence of traditional Chinese medicine physiotherapy and Western medicine rehabilitation training nodes.
2. The postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway according to claim 1, characterized in that, The state space modeling module includes a time alignment unit, used to resample the fast variable objective index data and the slow variable feature data according to a preset time base to generate time-aligned multimodal data; The weight calculation unit, connected to the time alignment unit, is used to extract nonlinear coupling features by calculating the mutual information entropy between the time series of fast variable objective index data and slow variable feature data in the time-aligned multimodal data, and to calculate the weight dynamic adjustment rate of the heterogeneous evaluation data fusion based on the nonlinear coupling features. The spatial mapping unit, connected to the weight calculation unit, is used to project the fast variable objective index data and slow variable feature data in the time-aligned multimodal data to a high-dimensional space using a preset mapping matrix to generate a first state vector and a second state vector. The first state vector and the second state vector are then weighted and superimposed using the weight dynamic adjustment rate of the heterogeneous evaluation data fusion to generate the current node. The multimodal rehabilitation state space is constructed based on the current node, and a multimodal state transition prediction model is pre-set in the multimodal rehabilitation state space.
3. The postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway according to claim 2, characterized in that, The path adaptive tracing module includes a topology distance calculation unit, which is used to obtain preset baseline path nodes in a pre-stored standard rehabilitation model from a preset static knowledge base, calculate the distance between the current node and the preset baseline path node in the multimodal rehabilitation state space, and output the topology offset distance. The tolerance assessment unit, connected to the topology distance calculation unit, is used to divide the preset maximum tolerance distance by the topology offset distance when the topology offset distance is not zero, so as to generate the tolerance of the clinical pathway node offset. A strategy generation unit, connected to the inclusion assessment unit, is used to compare the clinical pathway node offset inclusion with a preset inclusion threshold. The strategy generation unit is also used to trigger a strategy generation operation to generate the dynamic reconstruction intervention strategy when the clinical pathway node offset tolerance is greater than or equal to the preset tolerance threshold. The strategy generation unit is also used to generate a manual takeover alarm signal and terminate the generation of the dynamic reconstruction intervention strategy when the clinical pathway node offset tolerance is less than the preset tolerance threshold.
4. The postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway according to claim 3, characterized in that, The conflict resolution and smoothing control module includes: The conflict detection unit is used to compare the first state vector and the second state vector in the multimodal rehabilitation state space, and output the heterogeneous logic conflict feature when the cosine of the angle between the first state vector and the second state vector is less than a preset consistency threshold. An iterative backtracking unit, connected to the conflict detection unit, is used to initiate state machine backtracking calculation in response to the heterogeneous logic conflict feature, and record the number of iterations required to make the cosine of the angle between the first state vector and the second state vector greater than or equal to the preset consistency threshold, so as to generate the convergence steps for resolving the medical conflict. The strategy correction unit, connected to the iterative backtracking unit, is used to determine whether to adjust the execution nodes in the dynamic reconstruction intervention strategy in advance or delay them based on the phase difference direction between the first state vector and the second state vector, i.e. the deflection direction representing the leading or lagging relationship of recovery progress, and to determine the adjustment magnitude based on the number of convergence steps of the medical conflict resolution, so as to generate the updated dynamic reconstruction intervention strategy.
5. The postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway according to claim 4, characterized in that, The conflict resolution and smoothing control module further includes: a smoothness calculation unit, connected to the strategy correction unit, used to calculate the discrete difference rate of change of the state vector in the multimodal rehabilitation state space before and after the simulation execution of the updated dynamic reconstruction intervention strategy based on discrete time steps, and generate the smoothness of the rehabilitation trajectory state transition. The strategy output and management module is used to compare the smoothness of the rehabilitation trajectory state transition with a preset smoothness threshold; the strategy output and management module is also used to confirm the updated dynamic reconstruction intervention strategy as the final intervention strategy sequence when the smoothness of the rehabilitation trajectory state transition is greater than or equal to the preset smoothness threshold. The strategy output and management module is also used to send a readjustment instruction to the strategy correction unit when the smoothness of the recovery trajectory state transition is less than the preset smoothness threshold.
6. The postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway according to claim 1, characterized in that, The multimodal data acquisition module includes: The first acquisition unit is used to acquire the property data of the drainage fluid through a preset biochemical detection device; The second acquisition unit is used to acquire the pain score data through a preset interactive terminal; The third acquisition unit is used to acquire the tongue image feature data through a preset image sensor; The fourth acquisition unit is used to acquire the pulse characteristic data through a preset pressure sensor.
7. The postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway according to claim 6, characterized in that, The multimodal data acquisition module, state space modeling module, path adaptive tracing module, conflict resolution and smoothing control module, and strategy output and management module are integrated into a pre-set medical edge computing gateway. The medical edge computing gateway interacts with a pre-set hospital information system through a preset secure communication protocol.
8. The postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway according to claim 2, characterized in that, In the weight calculation unit, the formula for calculating the dynamic adjustment rate of the weights for the fusion of heterogeneous evaluation data is as follows: ;in, The weight dynamic adjustment rate; This is the preset sensitivity adjustment coefficient; The mutual information entropy between the time series of fast variable objective indicator data and slow variable characteristic data; This is the preset threshold for mutual information entropy.
9. The postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway according to claim 4, characterized in that, In the strategy correction unit, the formula for calculating the adjustment magnitude is: in, The time step for the adjustment; The number of convergence steps is used to resolve conflicts in medical theory. The system's built-in preset base time step.
10. The postoperative rehabilitation evaluation and management system for cholelithiasis based on the integrated traditional Chinese and Western medicine clinical pathway according to claim 7, characterized in that, When the heartbeat monitoring process of the medical edge computing gateway detects an unexpected disconnection of the secure communication link with the preset hospital information system, the medical edge computing gateway implements communication degradation isolation, temporarily stores dynamically generated intervention strategy execution logs using the built-in solid-state storage array, and directly sends control commands to external preset devices through the local physical serial port.