Special disease first-aid full-process auxiliary decision-making method based on multi-modal perception and AI

Abstract

The invention provides a special disease first-aid full-process auxiliary decision-making method based on multi-modal perception and AI, and relates to the technical field of medical first aid, and the method comprises the steps: obtaining first-aid field multi-modal data, constructing a feature dependence graph, and recognizing the type and symptom features of a special disease; indexing special disease diagnosis and treatment data based on the symptom feature vector, and determining a reference case and a treatment path; the action space is constructed in combination with clinical specifications, the expected income is predicted by applying Monte Carlo search, and the optimal treatment scheme is generated, so that the first-aid special disease recognition accuracy can be improved, the treatment decision time can be shortened, and the first-aid intervention effect can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical first aid, and in particular to a disease-specific first aid whole-process auxiliary decision-making method based on multi-modal perception and AI. BACKGROUND

[0002] First aid medicine is an important part of the medical service system, and rapid and accurate treatment of disease-specific first aid plays a decisive role in patient prognosis. With the development of information technology, multi-modal perception devices and artificial intelligence technology are increasingly widely used in the medical field, bringing new technical support to first aid medicine. With the application of data-driven decision support systems in first aid scenes, multi-modal information such as patient physiological data and environmental information can be collected, and artificial intelligence algorithms can provide assistance for first aid decision-making. However, the prior art still has problems such as only focusing on single modal data, lacking comprehensive analysis of multi-modal data, being difficult to fully grasp the patient's state, being difficult to effectively capture the temporal correlation between symptom evolution rules and characteristics, being unable to provide targeted support for first aid decision-making at different stages, lacking deep learning and optimization of historical cases, being unable to adaptively adjust treatment plans according to individual differences and dynamic changes of patients, lacking intelligent decision optimization mechanisms, and being difficult to provide precise personalized treatment decisions in complex first aid scenarios. SUMMARY

[0003] The present application provides a disease-specific first aid whole-process auxiliary decision-making method based on multi-modal perception and AI, which can at least solve some of the problems in the prior art.

[0004] In a first aspect, the present application provides a disease-specific first aid whole-process auxiliary decision-making method based on multi-modal perception and AI, comprising: Obtaining multi-modal data of a first aid scene and performing feature extraction to obtain an initial feature vector, calculating the temporal changes of the initial feature vector and constructing a feature dependency graph combining conditional probability, calculating the temporal correlation between feature nodes in the feature dependency graph, determining a key feature sequence based on the temporal correlation and identifying a disease type and corresponding symptom feature vector; Based on the symptom feature vector, the pre-set disease diagnosis and treatment data is indexed, a to-be-indexed subset is determined, the initial similarity between the symptom feature vector in the to-be-indexed subset and the historical cases is calculated, a candidate case is determined, the candidate case is analyzed to determine the symptom evolution rule, the treatment matching degree between the current physiological data and the symptom evolution rule is calculated, a reference case is determined based on the treatment matching degree, and the treatment path corresponding to the reference case is extracted as a reference path; An initial treatment scheme is generated based on the reference path, and an initial action space is constructed in combination with clinical norms; a state vector is constructed in combination with the current physiological data; a treatment action is selected in the initial action space, and an immediate reward value is calculated; an expected return is predicted in combination with Monte Carlo search; and the initial treatment scheme is updated based on the predicted expected return to obtain an optimal treatment scheme.

[0005] In an optional implementation, Multi-modal data of the emergency scene is acquired, and initial feature vectors are obtained through feature extraction; time series changes of the initial feature vectors are calculated, and a feature dependency graph is constructed in combination with conditional probabilities, including: The multi-modal data of the emergency scene is subjected to feature extraction to obtain initial feature vectors, first-order differential values of the initial feature vectors are calculated, and inflection points are determined based on the first-order differential values and a preset adaptive threshold; The initial feature vectors are divided into a plurality of feature segments according to the inflection points, each feature segment is modeled, and a segmented feature function is constructed; a time-varying weight transition connection function is constructed between adjacent feature segments based on the segmented feature function to obtain a time series change sequence; Conditional probabilities are calculated based on the time series change sequence; when the conditional probabilities exceed a preset dynamic threshold, directed edges are established between corresponding feature nodes; and the directed edges are repeatedly constructed to obtain the feature dependency graph.

[0006] In an optional implementation, The time series correlation degrees between feature nodes in the feature dependency graph are calculated, key feature sequences are determined based on the time series correlation degrees, and a disease type and a corresponding symptom feature vector are identified, including: For each feature node in the feature dependency graph, a PageRank centrality is calculated in combination with neighbor nodes of the current feature node, a connection probability between the current feature node and the neighbor nodes is calculated to determine an information entropy, and an importance weight of the current feature node is calculated based on the PageRank centrality and the information entropy; Distances between different feature nodes are calculated, and a node similarity is determined; an initial correlation degree is calculated based on the node similarity and the importance weight; and a time series correlation degree is obtained through time series adjustment based on the initial correlation degree and a preset time series decay factor; A feature mapping relationship corresponding to each disease type is extracted from a preset disease type knowledge base; feature nodes with a time series correlation degree greater than a preset correlation degree threshold are taken as key nodes; a combination correlation between the key nodes is analyzed, and the key feature sequences are constructed in combination with the feature mapping relationship; a matching degree between each feature in the key feature sequences and a disease type is calculated to obtain a disease type probability value, and a disease type corresponding to the current feature is determined; and a symptom feature vector is obtained by extracting a time series correlation degree corresponding to the disease type.

[0007] In an alternative embodiment, The partition indexing of the pre-set disease diagnosis and treatment data based on the symptom feature vector, the determination of the to-be-indexed subset, and the calculation of the initial similarity between the symptom feature vector in the to-be-indexed subset and the historical cases include: The difference between each symptom feature value in the symptom feature vector and the pre-set normal reference mean value is calculated to determine the abnormality degree of each symptom, and the corresponding risk level of each symptom is determined based on the abnormality degree; The symptom feature vector is divided into multiple priority partitions based on the risk level, and a first warning threshold is solved based on the risk level and a pre-set first balance parameter. For the symptom feature vector in each priority partition, a second warning threshold is solved by calculating the change trend and combining a pre-set second balance parameter. A combined warning value is obtained based on the first warning threshold and the second warning threshold, and an index structure is constructed based on the combined warning value and the symptom feature vector; Based on the index structure, the to-be-matched partitions are located from the disease diagnosis and treatment data, the partition matching degrees between the symptom feature vector and each to-be-matched partition are calculated, and the to-be-indexed subset is combined by selecting three to-be-matched partitions with the highest partition matching degrees; The Euclidean distance between the symptom feature vector and each historical case in the to-be-indexed subset is calculated to determine the feature similarity, the timeliness weight is calculated based on the time information of the historical cases, and the initial similarity is calculated by combining the feature similarity.

[0008] In an alternative embodiment, The candidate cases are determined, the time sequence analysis is performed on the candidate cases to determine the symptom evolution rule, the treatment matching degree between the current physiological data and the symptom evolution rule is calculated, the reference case is determined based on the treatment matching degree, and the treatment path corresponding to the reference case is extracted as the reference path. The candidate cases are determined in the to-be-indexed subset based on the initial similarity, the time sequence analysis is performed on the candidate cases by the dynamic time warping method, the time sequence distance corresponding to each candidate case is calculated, and the symptom evolution rule is determined based on the time sequence distance; The current physiological data is collected, the disease condition complexity matching value and the medical resource matching degree between the current physiological data and the symptom evolution rule are calculated and summed to obtain the treatment matching degree; selecting a candidate case with the highest treatment matching degree as a reference case, constructing a treatment optimization objective function according to a complication risk value, solving the treatment optimization objective function by a differential evolution algorithm to obtain an optimal solution, extracting a feasibility score corresponding to the optimal solution, and if the feasibility score is greater than a pre-set feasibility threshold, extracting a treatment path and outputting the treatment path as a reference path.

[0009] In an optional implementation, generating an initial treatment scheme based on the reference path and constructing an initial action space in combination with clinical norms, and constructing a state vector in combination with the current physiological data includes: performing time series decomposition on the reference path to obtain a plurality of treatment nodes, extracting treatment measures in each treatment node to construct an initial treatment scheme, constructing an initial action space in combination with pre-acquired clinical norms, and optimizing constraint parameters in the initial action space by an adaptive fuzzy reasoning method to obtain optimized constraint parameters; constructing a fuzzy decision tree based on the optimized constraint parameters, performing rationality evaluation on the initial action space based on the fuzzy decision tree to obtain a rationality score, and initializing a feature mapping matrix based on the rationality score; mapping pre-acquired current physiological data into a feature space corresponding to the optimized constraint parameters based on the feature mapping matrix to obtain a state vector.

[0010] In an optional implementation, selecting a treatment action in the initial action space and calculating an immediate reward value, predicting an expected return based on the predicted expected return, and updating the initial treatment scheme to obtain an optimal treatment scheme includes: collecting treatment actions in the initial action space and determining time series dependency relationships between different treatment actions, constructing an action transition graph based on the time series dependency relationships and the treatment actions; determining an action vertex set and a dependency relationship edge set based on the action transition graph, calculating a time series constraint function according to the dependency relationship edge set, and adding the state vector to the time series constraint function to calculate a predicted state transition result; calculating a state improvement degree of the predicted state transition result relative to the state vector, calculating an immediate reward value in combination with treatment resource consumption, performing Monte Carlo search on the immediate reward value, calculating a confidence upper limit interval and selecting a search node to obtain the predicted expected return; adding the predicted state transition result and the predicted expected return to a pre-set state-action value function and iteratively optimizing the state-action value function by a time series difference algorithm, updating the initial treatment scheme according to an optimization result to obtain an optimal treatment scheme.

[0011] In a second aspect of the embodiment of the present application, a multi-modal perception and AI-based special disease emergency whole-process auxiliary decision system is provided, comprising: A first unit is configured to acquire multi-modal data of an emergency scene and perform feature extraction to obtain an initial feature vector, calculate a time sequence change of the initial feature vector, and construct a feature dependency graph in combination with a conditional probability, calculate a time sequence correlation degree between feature nodes in the feature dependency graph, determine a key feature sequence based on the time sequence correlation degree, and identify a special disease type and a corresponding symptom feature vector; A second unit is configured to perform partition indexing on pre-set special disease diagnosis and treatment data based on the symptom feature vector, determine a to-be-indexed subset and calculate an initial similarity between the symptom feature vector in the to-be-indexed subset and a historical case, determine a candidate case and perform time sequence analysis on the candidate case to determine a symptom evolution rule, calculate a treatment matching degree between current physiological data and the symptom evolution rule, determine a reference case based on the treatment matching degree, and extract a treatment path corresponding to the reference case as a reference path; A third unit is configured to generate an initial treatment scheme based on the reference path, construct an initial action space in combination with a clinical standard, construct a state vector in combination with the current physiological data, select a treatment action in the initial action space and calculate an immediate reward value, predict an expected return in combination with a Monte Carlo search, and update the initial treatment scheme based on the predicted expected return to obtain an optimal treatment scheme.

[0012] In a third aspect of the embodiment of the present application, an electronic device is provided, comprising: A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0013] In a fourth aspect of the embodiment of the present application, a computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0014] In the present application, through multi-modal data perception and time sequence feature analysis, accurate identification of special disease types and effective extraction of symptom features are realized, the diagnosis accuracy of the emergency scene is effectively improved, the problem that the traditional method cannot comprehensively capture complex symptom manifestations is solved, the partition indexing strategy based on the symptom feature vector and the time sequence evolution analysis method are adopted, similar historical cases can be quickly located and the treatment matching degree can be accurately evaluated, the emergency decision-making time is greatly shortened, the emergency efficiency is improved, the clinical relevance and individual applicability of the treatment scheme are ensured, the treatment scheme is optimized and the patient state changes are adapted through immediate reward calculation and Monte Carlo search prediction of expected return, the timeliness and accuracy of emergency intervention are significantly improved, and the survival rate and prognosis quality of patients are improved. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 A flowchart of an embodiment of the application is shown in FIG. 1. The method comprises the following steps: Figure 2 A flowchart of an embodiment of the application is shown in FIG. 1. The method comprises the following steps: DETAILED DESCRIPTION

[0016] To make the objects, technical solutions and advantages of the embodiments of the application clearer, the technical solutions in the embodiments of the application will be described below in a clear and complete manner with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the application.

[0017] The technical solutions of the application will be described in detail below with reference to specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes can not be described in some embodiments.

[0018] Figure 1 A flowchart of an embodiment of the application is shown in FIG. 1. The method comprises the following steps: Figure 1 As shown in FIG. 1, the method comprises the following steps: obtaining multi-modal data of an emergency scene and performing feature extraction to obtain an initial feature vector, calculating a time sequence change of the initial feature vector and constructing a feature dependency graph in combination with a conditional probability, calculating a time sequence correlation degree between feature nodes in the feature dependency graph, determining a key feature sequence based on the time sequence correlation degree and identifying a special disease type and a corresponding symptom feature vector; performing partition indexing on pre-set special disease diagnosis and treatment data based on the symptom feature vector, determining a to-be-indexed subset and calculating an initial similarity of the symptom feature vector in the to-be-indexed subset to a historical case, determining a candidate case and performing time sequence analysis on the candidate case to determine a symptom evolution rule, calculating a treatment matching degree of current physiological data to the symptom evolution rule, determining a reference case based on the treatment matching degree and extracting a treatment path corresponding to the reference case as a reference path; generating an initial treatment scheme based on the reference path and constructing an initial action space in combination with clinical norms, constructing a state vector in combination with the current physiological data, selecting a treatment action in the initial action space and calculating an immediate reward value, predicting an expected return in combination with Monte Carlo search, and updating the initial treatment scheme based on the predicted expected return to obtain an optimal treatment scheme.

[0019] In an alternative embodiment, Obtaining multi-modal data of the emergency scene and performing feature extraction to obtain an initial feature vector, calculating the time series change of the initial feature vector and constructing a feature dependency graph combining conditional probability comprises: Performing feature extraction on multi-modal data of the emergency scene to obtain an initial feature vector, calculating the first-order difference value of the initial feature vector, and determining the inflection point based on the first-order difference value and a preset adaptive threshold value; According to the inflection point, the initial feature vector is divided into a plurality of feature segments, each feature segment is modeled and a segmented feature function is constructed, and a transition connection function of time-varying weight between adjacent feature segments is constructed based on the segmented feature function to obtain a time series change sequence; Based on the time series change sequence, the conditional probability is calculated, and when the conditional probability exceeds a preset dynamic threshold value, a directed edge is established between the corresponding feature nodes, and the directed edge is repeatedly constructed to obtain the feature dependency graph.

[0020] Multi-modal data of the emergency scene is collected, including physiological signals of the patient such as electrocardiogram, blood pressure, respiration, etc., image data such as ultrasound, tomography, etc., voice data such as patient symptom description, and video data such as patient expression, limb state, etc. Feature extraction algorithm is used to obtain initial feature vector. For physiological signal data, wavelet transform is used to extract time-frequency features; for image data, convolutional neural network is used to extract spatial features; for voice data, mel frequency cepstral coefficient feature extraction is used; for video data, dynamic features are obtained by combining optical flow method and depth feature extraction technology. The extracted features are integrated into an initial feature vector, represented as a multi-dimensional array, each dimension corresponding to a specific feature attribute.

[0021] After obtaining the initial feature vector, the first-order difference value of the vector is calculated. The first-order difference value reflects the change rate of the feature in the time dimension, and the calculation method is the difference value between adjacent time point feature values. For example, for the feature vector representing the patient's heart rate [120, 123, 128, 125, 122, 118, 115], the first-order difference value is [3, 5, -3, -3, -4, -3]. An adaptive threshold mechanism is introduced here, which dynamically adjusts the threshold value according to the historical statistical distribution of feature changes. The adaptive threshold is determined by calculating the weighted combination of the standard deviation and mean value of the historical data, and is updated as the data is continuously input. When the first-order difference value exceeds the adaptive threshold, the point is marked as an inflection point. In the foregoing example, if the adaptive threshold is ±4, the positions with difference values of 5 and -4 are marked as inflection points, corresponding to the 3rd and 5th elements in the original feature vector.

[0022] According to the determined inflection point, the initial feature vector is divided into multiple feature segments, and the original feature vector is divided into three segments: [120, 123, 128], [128, 125, 122, 118], and [118, 115]. Each feature segment is modeled and a segment feature function is constructed. The construction of the segment feature function adopts a piecewise linear regression or a polynomial fitting method, and the most suitable modeling method is selected according to the data characteristics within the segment. For heart rate data, it is assumed that the piecewise linear regression method is adopted, and the feature function of the first segment can be expressed as an increasing relationship, the second segment as a decreasing relationship, and the third segment also as a decreasing relationship.

[0023] A transition connection function with time-varying weights is constructed between adjacent feature segments to achieve smooth transition. The transition connection function adopts a weight decay strategy to make the adjacent segments smoothly transition at the connection point. In specific implementation, a time window size is defined, and within the window, the weight distribution ratio is gradually adjusted over time. For example, when transitioning from the first segment to the second segment, within the window near the inflection point, the closer the time is to the inflection point, the greater the weight of the second segment feature function and the smaller the weight of the first segment feature function. The time-varying weight transition connection function ensures the continuity of feature changes when transitioning between segments, and obtains a complete time series change sequence.

[0024] Based on the obtained time series change sequence, the conditional probability is calculated. In a first aid scenario, for example, given that the patient's heart rate is rapidly decreasing, the probability that the blood pressure will also decrease. The calculation of the conditional probability considers the feature change trend within multiple time windows, and extracts the probability distribution from historical data through statistical learning methods. For patient data of acute myocardial infarction, it can be concluded that the conditional probability between heart rate decrease and blood pressure decrease is 0.85, indicating that in the case of rapid heart rate decrease, there is an 85% probability that blood pressure will also decrease.

[0025] A dynamic threshold is set to judge the significance of the conditional probability. The dynamic threshold is dynamically adjusted according to the first aid situation and the importance of the features, and a lower threshold is set for key vital sign indicators and a higher threshold is set for secondary indicators. When the calculated conditional probability exceeds the preset dynamic threshold, a directed edge is established between the corresponding feature nodes. If the conditional probability between heart rate and blood pressure is 0.85, which exceeds the set dynamic threshold of 0.75, a directed edge is established between the heart rate node and the blood pressure node, indicating that heart rate changes may cause blood pressure changes. By repeatedly constructing directed edges between multiple feature nodes, a complete feature dependency graph is formed.

[0026] In this embodiment, by extracting features from multi-modal data of emergency scenes and introducing first-order difference and adaptive threshold to identify inflection points, the mutation points of features over time can be captured more sensitively, and key abnormalities can be avoided. By dividing the initial feature vector into multiple segments and constructing a segmented feature function, and then using a time-varying weight transition connection function for modeling, the dynamic transition relationship between different feature stages can be more accurately reflected, and the time sequence adaptability is higher. By calculating the conditional probability based on the time sequence change sequence and combining the dynamic threshold to establish a feature dependency graph, the dynamic dependency relationship between features can be effectively identified, and the accuracy and reliability of multi-modal feature correlation mining in complex scenes are improved.

[0027] In an optional implementation, The time sequence correlation degree between the feature nodes in the feature dependency graph is calculated, and the key feature sequence, the special disease type, and the corresponding symptom feature vector are determined based on the time sequence correlation degree, including: For each feature node in the feature dependency graph, the PageRank centrality is calculated in combination with the neighbor nodes of the current feature node, the connection probability between the current feature node and the neighbor nodes is calculated to determine the information entropy, and the importance weight of the current feature node is calculated based on the PageRank centrality and the information entropy; The distance between different feature nodes is calculated, and the similarity between nodes is determined, the initial correlation degree is calculated based on the similarity between nodes and the importance weight, and the time sequence correlation degree is obtained by time sequence adjustment based on the initial correlation degree and a pre-set time sequence decay factor; The feature mapping relationship corresponding to each special disease type is extracted in a pre-set special disease type knowledge base, the feature nodes with a time sequence correlation degree greater than a pre-set correlation degree threshold are taken as key nodes, the combination correlation between the key nodes is analyzed, and the key feature sequence is constructed in combination with the feature mapping relationship, the matching degree between each feature in the key feature sequence and the special disease type is calculated to obtain a special disease type probability value, and the special disease type corresponding to the current feature is determined, and the symptom feature vector is obtained by extracting the time sequence correlation degree corresponding to the special disease type.

[0028] For each feature node in the feature dependency graph, the PageRank algorithm is used to calculate the centrality, and the neighbor nodes of each feature node are traversed, and the number of neighbor nodes and their own centrality values are considered comprehensively. Taking the feature dependency graph of an acute myocardial infarction patient as an example, the electrocardiogram ST segment elevation node is connected to multiple important physiological indicator nodes, such as heart rate, blood pressure, etc., and the PageRank centrality value calculation result is 0.28, while the ordinary pain indicator node is only connected to two secondary nodes, and the centrality value is only 0.05.

[0029] The connection probability between the current feature node and its neighbor nodes is calculated to reflect the uncertainty degree of the relationship between the nodes, and then the information entropy is determined. The connection probability is obtained by normalizing the weight of the edge between the nodes, and the information entropy is calculated by the weighted sum of the connection probability. For example, the connection probabilities of the heart rate node and its three neighbor nodes are 0.5, 0.3 and 0.2, respectively, and the information entropy is calculated to be 1.03. The higher the information entropy, the more complex and variable the connection relationship of the node is, and there is more uncertainty in clinical judgment.

[0030] The importance weight of the feature node is calculated based on the PageRank centrality and the information entropy, and the centrality value and the reciprocal of the information entropy are weighted and combined. The weight coefficient is set according to the specific needs of the emergency scene. For the myocardial infarction situation, assuming that the weight coefficients are 0.7 and 0.3, respectively, the importance weight of the electrocardiogram ST segment elevation node is 0.28×0.7+1 / 1.03×0.3≈0.486, indicating that it has high importance in the judgment of myocardial infarction.

[0031] The distance between different feature nodes is calculated and the similarity between nodes is determined. The shortest path algorithm is used to calculate the distance between nodes, considering the influence of edge weight. The similarity is a function of distance, and the shorter the distance, the higher the similarity. In the feature dependency graph, the path distance between the electrocardiogram ST segment elevation node and the myocardial enzyme index node is 1.2, which is converted to a similarity of 0.833. The similarity between nodes reflects the close degree of clinical association between two features, and features with high similarity often point to the same pathological state.

[0032] The initial association degree is calculated based on the similarity and importance weight between nodes, and the importance weight of the two nodes is multiplied by the similarity between the nodes. Taking the electrocardiogram ST segment elevation node and the myocardial enzyme index node as an example, assuming that the importance weight of the myocardial enzyme index node is 0.420, the initial association degree is 0.486×0.420×0.833≈0.170.

[0033] The initial association degree is adjusted by introducing a time decay factor to obtain the time association degree. For rapidly changing physiological indicators in emergency scenes, a small decay factor such as 0.95 is set; for relatively stable indicators, a larger decay factor such as 0.99 is set. For the detected ST segment elevation and myocardial enzyme index, if the time interval is 10 minutes and the decay factor is 0.95, the time association degree is 0.170×0.95^10≈0.104.

[0034] Extract the corresponding feature mapping relationship of each special disease type from the pre-set special disease type knowledge base. The special disease knowledge base stores the feature patterns of various emergencies, such as the acute myocardial infarction knowledge entry containing ST segment elevation, myocardial enzyme elevation, chest pain, and other features and their typical performance value range, occurrence time sequence, and mutual correlation. The feature mapping relationship is stored in a structured form for quick retrieval and matching. The feature mapping relationship of acute myocardial infarction includes: ST segment elevation → myocardial enzyme elevation (time delay 30-60 minutes), chest pain → ST segment change (occur simultaneously or delay 5-10 minutes), etc.

[0035] Mark the feature nodes with a time sequence correlation greater than a pre-set correlation threshold as key nodes. If the pre-set threshold is 0.10, the time sequence correlation of the feature nodes ST segment elevation and myocardial enzyme is 0.104, which is greater than the threshold, and is marked as a key node.

[0036] Analyze the combination correlation between key nodes and construct a key feature sequence in combination with the feature mapping relationship. The combination correlation is calculated by combining the time sequence correlation between multiple key nodes, reflecting the clinical significance of the co-occurrence of multiple features. In the case of myocardial infarction, the key feature sequence may be: chest pain → ST segment elevation → myocardial enzyme elevation → heart rate instability → blood pressure drop. This sequence is matched with the feature mapping relationship of the myocardial infarction in the special disease knowledge base, and the matching degree is calculated.

[0037] Calculate the matching degree of each feature in the key feature sequence with the special disease type to obtain the special disease type probability value. The matching degree calculation considers multiple factors such as feature value, occurrence time sequence, and duration, and adopts a weighted similarity calculation method. For example, the patient's ST segment elevation is 2mm, lasting for 15 minutes, and the matching degree with the typical feature of myocardial infarction in the knowledge base (ST segment elevation ≥1.5mm, lasting ≥10 minutes) is 0.92; the myocardial enzyme index rises by 35%, and the matching degree with the typical feature (rise ≥30%) is 0.88. Comprehensive calculation of the matching degrees of all key features, the special disease type probability value of the patient's symptoms and acute myocardial infarction is 0.85.

[0038] Determine the special disease type corresponding to the current feature according to the special disease type probability value. When the probability value exceeds a pre-set threshold (such as 0.80), it is confirmed as the corresponding special disease type. After determining the special disease type, extract the time sequence correlation of all feature nodes corresponding to the special disease type from the feature dependency graph to form a symptom feature vector. For acute myocardial infarction, the symptom feature vector contains the time sequence correlation values between the key feature nodes.

[0039] In this embodiment, the importance weight of the feature node is calculated by combining the PageRank centrality and the information entropy in the feature dependency graph, which can improve the accuracy of feature node screening, avoid the deviation caused by a single indicator, and reflect the time evolution relationship between features by considering the node similarity and importance weight when calculating the node association and introducing a time decay factor for dynamic adjustment, thereby improving the timeliness and rationality of the association modeling. By combining the time sequence association degree with the feature mapping relationship in the special disease type knowledge base, key nodes can be effectively screened and key feature sequences can be constructed, thereby improving the matching accuracy between features and special disease types.

[0040] In an alternative embodiment, Based on the symptom feature vector, the pre-set special disease diagnosis and treatment data is partitioned and indexed, a to-be-indexed subset is determined, and the initial similarity between the symptom feature vector in the to-be-indexed subset and the historical cases is calculated, which includes: The difference between each symptom feature value in the symptom feature vector and the pre-set normal reference mean value is calculated to determine the abnormality degree of each symptom, and the corresponding risk level of each symptom is determined based on the abnormality degree; Based on the risk level, the symptom feature vector is divided into multiple priority partitions, and a first warning threshold is solved based on the risk level and a pre-set first balance parameter. For the symptom feature vector in each priority partition, a second warning threshold is solved by calculating the change trend and combining a pre-set second balance parameter. A combined warning value is obtained based on the first warning threshold and the second warning threshold, and an index structure is constructed based on the combined warning value and the symptom feature vector; Based on the index structure, the to-be-matched partitions are located from the special disease diagnosis and treatment data, the partition matching degree between the symptom feature vector and each to-be-matched partition is calculated, and the to-be-indexed subset is combined by selecting three to-be-matched partitions with the highest partition matching degree; The Euclidean distance between the symptom feature vector and each historical case in the to-be-indexed subset is calculated to determine the feature similarity, the timeliness weight is calculated based on the time information of the historical cases, and the initial similarity is calculated by combining the feature similarity.

[0041] The difference between each symptom feature value in the symptom feature vector and the pre-set normal reference mean value is calculated to determine the abnormality degree of each symptom. The normal reference mean value is statistically derived based on a large number of healthy population data, and different reference values are set for different genders and age groups. For an acute myocardial infarction patient, the ST segment elevation in the symptom feature vector is 2.5 mm, while the normal reference mean value is 0.1 mm, the difference is 2.4 mm, and the standardized abnormality degree is 0.96 after conversion; the myocardial enzyme index is 3.8 times the normal value, corresponding to an abnormality degree of 0.92; the blood pressure is 85 / 50 mmHg, which is lower than the normal reference value of 120 / 80 mmHg, and the abnormality degree is 0.74. The abnormality degree is represented by a standardized value of 0 to 1, and the closer the value is to 1, the higher the abnormality degree.

[0042] The risk level corresponding to each symptom is determined based on the abnormality degree. The risk level is divided into four levels: mild abnormality, moderate abnormality, severe abnormality, and extremely severe abnormality, corresponding to abnormality degree intervals of 0-0.25, 0.25-0.50, 0.50-0.75, and 0.75-1. In the foregoing example, the ST segment elevation and myocardial enzyme index belong to extremely severe abnormality, and the blood pressure belongs to severe abnormality. The risk level assessment not only considers the current abnormality degree, but also combines the symptom change rate, and the risk level is increased for rapidly worsening symptoms. For example, the ST segment elevation increases from 1.5 mm to 2.5 mm within 10 minutes, with a high change rate, and is confirmed as extremely severe abnormality; while the mild dizziness symptom lasts for 30 minutes without significant aggravation, and remains at the mild abnormality level.

[0043] The symptom feature vector is divided into multiple priority partitions based on the risk level. The priority partition is used to distinguish the urgency of treatment intervention. Extremely severe abnormality symptoms are assigned to the highest priority partition and require immediate intervention; severe abnormality symptoms are assigned to the high priority partition and require close monitoring and preparation for intervention; moderate and mild abnormality symptoms are assigned to the medium and low priority partitions, respectively. For myocardial infarction patients, the ST segment elevation and myocardial enzyme index are assigned to the highest priority partition, the blood pressure is assigned to the high priority partition, and other minor symptoms are assigned to a lower priority partition.

[0044] The first warning threshold is obtained based on the risk level and a pre-set first balance parameter. The first balance parameter reflects the medical resource situation and the overall risk tolerance of the patient, and the value range is 0 to 1. In the case of sufficient resources, the balance parameter takes a small value such as 0.3, making the system more sensitive; in the case of resource shortage, a larger value such as 0.7 is taken to avoid excessive warning. The first warning threshold is calculated by weighted summation, with the highest weight given to extremely severe abnormality, such as 0.5, severe abnormality weight 0.3, moderate abnormality weight 0.15, and mild abnormality weight 0.05.

[0045] For each symptom feature vector in each priority partition, a change trend is calculated and combined with a pre-set second balance parameter to solve a second early warning threshold. The change trend is calculated by symptom feature values at consecutive time points and is represented by a linear regression slope. For the ST segment elevation indicator, the change trend is a positive growth from 1.5 mm to 2.5 mm within 10 minutes, with a slope of 0.1 mm / min; for the myocardial enzyme indicator, the change trend is a slope of 0.04 times / min from 2.6 times to 3.8 times of the normal value within 30 minutes. The second balance parameter represents the sensitivity to the change trend, and a smaller value such as 0.25 is taken in the case of acute myocardial infarction, and the second early warning threshold of the highest priority partition is 0.78, and the second early warning threshold of the high priority partition is 0.65.

[0046] A combined early warning value is obtained based on the first early warning threshold and the second early warning threshold. For the highest priority partition, the first early warning threshold weight is 0.6 and the second early warning threshold weight is 0.4; for other priority partitions, the weights are 0.7 and 0.3 respectively, and the combined early warning value of the highest priority partition of the patient is 0.84x0.6+0.78x0.4=0.816, and the combined early warning value of the high priority partition is 0.84x0.7+0.65x0.3=0.783.

[0047] An index structure is constructed based on the combined early warning value and the symptom feature vector. The index structure is designed as a multi-level hash table, with the first index being the special disease type, the second index being the priority partition, and the third index being the combined early warning value range. For the current patient, the index path is: acute myocardial infarction→ highest priority partition→ combined early warning value 0.80-0.85. This index structure supports fast positioning of similar cases and improves retrieval efficiency.

[0048] Based on the index structure, a to-be-matched partition is located from the special disease diagnosis and treatment data. The special disease diagnosis and treatment data contains a large number of historical cases and their diagnosis and treatment process and result data. The to-be-matched partition located by the index structure contains multiple historical cases, and the historical cases have similarities with the current patient in terms of special disease type, symptom priority and early warning value. For example, through the aforementioned index path, a to-be-matched partition containing 12 historical cases is located.

[0049] The partition matching degree between the symptom feature vector and each to-be-matched partition is calculated. The partition matching degree is calculated by cosine similarity, considering the distribution of the symptom feature vector in each partition. For the three possible to-be-matched partitions, the calculated partition matching degrees are 0.92, 0.87 and 0.83 respectively, and the to-be-indexed subset is obtained by selecting the to-be-matched partition with the highest partition matching degree, containing a total of 35 historical cases.

[0050] The Euclidean distance of the symptom feature vector and each historical case in the to-be-indexed subset is calculated, and a feature similarity is determined. The Euclidean distance calculation considers the weight of each symptom feature, and a higher weight is given to key symptoms such as ST segment elevation and myocardial enzyme indicators. The feature similarity is a normalized conversion of the Euclidean distance, and the smaller the distance, the higher the similarity. For a historical case in the to-be-indexed subset, the calculated Euclidean distance is 0.23, and the converted feature similarity is 0.88.

[0051] The timeliness weight is calculated based on the time information of the historical cases, and the initial similarity is calculated in combination with the feature similarity. The timeliness weight reflects the time evolution of medical technology and treatment scheme, and the timeliness weight calculation adopts an exponential decay function, taking the current time as the reference, and the closer the case time, the higher the weight. For example, the timeliness weight of a case one year ago is 0.95, the timeliness weight of a case three years ago is 0.86, and the timeliness weight of a case five years ago is 0.78. The initial similarity is obtained by multiplying the feature similarity and the timeliness weight.

[0052] In this embodiment, the abnormality degree is determined by calculating the difference between the symptom feature value and the normal reference mean value, and the risk level is divided accordingly, which can realize the fine division of symptom severity, avoid misjudgment caused by single threshold judgment, dynamically solve the first and second warning thresholds in combination with the risk level and the balance parameter in different priority partitions, and generate a combined warning value, which can improve the flexibility and accuracy of the warning mechanism, enhance the adaptability to complex symptom change trend, construct an index structure based on the combined warning value and locate the to-be-matched partition, which can effectively reduce the search range, improve the matching efficiency, introduce partition matching degree screening in the to-be-indexed subset, calculate the feature similarity by Euclidean distance calculation, and adjust in combination with the timeliness weight of the historical cases, which can ensure the matching accuracy while improving the real-time and reliability of the matching result.

[0053] In an alternative embodiment, determining a candidate case and performing time sequence analysis on the candidate case to determine a symptom evolution rule, calculating a treatment matching degree of current physiological data and the symptom evolution rule, determining a reference case based on the treatment matching degree and extracting a treatment path corresponding to the reference case as a reference path, comprising: determining a candidate case in the to-be-indexed subset based on the initial similarity, performing time sequence analysis on the candidate case by dynamic time warping method, calculating a time sequence distance corresponding to each candidate case and determining a symptom evolution rule based on the time sequence distance; acquiring current physiological data, calculating a disease complexity matching value of the current physiological data and the symptom evolution rule and a medical resource matching degree and summing to obtain a treatment matching degree; Select the candidate case with the highest treatment matching degree as the reference case, construct a treatment optimization objective function based on the complication risk value, solve the treatment optimization objective function by a differential evolution algorithm to obtain an optimal solution, extract the feasibility score corresponding to the optimal solution, and if the feasibility score is greater than a pre-set feasibility threshold, extract the treatment path and output it as the reference path.

[0054] An initial similarity is obtained, and historical cases with an initial similarity greater than a pre-set threshold are selected from the to-be-indexed set as candidate cases. For an acute myocardial infarction patient, the initial similarity threshold is set to 0.80, and 8 candidate cases are selected from 35 historical cases, with initial similarities of 0.836, 0.825, 0.818, 0.814, 0.809, 0.805, 0.802, and 0.801, respectively.

[0055] The candidate cases are subjected to time series analysis by a dynamic time warping method, the time series distance of symptom evolution is calculated, and the symptom development sequence is aligned by time axis stretching. Taking ST segment elevation as an example, the current patient takes 40 minutes to reach 2.5 millimeters from the onset of symptoms, while a certain candidate case takes 55 minutes to reach a similar degree, and the time series distance calculated by dynamic time warping is 0.24. Similar calculations are performed on all key symptoms, and the total time series distance of each candidate case is obtained, such as 0.28, 0.31, 0.25, etc. The smaller the time series distance, the more similar the symptom evolution process.

[0056] The symptom evolution rule is determined based on the time series distance. The symptom change curve of each candidate case is analyzed, and the key time nodes and turning points are extracted. For example, the ST segment elevation evolution rule of a certain candidate case is: initial slight elevation (0.8 millimeters), rapid increase to 2.0 millimeters within 30 minutes, then slow increase to 2.8 millimeters within 60 minutes, and gradual decrease to 1.2 millimeters within 90 minutes after medication. Through clustering analysis of the symptom evolution rules of multiple candidate cases, three typical evolution patterns are obtained: rapid deterioration type, slow progression type, and fluctuating progression type. The symptom evolution characteristics of the current patient are closest to the rapid deterioration type, indicating that the disease develops rapidly and requires urgent intervention.

[0057] The physiological data of the current patient are collected, including real-time indicators such as electrocardiogram, blood pressure, heart rate, respiratory rate, and blood oxygen saturation. The electrocardiogram of the current patient shows that the ST segment is elevated by 2.5 millimeters and continues to increase, the heart rate is 110 beats per minute, the blood pressure is 85 / 50 millimeters of mercury, the respiratory rate is 22 times per minute, and the blood oxygen saturation is 92%.

[0058] The current physiological data and the symptom evolution rule are calculated to match the disease complexity value. The disease complexity considers multiple dimensions: symptom severity, multi-system involvement, complication risk, and patient's basic condition. The current patient's symptom severity score is 8.5 points (out of 10), the multi-system involvement is moderate (heart, circulatory system), the complication risk is high (arrhythmia risk), the patient has a history of hypertension and diabetes, and the basic condition score is 7 points (out of 10). The comprehensive calculation of the disease complexity is 8.2 points. Compared with the disease complexity of each candidate case (such as 7.8 points, 8.4 points, 7.5 points, etc.), the disease complexity matching value is calculated, such as 0.95, 0.98, 0.91, etc. The closer the matching value is to 1, the closer the disease complexity is.

[0059] The medical resource matching degree is calculated, considering factors such as equipment availability, professional configuration, drug supply, and treatment space. The current emergency environment is a pre-hospital emergency vehicle, equipped with basic electrocardiogram monitoring equipment, defibrillator, emergency medicine box, and one senior emergency technician and one ordinary emergency worker. Compared with the medical resources required by the candidate cases, the medical resource matching degree is calculated, such as 0.65, 0.72, 0.58, etc. The higher the matching degree, the more the existing resources can meet the treatment needs.

[0060] The disease complexity matching value and the medical resource matching degree are weighted and summed to obtain the treatment matching degree. The weight setting is dynamically adjusted according to the emergency environment, and the disease complexity matching value weight is 0.6 and the medical resource matching degree weight is 0.4 in the pre-hospital emergency stage. The treatment matching degrees of the 8 candidate cases are calculated as 0.83, 0.87, 0.78, 0.85, 0.81, 0.76, 0.79, and 0.74, respectively. The treatment matching degree reflects the applicability of the treatment plan of the candidate cases to the current patient.

[0061] The candidate case with the highest treatment matching degree is selected as the reference case. In the aforementioned example, the second candidate case has a treatment matching degree of 0.87, which is the highest, so this case is selected as the reference case. The reference case is a 62-year-old male patient, also with acute myocardial infarction, similar ST segment elevation position and degree to the current patient, and obtained good treatment effect under similar medical resource conditions.

[0062] The treatment optimization objective function is constructed according to the complication risk value. The complication risk value is based on the physiological data and the basic disease of the current patient, mainly considering the risk of arrhythmia, shock, and heart failure. The current patient's arrhythmia risk is 0.75, the shock risk is 0.68, and the heart failure risk is 0.56. The treatment optimization objective function considers the maximization of treatment effect and the minimization of complication risk, and the function variables include drug selection, administration timing, administration dose, referral timing, and other treatment parameters.

[0063] The optimal solution is obtained by solving the treatment optimization objective function by the differential evolution algorithm. The algorithm initializes the population size to 50, the maximum number of iterations to 100, the crossover probability to 0.8, and the mutation factor to 0.5. The algorithm converges after about 68 iterations to obtain the optimal solution: immediate dual antiplatelet therapy (aspirin 300mg orally, clopidogrel 300mg orally), intravenous injection of nitroglycerin 0.4mg, and establishment of intravenous access for thrombolytic therapy, and immediate contact with the nearest medical institution with a catheter room for transfer.

[0064] The feasibility score corresponding to the optimal solution is extracted. The feasibility score evaluates the implementation possibility of the optimal solution under the current conditions, considering factors such as drug accessibility, operation difficulty, time requirements, and risk control. The calculated feasibility score is 0.92, which is higher than the pre-set feasibility threshold of 0.85, indicating that the optimal solution has a high implementation possibility under the current conditions.

[0065] The optimal solution corresponding to the treatment path is extracted as the reference path output. The complete reference path includes: T0 time: dual antiplatelet therapy + intravenous injection of nitroglycerin + establishment of intravenous access; T0+10 minutes: evaluate thrombolytic conditions, if met, implement thrombolytic therapy; T0+15 minutes: start the transfer program, contact the receiving hospital; T0+20 minutes: start transfer, continuously monitor vital signs during transit; T0+40 minutes: arrive at the receiving hospital, hand over the patient to the catheter room team; T0+60 minutes: complete coronary intervention.

[0066] In this embodiment, by introducing the dynamic time warping method for time series analysis in candidate cases, and determining the symptom evolution rule based on the time series distance, the dynamic characteristics of patient condition development over time can be more accurately reflected, avoiding the deviation caused by static characteristic comparison. By combining the current physiological data with the symptom evolution rule, the disease complexity matching value and the medical resource matching degree are calculated respectively, the comprehensive evaluation of the reference value of the case can be realized, the clinical applicability of the matching result is improved, and the scientificity and feasibility of the treatment path recommendation are effectively improved through the feasibility score screening and the output of the reference path.

[0067] In an alternative embodiment, generating an initial treatment plan based on the reference path and constructing an initial action space in combination with clinical norms, and constructing a state vector in combination with the current physiological data includes: time series decomposition of the reference path to obtain a plurality of treatment nodes, extraction of treatment measures in each treatment node to construct an initial treatment plan, construction of an initial action space in combination with pre-acquired clinical norms, and optimization of constraint parameters in the initial action space by an adaptive fuzzy reasoning method to obtain optimized constraint parameters; constructing a fuzzy decision tree based on the optimization constraint parameter, performing rationality evaluation on the initial action space based on the fuzzy decision tree, obtaining a rationality score and initializing a feature mapping matrix based on the rationality score; mapping pre-acquired current physiological data into a feature space corresponding to the optimization constraint parameter based on the feature mapping matrix to obtain a state vector.

[0068] The reference path is time-decomposed into multiple treatment nodes. For example, the reference path of an acute myocardial infarction patient is decomposed into six treatment nodes: initial assessment and administration node T0, thrombolytic evaluation node T0+10 minutes, transfer preparation node T0+15 minutes, transfer implementation node T0+20 minutes, hospital handover node T0+40 minutes, and coronary intervention treatment node T0+60 minutes. Each node has a clear time marker and corresponding treatment goals.

[0069] The treatment measures in each treatment node are extracted to construct an initial treatment plan. The treatment measures for the initial assessment and administration node T0 include: establishing a venous access, giving aspirin 300mg orally, clopidogrel 300mg orally, nitroglycerin 0.4mg intravenous injection, and continuous electrocardiogram monitoring. The treatment measures for the thrombolytic evaluation node T0+10 minutes include: evaluating thrombolytic indications and contraindications, and preparing alteplase 50mg. The treatment measures for the transfer preparation node T0+15 minutes include: contacting the receiving hospital, confirming the preparation of the catheter room, and preparing the transfer equipment. During the extraction of treatment measures, specific details of operations, drug dosages, and administration routes are considered to ensure the completeness and executability of the treatment plan.

[0070] An initial action space is constructed in combination with pre-acquired clinical guidelines. Clinical guidelines are derived from authoritative first aid guidelines and expert consensus, providing standardized processing procedures for various emergency situations. For acute myocardial infarction, clinical guidelines specify drug usage order, dosage range, contraindications, and standard procedures for various operations. The initial action space is represented in the form of a multi-dimensional matrix, including dimensions such as action type, execution time, operator, required resources, and expected effect. For example, the action space of anti-platelet therapy includes drug selection such as aspirin and clopidogrel, dosage range such as aspirin 100 to 300mg and clopidogrel 75 to 300mg, administration route such as oral, intravenous, and administration timing such as immediately, within 10 minutes, etc.

[0071] The constraint parameters in the initial action space are optimized by an adaptive fuzzy inference method. The constraint parameters include dose range, time window, operation sequence, and other limiting conditions. The adaptive fuzzy inference method is based on fuzzy set theory, and the constraint parameters are represented as fuzzy membership functions. For aspirin dose, the membership function maps the dose in the range of 100 to 300 mg to a suitability score of 0 to 1. The membership function is dynamically adjusted according to the specific situation of the patient, for example, for patients with high risk of bleeding, the membership function of aspirin dose is shifted to the low dose direction, so that the membership value of 150 mg dose is higher than that of 300 mg. Through the feedback mechanism, the membership function is automatically adjusted to adapt to the changes when new patient data is obtained.

[0072] For the constraint parameter of thrombolytic time window, the initial setting is within 6 hours after the onset of symptoms, and after optimization by adaptive fuzzy inference, the window is reduced to within 4 hours for patients over 70 years old, and the membership value gradually decreases as the time extends. The optimized constraint parameter is no longer a hard limit, but a fuzzy range with membership, which is more in line with the complexity of actual clinical decision-making. For example, the optimized aspirin dose constraint is: less than 100 mg membership 0.2, 100 to 200 mg membership 0.7, 200 to 300 mg membership 0.9, and more than 300 mg membership 0.1.

[0073] A fuzzy decision tree is constructed based on the optimized constraint parameters. The fuzzy decision tree is an extension of the traditional decision tree, and the node splitting and path selection are based on fuzzy logic, including feature selection, node splitting, and path connection. For acute myocardial infarction treatment decision, the root node is the initial evaluation result, the first layer branch node is the ST segment elevation degree and duration, the second layer branch node is the patient's age and underlying disease condition, and the leaf node is the specific treatment decision. For example, a path may be: ST segment elevation greater than 2 mm membership 0.9, duration greater than 30 minutes membership 0.8, age less than 65 years membership 0.7, no severe underlying disease membership 0.9, and recommended direct transport to the catheter room for coronary intervention treatment.

[0074] Each node of the fuzzy decision tree contains a fuzzy set defined based on the optimized constraint parameters, and the connection strength between nodes is determined by the combination of the membership values of each node. For a patient with ST segment elevation 2.5 mm, duration 40 minutes, and history of hypertension and diabetes, the path in the fuzzy decision tree is represented as: ST segment elevation greater than 2 mm membership 0.95, duration greater than 30 minutes membership 0.85, and moderate underlying disease membership 0.7. The comprehensive recommendation is to immediately give anti-platelet therapy and prepare for transport to the catheter room.

[0075] The initial action space is evaluated for rationality based on a fuzzy decision tree, and a rationality score is calculated by traversing the decision tree. In the evaluation process, the matching degree of the action and the patient's state, resource accessibility, time urgency, and risk-reward ratio are considered. For example, for an example patient, the rationality score of anti-platelet therapy with aspirin 300 mg plus clopidogrel 300 mg is 0.88, the rationality score of thrombolytic therapy is 0.65, and the rationality score of direct transport to the catheter room is 0.92. The rationality score ranges from 0 to 1, and the higher the score, the more reasonable the action.

[0076] The feature mapping matrix is initialized based on the rationality score, and the matrix elements represent the influence weight of a specific physiological feature on a specific treatment decision. The matrix initialization is weighted based on the rationality score, and the treatment decision with a higher rationality score obtains a higher initial weight in the matrix. For example, the initial mapping weight of the ST segment elevation feature on the direct transport to the catheter room decision is 0.85, and the initial mapping weight on the thrombolytic therapy decision is 0.75. The dimension of the feature mapping matrix is determined by the number of physiological features and the number of possible treatment decisions. For the current patient, the matrix dimension is 15 by 8, containing 15 key physiological features and 8 possible treatment decisions.

[0077] The pre-acquired current physiological data is mapped into the feature space corresponding to the optimization constraint parameter based on the feature mapping matrix, and a state vector is obtained.

[0078] In this embodiment, by performing time sequence decomposition on the reference path and extracting treatment nodes to construct an initial treatment scheme, fine-grained disassembly of the treatment path can be realized, the integrity and operability of the scheme generation can be improved, the initial action space is constructed in combination with clinical specifications, and the flexibility and adaptability of the scheme are enhanced under the premise of ensuring compliance with medical specifications by optimizing the constraint parameters. Based on the optimization of the constraint parameters, a fuzzy decision tree is constructed and rationality is evaluated, which can effectively eliminate unreasonable action combinations and improve the scientificity and reliability of the scheme design. By initializing the feature mapping matrix based on the rationality score and mapping the current physiological data into the optimized feature space, accurate expression of the individual state of the patient can be realized, and the subsequent decision process can better match the individualized treatment needs.

[0079] Figure 2 The reference path-based treatment decision generation process of the multi-modal perception and AI-based special disease emergency whole-process auxiliary decision method of the embodiment of the application is shown in the flowchart.

[0080] In an alternative embodiment, The initial action space selects a treatment action and calculates an immediate reward value, combines a Monte Carlo search to predict an expected return, and updates the initial treatment scheme based on the predicted expected return to obtain an optimal treatment scheme, which includes: collecting treatment actions in the initial action space and determining the time-dependent relationship between different treatment actions, constructing an action transition graph based on the time-dependent relationship and the treatment actions; determining an action vertex set and a dependent relationship edge set based on the action transition graph, calculating a time constraint function according to the dependent relationship edge set, adding the state vector to the time constraint function to obtain a predicted state transition result; calculating the state improvement degree of the predicted state transition result with respect to the state vector, calculating an immediate reward value in combination with treatment resource consumption, performing Monte Carlo search on the immediate reward value, calculating a confidence upper limit interval and selecting a search node to obtain the predicted expected return; adding the predicted state transition result and the predicted expected return to a preset state-action value function and iteratively optimizing by a time difference algorithm, updating the initial treatment scheme according to the optimization result to obtain an optimal treatment scheme.

[0081] Collecting treatment actions in the initial action space and determining the time-dependent relationship between different treatment actions. Taking acute myocardial infarction first aid as an example, the treatment actions extracted from the initial action space include establishing venous access, giving antiplatelet drugs, intravenous injection of nitroglycerin, evaluating thrombolytic indications, performing thrombolytic therapy, and transporting to the catheter room. The time-dependent relationship describes the sequential execution order constraints between treatment actions, such as establishing venous access must be performed before intravenous injection of nitroglycerin, and antiplatelet therapy can be performed in parallel with intravenous injection of nitroglycerin. By analyzing clinical standards and expert experience, the time-dependent relationship between all treatment actions is determined and recorded as a directed relationship pair.

[0082] Based on the time-dependent relationship and the treatment actions, an action transition graph is constructed, in which the nodes represent the treatment actions and the directed edges represent the dependent relationship between the actions. In the action transition graph, the directed edge from node A to node B indicates that action A must be performed before action B. For the treatment process of acute myocardial infarction, the established action transition graph contains 13 action nodes and 21 dependent relationship edges. For example, there is a directed edge between the node "establishing venous access" and the node "intravenous injection of nitroglycerin", indicating that venous access must be established before drug injection. The action transition graph also contains parallel execution paths, such as "giving aspirin" and "giving clopidogrel" can be performed in parallel, and there is no direct dependent edge between them.

[0083] The action vertex set and the dependency edge set are determined based on the action transition graph. The action vertex set is a collection of all the treatment action nodes in the transition graph, and the dependency edge set is a collection of all the directed edges representing the dependency relationship. For the treatment of acute myocardial infarction, the action vertex set contains 13 vertices such as "establishing venous access", "administering aspirin 300mg", "administering clopidogrel 300mg", "intravenous injection of nitroglycerin 0.4mg", "evaluating thrombolytic indications", "preparing thrombolytic drugs", "performing thrombolytic therapy", "contacting receiving hospitals", "preparing transport equipment", "starting transport", "monitoring vital signs", "handing over patients", and "coronary intervention". The dependency edge set contains 21 directed edges such as "establishing venous access → intravenous injection of nitroglycerin", "evaluating thrombolytic indications → preparing thrombolytic drugs", and "preparing thrombolytic drugs → performing thrombolytic therapy".

[0084] The timing constraint function is calculated according to the dependency edge set. The timing constraint function describes the time relationship constraints between treatment actions, including the earliest start time, the latest end time, and the execution duration. The timing constraint function is obtained by analyzing the dependency edge set. For each action node, the earliest start time is not earlier than the latest end time of all the preceding actions. For example, the earliest start time of "intravenous injection of nitroglycerin" is the latest end time of "establishing venous access" plus the time for preparing the drug. For the treatment of acute myocardial infarction, the execution duration of each action is set according to clinical experience, such as "establishing venous access" lasting for 2 minutes, "administering aspirin" lasting for 1 minute, and "evaluating thrombolytic indications" lasting for 5 minutes. The timing constraint function also takes into account the urgency and priority. For patients with unstable vital signs, the execution time of key actions will be shortened accordingly.

[0085] The state vector is added to the timing constraint function to calculate the predicted state transition result. The state vector describes the current physiological state of the patient, including the ST segment elevation, heart rate, blood pressure, and other indicators. By combining the state vector with the timing constraint function, the change in the patient's state after executing a specific sequence of treatment actions can be predicted. For example, for a patient with a state vector of 0.92, 0.85, 0.78, 0.65, 0.88, 0.72, 0.81, 0.59, after executing "administering aspirin 300mg" and "administering clopidogrel 300mg", the predicted state vector becomes 0.92, 0.87, 0.81, 0.68, 0.90, 0.75, 0.83, 0.62, indicating the expected changes in the patient's physiological indicators.

[0086] The state improvement degree of the predicted state transition result relative to the state vector is calculated. The state improvement degree measures the degree of positive influence of the treatment action on the patient's state, which is calculated by comparing the state vectors before and after the execution of the action. The state improvement degree considers multiple dimensions, including vital sign stability, symptom relief degree, and disease progression control. For acute myocardial infarction patients, the improvement of indicators such as ST segment change, myocardial enzyme index, heart rate, blood pressure, and blood oxygen saturation is focused on. The state improvement degree is calculated in the form of weighted sum, and the weight of each indicator is determined according to the clinical importance. For example, the weight of ST segment regression is 0.35, the weight of heart rate normalization is 0.25, the weight of blood pressure stability is 0.20, and the weight of other indicators is 0.20. For the above example, the state improvement degree after executing anti-platelet therapy is calculated as 0.15.

[0087] The immediate reward value is calculated in combination with the treatment resource consumption. The treatment resource consumption includes drug cost, equipment use, manpower investment, and time cost, etc. The immediate reward value is a comprehensive evaluation of the state improvement degree and resource consumption, reflecting the cost-effectiveness of the treatment action. The calculation uses the state improvement degree minus the weighted value of resource consumption, and the weight is dynamically adjusted according to the resource tightness. For example, the resource consumption score of anti-platelet therapy is 0.05, and the calculated immediate reward value is 0.15-0.05=0.10. The immediate reward value greater than 0 indicates that the benefit of the treatment action is greater than the cost, and the greater the value, the more valuable the action.

[0088] Monte Carlo search is performed on the immediate reward value to calculate the upper confidence interval and select the search node, and the predicted expected return is obtained. Monte Carlo search evaluates the long-term return of different action sequences by simulating multiple possible treatment paths. During the search process, the upper confidence bound algorithm is used to select the most potential node for expansion, balancing exploration and utilization. For each action node, the upper confidence interval is calculated, and the node with the highest upper confidence bound is selected for search. The upper confidence bound takes into account the expected return and uncertainty of the node, encouraging the algorithm to explore actions that have not been fully evaluated. For example, through 1000 times of Monte Carlo simulation, the predicted expected return of the path "anti-platelet therapy first, then thrombolytic therapy, and finally transfer to catheter room" is 0.72, and the confidence interval is 0.68 to 0.76; while the predicted expected return of the path "anti-platelet therapy first, direct transfer to catheter room" is 0.85, and the confidence interval is 0.81 to 0.89. The optimal path is selected.

[0089] The predicted state transition result and the predicted expected return are added to the preset state-action value function and iteratively optimized by a time difference algorithm. The time difference algorithm updates the value function by the actually observed reward and the estimated future return. For each treatment decision, the current state, the executed action, the obtained immediate reward and the transitioned new state are recorded, the time difference error is calculated and the value function is updated. The learning rate is set to 0.05, and the discount factor is 0.95. Through iterative optimization of a large amount of case data, the value function gradually converges, and the long-term value of different treatment actions can be accurately evaluated.

[0090] The initial treatment scheme is updated according to the optimization result to obtain an optimal treatment scheme. The optimal treatment scheme is a sequence of actions with the highest expected return under the current state. By querying the optimized state-action value function, the sequence of actions with the highest value is selected as the optimal scheme. For an acute myocardial infarction patient, the optimal treatment scheme is: immediately give aspirin 300mg and clopidogrel 300mg at T0, and establish a venous access and intravenously inject nitroglycerin 0.4mg; start contacting the receiving hospital with a catheter room at T0+5 minutes; prepare the transfer equipment at T0+10 minutes; start the transfer at T0+15 minutes, continuously monitor the vital signs during the transfer; arrive at the receiving hospital at T0+40 minutes, complete the patient handover; and perform coronary intervention treatment at T0+60 minutes.

[0091] In this embodiment, by introducing the time dependence relationship in the initial action space and constructing the action transition graph, the sequence and mutual restriction between different treatment actions can be accurately described, and unreasonable action combinations in the scheme generation process can be avoided. The time constraint function is calculated based on the action transition graph and combined with the state vector to obtain the predicted state transition result, which can realize the dynamic prediction of the treatment effect. Compared with the static rule derivation method, the method is more flexible and forward-looking. The state improvement degree of the predicted state transition result is calculated and combined with the resource consumption to generate the immediate reward value. The search nodes are selected by using the Monte Carlo search and the confidence upper limit interval, so that the balance between exploration and utilization can be achieved, and the convergence efficiency and accuracy of the scheme optimization can be improved. The predicted state transition result and the predicted expected return are introduced into the state-action value function and iteratively optimized by the time difference algorithm, so that the quality of the scheme can be continuously improved, and the optimal balance can be realized.

[0092] In a second aspect, the embodiment of the present application provides a special disease emergency whole-process auxiliary decision system based on multi-modal perception and AI, comprising: A first unit is configured to acquire multi-modal data of an emergency scene and perform feature extraction to obtain an initial feature vector, calculate the time sequence change of the initial feature vector, construct a feature dependency graph in combination with a conditional probability, calculate the time sequence correlation degree between feature nodes in the feature dependency graph, determine a key feature sequence based on the time sequence correlation degree, and identify a special disease type and a corresponding symptom feature vector. The second unit is configured to partition and index the pre-set disease diagnosis and treatment data based on the symptom feature vector, determine a to-be-indexed subset, calculate the initial similarity between the symptom feature vector in the to-be-indexed subset and a historical case, determine a candidate case, perform time sequence analysis on the candidate case to determine a symptom evolution rule, calculate the treatment matching degree between the current physiological data and the symptom evolution rule, determine a reference case based on the treatment matching degree, and extract a treatment path corresponding to the reference case as a reference path; The third unit is configured to generate an initial treatment scheme based on the reference path, construct an initial action space in combination with clinical norms, construct a state vector in combination with the current physiological data, select a treatment action in the initial action space and calculate an immediate reward value, predict an expected return in combination with Monte Carlo search, and update the initial treatment scheme based on the predicted expected return to obtain an optimal treatment scheme.

[0093] In a third aspect, an electronic device is provided, including: A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0094] In a fourth aspect, a computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0095] The present application can be a method, device, system and / or computer program product. The computer program product can include a computer readable storage medium having computer readable program instructions stored therein, which are used to perform various aspects of the present application.

[0096] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for assisting decision-making throughout the entire process of emergency care for specific diseases based on multimodal perception and AI, characterized in that: include: Multimodal data from the emergency scene is acquired and features are extracted to obtain an initial feature vector. The temporal variation of the initial feature vector is calculated and a feature dependency graph is constructed by combining it with conditional probability. The temporal correlation degree between feature nodes in the feature dependency graph is calculated. Based on the temporal correlation degree, key feature sequences are determined and disease types and corresponding symptom feature vectors are identified. Based on the symptom feature vector, the pre-set disease diagnosis and treatment data is partitioned and indexed to determine the subset to be indexed and calculate the initial similarity between the symptom feature vector in the subset to be indexed and historical cases. Candidate cases are determined and time series analysis is performed on the candidate cases to determine the symptom evolution pattern. The treatment matching degree between the current physiological data and the symptom evolution pattern is calculated. Based on the treatment matching degree, reference cases are determined and the treatment path corresponding to the reference cases is extracted as a reference path. An initial treatment plan is generated based on the reference path, and an initial action space is constructed in conjunction with clinical guidelines. A state vector is constructed in conjunction with the current physiological data. A treatment action is selected in the initial action space and an immediate reward value is calculated. The expected return is predicted by combining Monte Carlo search. The initial treatment plan is updated based on the predicted expected return to obtain the optimal treatment plan.

2. The method according to claim 1, characterized in that, Acquiring multimodal data from the emergency scene and extracting features to obtain an initial feature vector, calculating the temporal variation of the initial feature vector, and constructing a feature dependency graph by combining conditional probabilities include: Feature extraction is performed on multimodal data from the emergency scene to obtain an initial feature vector. The first-order difference value of the initial feature vector is calculated, and the inflection point is determined based on the first-order difference value and a preset adaptive threshold. The initial feature vector is divided into multiple feature segments based on the inflection point. Each feature segment is modeled and a segmented feature function is constructed. A time-varying weighted transition connection function is constructed between adjacent feature segments based on the segmented feature function to obtain the time-series change sequence. The conditional probability is calculated based on the time-series change sequence. When the conditional probability exceeds a preset dynamic threshold, a directed edge is established between the corresponding feature nodes. The directed edge is repeatedly constructed to obtain the feature dependency graph.

3. The method according to claim 1, characterized in that, Calculating the temporal correlation degree between feature nodes in the feature dependency graph, and determining key feature sequences and identifying disease types and corresponding symptom feature vectors based on the temporal correlation degree, includes: For each feature node in the feature dependency graph, the PageRank centrality is calculated by combining the neighboring nodes of the current feature node, the connection probability between the current feature node and its neighboring nodes is calculated to determine the information entropy, and the importance weight of the current feature node is calculated based on the PageRank centrality and the information entropy. Calculate the distance between different feature nodes and determine the similarity between nodes. Calculate the initial correlation degree based on the node similarity and the importance weight. Perform time-series adjustment based on the initial correlation degree and a pre-set time-series decay factor to obtain the time-series correlation degree. In a pre-set disease type knowledge base, feature mapping relationships corresponding to each disease type are extracted. Feature nodes with temporal correlation greater than a preset correlation threshold are taken as key nodes. The combination correlation between the key nodes is analyzed and key feature sequences are constructed by combining the feature mapping relationships. The matching degree between each feature in the key feature sequence and the disease type is calculated to obtain the disease type probability value and determine the disease type corresponding to the current feature. The temporal correlation degree corresponding to the disease type is extracted to obtain the symptom feature vector.

4. The method according to claim 1, characterized in that, Based on the symptom feature vectors, the pre-set disease-specific diagnosis and treatment data is partitioned and indexed. The process of determining the subset to be indexed and calculating the initial similarity between the symptom feature vectors in the subset and historical cases includes: The degree of abnormality of each symptom is determined by calculating the difference between each symptom feature value in the symptom feature vector and the pre-set normal reference mean, and the risk level corresponding to each symptom is determined based on the degree of abnormality. Based on the risk level, the symptom feature vector is divided into multiple priority partitions. A first warning threshold is obtained based on the risk level and a pre-set first balance parameter. For the symptom feature vector in each priority partition, a second warning threshold is obtained by calculating the trend and combining it with a pre-set second balance parameter. A combined warning value is obtained based on the first warning threshold and the second warning threshold. An index structure is constructed based on the combined warning value and the symptom feature vector. Based on the index structure, the partitions to be matched are located from the disease diagnosis and treatment data. The partition matching degree between the symptom feature vector and each partition to be matched is calculated. The three partitions to be matched with the highest partition matching degree are selected to obtain the subset to be indexed. Calculate the Euclidean distance between the symptom feature vector and each historical case in the subset to be indexed, and determine the feature similarity. Calculate the timeliness weight based on the time information of the historical cases, and combine it with the feature similarity to obtain the initial similarity.

5. The method according to claim 1, characterized in that, Candidate cases are identified, and time-series analysis is performed on them to determine the symptom evolution pattern. The treatment matching degree between the current physiological data and the symptom evolution pattern is calculated. Based on the treatment matching degree, reference cases are identified, and the treatment paths corresponding to the reference cases are extracted as reference paths, including: Based on the initial similarity, candidate cases are identified in the subset to be indexed. The candidate cases are then subjected to time series analysis using a dynamic time warping method. The time series distance for each candidate case is calculated, and the symptom evolution pattern is determined based on the time series distance. Collect current physiological data, calculate the disease complexity matching value and medical resource matching degree between the current physiological data and the symptom evolution pattern, and sum them to obtain the treatment matching degree; The candidate case with the highest treatment matching degree is selected as the reference case. A treatment optimization objective function is constructed based on the complication risk value. The optimal solution is obtained by solving the treatment optimization objective function through the differential evolution algorithm. The feasibility score corresponding to the optimal solution is extracted. If the feasibility score is greater than the preset feasibility threshold, the treatment path is extracted and output as the reference path.

6. The method according to claim 1, characterized in that, An initial treatment plan is generated based on the reference path, and an initial action space is constructed in conjunction with clinical guidelines. A state vector is constructed using the current physiological data, including: The reference path is decomposed into multiple treatment nodes by time sequence. Treatment measures in each treatment node are extracted to construct an initial treatment plan. An initial action space is constructed by combining the pre-acquired clinical guidelines. The constraint parameters in the initial action space are optimized by an adaptive fuzzy inference method to obtain optimized constraint parameters. A fuzzy decision tree is constructed based on the optimization constraint parameters. The rationality of the initial action space is evaluated based on the fuzzy decision tree to obtain a rationality score. The feature mapping matrix is ​​then initialized based on the rationality score. Based on the feature mapping matrix, the pre-acquired current physiological data is mapped to the feature space corresponding to the optimization constraint parameters to obtain the state vector.

7. The method according to claim 1, characterized in that, Selecting a treatment action from the initial action space and calculating the immediate reward value, combined with Monte Carlo search to predict the expected return, and updating the initial treatment plan based on the predicted expected return to obtain the optimal treatment plan includes: Collect treatment actions in the initial action space and determine the temporal dependencies between different treatment actions; construct an action transition graph based on the temporal dependencies and the treatment actions. Based on the action transition graph, determine the action vertex set and dependency edge set, calculate the temporal constraint function according to the dependency edge set, and add the state vector to the temporal constraint function to calculate the predicted state transition result; The state improvement degree of the predicted state transition result relative to the state vector is calculated, and the immediate reward value is calculated in combination with the treatment resource consumption. Monte Carlo search is performed on the immediate reward value, the upper confidence interval is calculated and the search node is selected to obtain the predicted expected return. The predicted state transition results and the predicted expected returns are added to a pre-set state-action value function and iteratively optimized using a temporal difference algorithm. The initial treatment plan is then updated based on the optimization results to obtain the optimal treatment plan.

8. A specialized emergency care decision support system based on multimodal perception and AI, used to implement the method of any one of claims 1-7, characterized in that, include: The first unit is used to acquire multimodal data from the emergency scene and extract features to obtain an initial feature vector, calculate the temporal variation of the initial feature vector and construct a feature dependency graph by combining conditional probability, calculate the temporal correlation degree between feature nodes in the feature dependency graph, determine key feature sequences based on the temporal correlation degree and identify disease types and corresponding symptom feature vectors; The second unit is used to partition and index pre-set disease diagnosis and treatment data based on the symptom feature vector, determine the subset to be indexed and calculate the initial similarity between the symptom feature vector in the subset to be indexed and historical cases, determine candidate cases and perform time series analysis on the candidate cases to determine the symptom evolution pattern, calculate the treatment matching degree between the current physiological data and the symptom evolution pattern, determine reference cases based on the treatment matching degree and extract the treatment path corresponding to the reference cases as reference paths; The third unit is used to generate an initial treatment plan based on the reference path and construct an initial action space in combination with clinical standards, construct a state vector in combination with the current physiological data, select a treatment action in the initial action space and calculate the immediate reward value, predict the expected benefit in combination with Monte Carlo search, and update the initial treatment plan based on the predicted expected benefit to obtain the optimal treatment plan.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images