Gastrointestinal disease postoperative rehabilitation method and system based on big data

Abstract

The invention provides a gastrointestinal tract disease postoperative rehabilitation method and system based on big data, and relates to the technical field of gastrointestinal tract postoperative rehabilitation, historical rehabilitation patient data is acquired and clustered, historical rehabilitation patient clustering results and disease cluster similarities corresponding to different types of gastrointestinal tract disease clusters in the clustering results are acquired, and the disease cluster similarities are obtained; generating a gastrointestinal tract disease candidate set based on the disease cluster similarity, fusing a preset gastrointestinal tract disease standard vector with the gastrointestinal tract disease candidate set based on a gastrointestinal tract constraint fusion mechanism to generate a gastrointestinal tract disease rehabilitation prediction curve, and constructing a recombinant gastrointestinal tract disease candidate set; performing threshold calculation based on the recombinant gastrointestinal disease candidate set to obtain a dynamic threshold, obtaining a real-time data vector of a to-be-rehabilitated patient, matching a corresponding gastrointestinal disease rehabilitation prediction curve, comparing the real-time data vector of the to-be-rehabilitated patient with the gastrointestinal disease rehabilitation prediction curve, and implementing a corresponding strategy based on a comparison result. The method can assist a physician in generating a treatment scheme.

Description

Technical Field

[0001] The present invention relates to the technical field of gastrointestinal postoperative rehabilitation. Specifically, it relates to a method and system for gastrointestinal disease postoperative rehabilitation based on big data. Background Art

[0002] Currently, the postoperative rehabilitation monitoring of gastrointestinal diseases mainly relies on doctors' experience and static evaluation of single-dimensional indicators. However, this method has obvious limitations.

[0003] On the one hand, traditional methods rely on doctors to manually integrate fragmented data, such as scattered test reports, imaging reports, medication records, etc. It is necessary to manually consult multi-system reports and subjectively correlate key indicators. This process is time-consuming and vulnerable to experience differences. Inexperienced doctors may overlook cross-modal associations, while over-reliance on experience may lead to cognitive biases, such as the inertial underestimation of "common" symptoms. In addition, the low efficiency of manual integration results in a lag in key indicator responses, delaying the intervention window period. At the same time, in the face of a surge in data volume or complex cases, manual processing is prone to information omission, further reducing the accuracy and timeliness of judgment.

[0004] On the other hand, traditional methods are difficult to generate reliable thresholds because they rely on fixed values or simple statistics and cannot adapt to the dynamic fluctuations of different rehabilitation stages and patient subgroups. For example, CRP is generally high in the early postoperative period, and a fixed threshold is likely to misjudge normal recovery as infection; while the threshold in the maintenance period is too wide, which may mask the recurrence of chronic inflammation. At the same time, a single indicator can only reflect the local pathological state and cannot comprehensively evaluate the condition: a patient may have normal CRP but imaging shows thickening of the intestinal wall, or a sudden drop in hemoglobin indicates hidden bleeding. These key signals are easily overlooked in single-dimensional analysis. Traditional methods lack the ability to co-model multi-modal data, namely symptoms, biochemistry, imaging, and medication, resulting in one-sided evaluation and difficulty in capturing the evolution law of complex conditions.

[0005] Therefore, the existing gastrointestinal postoperative rehabilitation technologies are unable to cope with complex and changing patient data and often have certain limitations in the stability and accuracy of the provided rehabilitation treatment plans. Summary of the Invention

[0006] In view of the above-mentioned problems, in combination with the first aspect of the present invention, embodiments of the present invention provide a method for gastrointestinal disease postoperative rehabilitation based on big data. The method includes: Obtain historical rehabilitation patient data and real-time data vectors of patients to be rehabilitated, cluster the historical rehabilitation patient data, and obtain the clustering results of historical rehabilitation patients. The clustering is expressed as dividing the historical rehabilitation patient data into different types of gastrointestinal disease clusters according to the types of gastrointestinal diseases. Obtain the disease cluster similarity corresponding to different types of gastrointestinal disease clusters, and generate a candidate set of gastrointestinal diseases based on the disease cluster similarity; Obtain a preset standard vector of gastrointestinal diseases and perform constraint fusion on the preset standard vector of gastrointestinal diseases and the candidate set of gastrointestinal diseases based on the gastrointestinal constraint fusion mechanism to generate a gastrointestinal disease rehabilitation prediction curve; Conduct guided recombination on the candidate set of gastrointestinal diseases, obtain multiple groups of recombined gastrointestinal disease vectors, construct a candidate set of recombined gastrointestinal diseases based on the multiple groups of recombined gastrointestinal disease vectors, and calculate the gastrointestinal disease threshold based on the candidate set of recombined gastrointestinal diseases to generate a gastrointestinal disease threshold; Match the real-time data vector of the patient to be rehabilitated with the corresponding gastrointestinal disease rehabilitation prediction curve, and compare the real-time data vector of the patient to be rehabilitated with the corresponding time period in the gastrointestinal disease rehabilitation prediction curve; If the real-time data vector of the patient to be rehabilitated exceeds the gastrointestinal disease threshold in the corresponding time period, perform a judgment on the deterioration of the gastrointestinal disease to obtain a judgment result on the deterioration of the gastrointestinal disease, and the judgment result on the deterioration of the gastrointestinal disease can assist external physicians in generating a corresponding treatment plan for the deterioration of the gastrointestinal disease.

[0007] As a further solution of the present invention, the obtaining of the disease cluster similarity corresponding to different types of gastrointestinal disease clusters and generating a candidate set of gastrointestinal diseases based on the disease cluster similarity includes: Calculate the maximum distance between data within the gastrointestinal disease cluster, generate a cluster density neighborhood based on the maximum distance, calculate the proportion of data within the gastrointestinal disease cluster included in the corresponding cluster density neighborhood of its adjacent gastrointestinal disease cluster, obtain the inter-cluster data coverage, calculate the inter-cluster similarity between any two gastrointestinal disease clusters based on the inter-cluster data coverage, compare the inter-cluster similarity with a preset threshold, obtain an inter-cluster similarity judgment result, and generate a candidate set of gastrointestinal diseases based on the gastrointestinal disease iteration mechanism and the inter-cluster similarity judgment result.

[0008] As a further solution of the present invention, the generating of a candidate set of gastrointestinal diseases based on the gastrointestinal disease iteration mechanism and the inter-cluster similarity judgment result includes: Construct a gastrointestinal disease network with gastrointestinal disease clusters as nodes and inter-cluster similarity as edge weights, merge the gastrointestinal disease clusters through an iterative update rule to obtain a candidate set of gastrointestinal diseases. The iterative update rule is expressed as assigning a disease label to each gastrointestinal disease cluster during initialization, and during the iteration process, the disease label of each gastrointestinal disease cluster will be replaced by the disease label of its strongest associated neighbor. The strongest associated neighbor is the gastrointestinal disease cluster with the largest inter-cluster similarity to the gastrointestinal disease cluster with the replaced disease label. The iteration process continues until the disease label distribution is stable, and finally, the same-label gastrointestinal disease clusters with the same disease label are merged into a candidate set of gastrointestinal diseases.

[0009] As a further solution of the present invention, the obtaining of the preset gastrointestinal disease standard vector and the constraint fusion of the preset gastrointestinal disease standard vector with the gastrointestinal disease candidate set based on the gastrointestinal constraint fusion mechanism to generate a gastrointestinal disease rehabilitation prediction curve includes: Obtain the stage reference data of the historical rehabilitation patient data vectors included in the gastrointestinal disease candidate set at each rehabilitation stage. The stage reference data at each rehabilitation stage is represented as the feature median corresponding to all historical rehabilitation patient data vectors in the gastrointestinal disease candidate set at each rehabilitation stage. Calculate the fusion similarity between the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set through the fusion similarity function, and obtain the fusion similarity result; Generate a fusion factor for the gastrointestinal disease candidate set based on the fusion similarity result. Perform fusion on the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set according to the fusion function and the fusion factor to obtain the fusion result, and generate a temporally continuous gastrointestinal disease rehabilitation prediction curve for the fusion results corresponding to each rehabilitation stage based on multi-dimensional constraints.

[0010] As a further solution of the present invention, the generating of a fusion factor for the gastrointestinal disease candidate set based on the fusion similarity result includes: When the fusion similarity result is not less than 0.8, the generated fusion factor is 0.2. At this time, it indicates that the gastrointestinal disease candidate set fits the preset gastrointestinal disease standard vector, and the gastrointestinal disease candidate set is mainly used for fusion; When the fusion similarity result is in the interval [0.5, 0.8), the generated fusion factor is 0.5, and the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set are evenly fused during fusion; When the fusion similarity result is less than 0.5, the generated fusion factor is 0.8. At this time, it indicates that the gastrointestinal disease candidate set deviates from the preset gastrointestinal disease standard vector, and the preset gastrointestinal disease standard vector is mainly used for fusion.

[0011] As a further solution of the present invention, the guiding recombination of the gastrointestinal disease candidate set, obtaining multiple groups of recombined gastrointestinal disease vectors, constructing a recombined gastrointestinal disease candidate set based on the multiple groups of recombined gastrointestinal disease vectors, and calculating the gastrointestinal disease threshold based on the recombined gastrointestinal disease candidate set to generate a gastrointestinal disease threshold includes: Guidingly recombine the gastrointestinal disease candidate set based on the gastrointestinal disease recombination strategy. The gastrointestinal disease recombination strategy is represented as generating multiple groups of recombined gastrointestinal disease vectors by introducing gradient perturbations and injecting noise, and constructing a recombined gastrointestinal disease candidate set based on the multiple groups of recombined gastrointestinal disease vectors; Perform dynamic threshold calculation on the recombined gastrointestinal disease candidate set based on the Manhattan distance calculation and the quantile threshold selection mechanism to generate a gastrointestinal disease threshold.

[0012] As a further aspect of the present invention, matching the real-time data vector of the patient to be rehabilitated with the corresponding gastrointestinal disease rehabilitation prediction curve, and comparing the real-time data vector of the patient to be rehabilitated with the corresponding time period in the gastrointestinal disease rehabilitation prediction curve, including: Calculating the centroid distance between the real-time data vector of the patient to be rehabilitated and each gastrointestinal disease candidate set, selecting the gastrointestinal disease candidate set with the smallest linear centroid distance as the main matching object, taking the gastrointestinal disease rehabilitation prediction curve corresponding to the main matching object as the individualized rehabilitation benchmark of the patient to be rehabilitated, and comparing the real-time data vector of the patient to be rehabilitated with the individualized rehabilitation benchmark.

[0013] As a further aspect of the present invention, clustering the historical rehabilitation patient data to obtain the historical rehabilitation patient clustering result, including: Clustering the historical rehabilitation patient data vectors based on a clustering algorithm, clustering the historical rehabilitation patient data vectors into multiple different types of gastrointestinal disease clusters according to the gastrointestinal disease type, and generating the historical rehabilitation patient clustering result based on the gastrointestinal disease clusters.

[0014] As a further aspect of the present invention, if the real-time data vector of the patient to be rehabilitated exceeds the gastrointestinal disease threshold for the corresponding time period, then performing a gastrointestinal disease deterioration judgment to obtain the gastrointestinal disease deterioration judgment result, including: Calculating and quantifying the multi-dimensional absolute deviation between the real-time data vector of the patient to be rehabilitated and the gastrointestinal disease rehabilitation prediction curve at each time point based on the Manhattan distance, performing a threshold judgment based on the multi-dimensional absolute deviation. If the multi-dimensional absolute deviation of the patient exceeds the preset gastrointestinal disease threshold three times continuously or the single deviation exceeds 150% of the preset gastrointestinal disease threshold, then triggering the gastrointestinal disease deterioration judgment to obtain the gastrointestinal disease deterioration judgment result.

[0015] On the other hand, an embodiment of the present invention further provides a postoperative rehabilitation system for gastrointestinal diseases based on big data, including: An acquisition module, which is used to acquire historical rehabilitation patient data, the real-time data vector of the patient to be rehabilitated, and a preset gastrointestinal disease standard vector, and cluster the historical rehabilitation patient data to obtain the historical rehabilitation patient clustering result; A generation module, which is used to calculate the disease cluster similarity corresponding to the gastrointestinal disease cluster and generate a gastrointestinal disease candidate set based on the disease cluster similarity; A construction module, which is used to perform constraint fusion on the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set to generate a gastrointestinal disease rehabilitation prediction curve; A recombination module, which is used to guide the recombination of the gastrointestinal disease candidate set, obtain multiple groups of recombined gastrointestinal disease vectors, construct a recombined gastrointestinal disease candidate set based on the multiple groups of recombined gastrointestinal disease vectors, and calculate the gastrointestinal disease threshold to generate the gastrointestinal disease threshold; A comparison module, which is used to match the real-time data vector of the patient to be rehabilitated with the corresponding gastrointestinal disease rehabilitation prediction curve in real time, and perform a comparison operation between the real-time data vector of the patient to be rehabilitated and the corresponding time period in the gastrointestinal disease rehabilitation prediction curve; A judgment module, which is used to judge the deterioration of the gastrointestinal disease for the real-time data vector of the patient to be rehabilitated that exceeds the gastrointestinal disease threshold in the corresponding time period, and obtain the gastrointestinal disease deterioration judgment result.

[0016] Based on the above aspects, the embodiment of the present application obtains the historical rehabilitation patient data and the real-time data vector of the patient to be rehabilitated, clusters the historical rehabilitation patient data to obtain the historical rehabilitation patient clustering result, obtains the disease cluster similarity corresponding to different types of gastrointestinal disease clusters, generates a gastrointestinal disease candidate set based on the disease cluster similarity, obtains the preset gastrointestinal disease standard vector, and performs constraint fusion on the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set based on the gastrointestinal constraint fusion mechanism. By fusing multi-source data such as symptoms, images, and medications through multi-dimensional absolute deviation, the overall deviation degree between the patient's rehabilitation progress and the expected rehabilitation progress is quantified, a gastrointestinal disease rehabilitation prediction curve is generated, the gastrointestinal disease candidate set is guided to be recombined, the reasonable boundaries of each data in the patient's rehabilitation process are expanded, multiple groups of recombined gastrointestinal disease vectors are obtained, a recombined gastrointestinal disease candidate set is constructed based on the multiple groups of recombined gastrointestinal disease vectors, and the gastrointestinal disease threshold is calculated based on the recombined gastrointestinal disease candidate set to generate the gastrointestinal disease threshold. The real-time data vector of the patient to be rehabilitated is matched with the corresponding gastrointestinal disease rehabilitation prediction curve, and a comparison operation is performed between the real-time data vector of the patient to be rehabilitated and the corresponding time period in the gastrointestinal disease rehabilitation prediction curve. If the real-time data vector of the patient to be rehabilitated exceeds the gastrointestinal disease threshold in the corresponding time period, then a gastrointestinal disease deterioration judgment is performed to obtain the gastrointestinal disease deterioration judgment result, covering the complex disease condition evolution and shortening the diagnosis and decision-making time.

[0017] In summary, this method improves the limitations of the lack of accuracy and timeliness caused by manual judgment by doctors and the lack of the ability to cooperate with multi-modal data in traditional methods, resulting in one-sided evaluation and difficulty in capturing the evolution law of complex disease conditions, and provides assistance for doctors to generate a rehabilitation treatment plan that is more suitable for postoperative patients. Description of the Drawings

[0018] Figure 1 It is a schematic execution flowchart of a method for postoperative rehabilitation of gastrointestinal diseases based on big data provided by an embodiment of the present invention.

[0019] Figure 2 It is a schematic diagram of a postoperative rehabilitation system for gastrointestinal diseases based on big data provided by an embodiment of the present invention. Specific implementation manners

[0020] The present invention will be specifically described below in conjunction with the accompanying drawings of the specification. Figure 1 It is a schematic execution flowchart of a postoperative rehabilitation method for gastrointestinal diseases based on big data provided by an embodiment of the present invention. The following will introduce this postoperative rehabilitation method for gastrointestinal diseases based on big data in detail.

[0021] Step S1: Obtain historical rehabilitation patient data and real-time data vectors of patients to be rehabilitated, cluster the historical rehabilitation patient data, and obtain the clustering result of historical rehabilitation patients. The clustering means dividing the historical rehabilitation patient data into different types of gastrointestinal disease clusters according to the types of gastrointestinal diseases.

[0022] In this embodiment, step S1 includes: Step S1-1: Obtain historical rehabilitation patient data and real-time data vectors of patients to be rehabilitated.

[0023] Specifically, extract features from the real-time data of historical rehabilitation patients and patients to be rehabilitated, quantify the extracted features, and construct historical rehabilitation patient data and real-time data vectors of patients to be rehabilitated. The historical rehabilitation patient data and real-time data vectors of patients to be rehabilitated can be expressed as {patient symptoms, biochemical indicators, imaging features, medication records, rehabilitation period}.

[0024] For example, for a patient with ulcerative colitis, the corresponding patient symptoms are abdominal pain 3 times a day, diarrhea 6 times a day, and visible fresh blood in the feces. The biochemical indicators are CRP of 25 mg / L, hemoglobin of 105 mg / L, and fecal calprotectin of 1200 μg / g. The imaging features are a colon wall thickness of 6.5 mm, moderate enhancement of the intestinal wall, and increased density of mesenteric fat. The medication record is taking 3 g of mesalazine per day, taking 20 mg of prednisone per day, and injecting infliximab at a concentration of 5 mg / kg every 8 weeks. The rehabilitation period indicates that the current patient is in the 4th week of postoperative treatment and the expected total rehabilitation period is 12 weeks.

[0025] The symptoms of the patient are quantified into symptom scores based on the symptom quantification relationship. The symptom quantification relationship is expressed as follows: the abdominal pain frequency is quantified as 0 points for no abdominal pain symptoms, 1 point for 1 to 2 times a day, 2 points for 3 to 5 times a day, and 3 points for more than 5 times a day; the diarrhea frequency is quantified as 0 points for no diarrhea symptoms, 1 point for 1 to 3 times a day, 2 points for 4 to 6 times a day, and 3 points for more than 6 times a day; the degree of blood in the stool is quantified as 0 points for no blood in the stool symptoms, 1 point for occult blood positive, 2 points for a small amount of blood in the stool, and 3 points for a large amount of blood in the stool. The full score of the symptom score is 9 points and only two decimal places are retained. Then the symptom score of this patient is (2 + 3 + 3) / 9 = 0.89. The biochemical indicators are quantified by performing data standardization on the biochemical indicators. The results of data standardization for this patient's biochemical indicators are CRP = 1.625, hemoglobin = 0.25, and fecal calprotectin = 3.079. The imaging features are quantified by performing texture analysis on the patient's images. The texture analysis can calculate the uniformity result and the structural difference result of the image by constructing a gray-level co-occurrence matrix, and calculate the Z-score of the colon wall thickness. Combining the uniformity result, the structural difference result of the image, and the Z-score of the colon wall thickness, two comprehensive image scores are output. The comprehensive image scores are used to represent the uniformity and structural difference of the image. The comprehensive image scores of this patient are 0.65 and -0.21. The medication record is quantified by scoring the patient's medication record based on the drug intensity to obtain the medication record score. The score for basic drugs is 1 point, the score for glucocorticoid drugs is 2 to 3 points, and the score for biological agents is 4 to 6 points. The full score of the medication record score is 10 points. Then the corresponding medication record score of this patient is 0.9. The recovery period is quantified by aligning the current treatment stage with the expected total recovery period on the time axis and calculating the recovery period score. Then the corresponding recovery period score of this patient is 0.33. The real-time data vector of this patient can be expressed as [0.89, (1.625, 0.25, 3.079), (0.65, -0.21), 0.9, 0.33].

[0026] Step S1-2: Cluster the historical recovered patient data vectors to obtain the clustering results of historical recovered patients.

[0027] Specifically, cluster the historical recovered patient data vectors based on the clustering algorithm, cluster the historical recovered patient data vectors into multiple different types of gastrointestinal disease clusters according to the types of gastrointestinal diseases, and generate the clustering results of historical recovered patients based on the gastrointestinal disease clusters.

[0028] Step S2: Obtain the disease cluster similarities corresponding to different types of gastrointestinal disease clusters, and generate a candidate set of gastrointestinal diseases based on the disease cluster similarities.

[0029] Specifically, calculate the maximum distance between data within a cluster and the coverage of data between clusters. Generate a cluster density neighborhood based on the maximum distance, calculate the similarity between any two gastrointestinal disease clusters based on the coverage of data between clusters, and compare the similarity between clusters with a preset threshold to obtain a judgment result on the similarity between clusters. The maximum distance between data within a cluster is expressed as the maximum value of the distances between any two historical recovered patient data vectors within the gastrointestinal disease cluster. The calculation of the coverage of data between clusters is expressed as calculating the proportion of the coverage of the historical recovered patient data vectors within the gastrointestinal disease cluster in the cluster density neighborhood of the adjacent gastrointestinal disease cluster.

[0030] Furthermore, construct a gastrointestinal disease network with gastrointestinal disease clusters as nodes and the similarity between clusters as edge weights, and merge the gastrointestinal disease clusters through an iterative update rule to obtain a candidate set of gastrointestinal diseases. The iterative update rule is expressed as assigning independent labels to each gastrointestinal disease cluster at initialization, and during the iteration, the label of each gastrointestinal disease cluster will be replaced with the label of its strongest associated neighbor. The iteration process continues until the label distribution is stable, and finally, the clusters with the same label are merged into the candidate set of gastrointestinal diseases.

[0031] For example, there are currently 3 gastrointestinal disease clusters C1, C2, and C3. Among them, the disease type corresponding to the gastrointestinal disease cluster C1 is moderately to severely ulcerative colitis - hormone-sensitive, and this disease cluster contains 12 patients. The disease type corresponding to the gastrointestinal disease cluster C2 is chronic relapsing Crohn's disease - biologic-responsive, and this disease cluster contains 8 patients. The disease type corresponding to the gastrointestinal disease cluster C3 is acute infectious enteritis - antibiotic-treated, and this disease cluster contains 5 patients.

[0032] Calculate the similarity between any two of the three gastrointestinal disease clusters C1, C2, and C3. First, obtain the maximum distance between the intra-cluster data corresponding to the three gastrointestinal disease clusters C1, C2, and C3. The maximum distances between the intra-cluster data of C1, C2, and C3 are 4.2, 3.8, and 2.4 respectively. Then, the neighborhood radii of the cluster density neighborhoods corresponding to C1, C2, and C3 are 2.1, 1.9, and 1.2 respectively. Among them, 7 patients in C1 fall into the cluster density neighborhood of C2, 5 patients in C2 fall into the cluster density neighborhood of C1, 2 patients in C1 fall into the cluster density neighborhood of C3, 1 patient in C3 falls into the cluster density neighborhood of C1, 1 patient in C2 falls into the cluster density neighborhood of C3, and 0 patients in C3 fall into the cluster density neighborhood of C2. Calculate the coverage of C1 to C2 as 7 / 12 ≈ 0.583, the coverage of C2 to C1 as 5 / 8 = 0.625, the coverage of C1 to C3 as 2 / 12 ≈ 0.167, the coverage of C3 to C1 as 1 / 5 = 0.2, the coverage of C2 to C3 as 1 / 8 = 0.125, and the coverage of C3 to C2 as 0 / 5 = 0. Then, the similarity between C1 and C2 is Sim(C1, C2) = (0.583 + 0.625) / 2 = 0.604. The similarity between C1 and C3 is Sim(C1, C3) = (0.167 + 0.2) / 2 = 0.1835. The similarity between C2 and C3 is Sim(C2, C3) = (0.125 + 0) / 2 = 0.0625. Assume the preset inter-cluster similarity threshold θ = 0.5. Since Sim(C1, C2) > θ, it is determined that C1 and C2 are similar. Since Sim(C1, C3) < θ, it is determined that C1 and C3 are not similar. Since Sim(C2, C3) < θ, it is determined that C2 and C3 are not similar. Based on the determination results of C1, C2, and C3, C1 and C2 are defined as strongly associated clusters, and C3 is defined as an independent cluster. The strongly associated clusters C1 and C2 are merged into the same gastrointestinal disease candidate set, representing the disease group sharing the treatment response mechanism, while the independent cluster C3 is separately used as a gastrointestinal disease candidate set.

[0033] Step S3, obtain the preset gastrointestinal disease standard vector and perform constraint fusion on the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set based on the gastrointestinal constraint fusion mechanism to generate a gastrointestinal disease rehabilitation prediction curve.

[0034] In this embodiment, step S3 includes: Step S3-1, obtain the preset gastrointestinal disease standard vector and preprocess the gastrointestinal disease candidate set.

[0035] Specifically, based on professional materials in the field of gastrointestinal diseases and the research consensus of multiple medical centers, define the ideal rehabilitation indicators for gastrointestinal diseases, extract the features of the ideal rehabilitation indicators for gastrointestinal diseases, and obtain the preset gastrointestinal disease standard vector. For example, in the ideal rehabilitation indicators for gastrointestinal diseases, the ideal biochemical indicator is CRP ≤ 5 mg / L, the hemoglobin of male patients ≥ 130 g / L, the hemoglobin of female patients ≥ 120 g / L, and the ideal imaging feature in the ideal rehabilitation indicators for gastrointestinal diseases is that the colon wall thickness ≤ 3mm and there is no intestinal wall enhancement symptom.

[0036] Further, divide the patient's rehabilitation cycle into stages. For example, the estimated rehabilitation cycle of a certain patient is 12 weeks. Divide 0 - 2 weeks into the initial treatment period, 3 - 6 weeks into the inflammation control period, and 7 - 12 weeks into the stable maintenance period. Map each index in the historical rehabilitation patient data vector included in the gastrointestinal disease candidate set to the standardized time axis according to the time ratio according to the rehabilitation cycle, and linearly interpolate and supplement the missing data. The linear interpolation supplement is expressed as using the adjacent data segments of the missing data for linear interpolation filling.

[0037] Step S3 - 2, based on the gastrointestinal constraint fusion mechanism, perform constraint fusion on the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set.

[0038] Specifically, obtain the stage benchmark data of each rehabilitation stage of the historical rehabilitation patient data vector in the gastrointestinal disease candidate set. The stage benchmark data is expressed as the feature median corresponding to the four feature dimensions of the patient's symptoms, biochemical indicators, imaging features, and medication records of all historical rehabilitation patient data vectors in the gastrointestinal disease candidate set. Calculate the fusion similarity between the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set through the fusion similarity function, and obtain the fusion similarity result. The fusion similarity function can be expressed as: ; Wherein, represents the gastrointestinal disease candidate set, represents the preset gastrointestinal disease standard vector, represents the fusion similarity result, represents the weight of the k - th feature dimension in the gastrointestinal disease candidate set, is the feature median of the data of the k - th feature dimension in the gastrointestinal disease candidate set, represents the feature data of the k - th feature dimension of the preset gastrointestinal disease standard vector, represents the Euclidean length of the median vector of the gastrointestinal disease candidate set, represents the Euclidean length of the preset gastrointestinal disease standard vector.

[0039] For example, a candidate set P for a gastrointestinal disease is an independent cluster, and the corresponding historical data of recovered patients is [0.89, (1.625, 0.25, 3.079), (0.65, -0.21), 0.9, 1.0]. Let the preset standard vector S for gastrointestinal diseases be [0, (0, 0, 0), (1.0, 0.0), 0, 1.0]. In the preset standard vector for gastrointestinal diseases, a symptom score of 0 indicates no symptoms, a biochemical index of (0, 0, 0) indicates CRP = 0, hemoglobin = 0, and fecal calprotectin = 0, and a comprehensive imaging feature score of (1.0, 0.0) indicates uniformity = 1.0 and structural difference = 0.0, that is, the patient's colon wall is completely uniform and there is no structural difference, and a medication record score of 0 indicates no medication required, that is, the patient has fully recovered. Among them, the weights corresponding to the symptom score, CRP, hemoglobin, fecal calprotectin, uniformity, structural difference, and medication record are 0.3, 0.4, 0.2, 0.4, 0.1, 0.1, and 0.1 respectively. Then ≈0.02.

[0040] Furthermore, based on the fusion similarity result, a fusion factor is generated for the candidate set of gastrointestinal diseases. The fusion factor is expressed as follows: when the fusion similarity result is not less than 0.8, the generated fusion factor is 0.2, which indicates that the data included in the candidate set of gastrointestinal diseases highly matches the preset standard vector for gastrointestinal diseases, and the data included in the candidate set of gastrointestinal diseases is mainly used for fusion when performing fusion; when the fusion similarity result is in the interval [0.5, 0.8), the generated fusion factor is 0.5, and when performing fusion, the preset standard vector for gastrointestinal diseases and the candidate set of gastrointestinal diseases are evenly fused; when the fusion similarity result is less than 0.5, the generated fusion factor is 0.8, which indicates that the data included in the candidate set of gastrointestinal diseases deviates from the preset standard vector for gastrointestinal diseases, and the preset standard vector for gastrointestinal diseases is mainly used for fusion when performing fusion. According to the fusion function, the preset standard vector for gastrointestinal diseases and the candidate set of gastrointestinal diseases are fused. The fusion function can be expressed as: ; Among them, is expressed as the recovery stage the fusion result of the k-th dimensional feature dimension of the preset standard vector for gastrointestinal diseases and the candidate set of gastrointestinal diseases, is the fusion factor, is expressed as the recovery stage the feature median of the data of the k-th dimensional feature dimension in the candidate set of gastrointestinal diseases. For example, the preset standard of CRP in the recovery stage T1 is 20 mg / L, the CRP feature median in the candidate set of gastrointestinal diseases is 35 mg / L, and the fusion factor is 0.2, then the fusion result of the CRP dimension is 32 mg / L.

[0041] Step S3-3: Generate a gastrointestinal disease recovery prediction curve.

[0042] Specifically, based on multi-dimensional constraints, generate a temporally continuous gastrointestinal disease recovery prediction curve from the fusion results corresponding to each recovery stage. The multi-dimensional constraints are expressed as follows: the second derivative of the gastrointestinal disease recovery prediction curve is continuous; when there is a unidirectional change in the dimensional index, the curve is constrained to change monotonically; when in the initial stage of treatment, the slope of the curve is negative. For example, when the patient's CRP continuously decreases, the CRP curve is forced to decrease monotonically, and the slope of the curve at the initial stage of the patient's treatment is negative, that is, the patient's condition should not deteriorate at the start of treatment.

[0043] Step S4: Guide the recombination of the gastrointestinal disease candidate set, obtain multiple groups of recombined gastrointestinal disease vectors, construct a recombined gastrointestinal disease candidate set based on the multiple groups of recombined gastrointestinal disease vectors, and calculate the gastrointestinal disease threshold based on the recombined gastrointestinal disease candidate set to generate a gastrointestinal disease threshold.

[0044] In this embodiment, Step S4 includes: Step S4-1: Guide the recombination of the gastrointestinal disease candidate set based on the gastrointestinal disease recombination strategy. The gastrointestinal disease recombination strategy is expressed as generating multiple groups of recombined gastrointestinal disease vectors by introducing gradient perturbations and injecting noise, and constructing a recombined gastrointestinal disease candidate set based on the multiple groups of recombined gastrointestinal disease vectors.

[0045] Specifically, calculate the difference vector between the data vector corresponding to each patient included in the gastrointestinal disease candidate set and the gastrointestinal disease recovery prediction curve at the corresponding time point, and at the same time inject Gaussian noise to generate a noise vector. The noise vector is expressed as calculating the standard deviation for each feature dimension of the gastrointestinal disease candidate set, and at the same time injecting a noise factor into the standard deviation of each dimension to generate a noise vector. Recombine the gastrointestinal disease candidate set based on the difference vector and the noise vector. For example, a certain gastrointestinal disease candidate set is an independent cluster, and the CRP of the patients included in it at the 2nd week at the start of treatment is 25 mg / L, and the CRP corresponding to the gastrointestinal disease recovery prediction curve in the corresponding time period, that is, the 2nd week at the start of treatment, is 20 mg / L. Calculate the difference vector between the two as 0.3×(20 - 25) = -1.5. The 0.3 is the adjustment amplitude coefficient, which represents the simulated progressive improvement trend of the patient under the ideal treatment path. Then the perturbed CRP = 25 + (-1.5) = 23.5 mg / L. Suppose the noise factor and standard deviation of the CRP dimension are 0.1 and 5 respectively, then the generated noise vector is N(0, 0.5). Then the final recombined CRP = 23.5 + N(0, 0.5). The N(0, 0.5) represents any number randomly generated within the range of (0, 0.5).

[0046] It is understandable that, based on the boundary constraints of gastrointestinal diseases, boundary constraints are imposed on the recombinant gastrointestinal disease vectors, and the recombinant gastrointestinal disease vectors exceeding the boundaries of gastrointestinal diseases are forcibly truncated. The gastrointestinal disease boundary constraints are expressed as defining the boundaries of gastrointestinal diseases based on professional materials in the field of gastrointestinal diseases and research consensus of multiple medical centers. For example, it is defined that CRP should not exceed [5, 50 mg / L]. For a certain recombinant gastrointestinal disease vector with CRP = 55 mg / L, since CRP exceeds the defined upper limit, the CRP of this vector is forcibly truncated and corrected to 50 mg / L.

[0047] Furthermore, the data included in the original gastrointestinal disease candidate set is merged one-to-one with the recombinant gastrointestinal disease vectors to generate a recombinant gastrointestinal disease candidate set.

[0048] Step S4-2: Calculate the dynamic threshold for the recombinant gastrointestinal disease candidate set based on the Manhattan distance calculation and the quantile threshold selection mechanism to generate the gastrointestinal disease threshold.

[0049] It is understandable that the Manhattan distance calculation is expressed as quantifying the overall deviation degree between the data included in the recombinant gastrointestinal disease candidate set and the prediction curve by accumulating the absolute deviations between the data of each dimension of the recombinant gastrointestinal disease candidate set and the data of the corresponding gastrointestinal disease recovery prediction curve.

[0050] It is understandable that the quantile threshold selection mechanism is expressed as extracting the Manhattan distance sets of each recovery stage based on the recombinant gastrointestinal disease candidate set, sorting the Manhattan distance sets in ascending order, and obtaining the benchmark threshold using the quantile threshold calculation method. The quantile threshold calculation method is expressed as multiplying the total amount of data in the Manhattan distance set by 0.9 to determine the position index, and obtaining the exact quantile value through linear interpolation of adjacent data, so as to cover 90% of the normal fluctuation range. And the benchmark threshold is dynamically adjusted based on the threshold buffer mechanism. The threshold buffer mechanism is expressed as calculating the standard deviation of the Manhattan distance set and expanding the benchmark threshold by 0.2 times the standard deviation. At the same time, the peak phenomenon is identified by evaluating the distribution kurtosis of the gastrointestinal disease recovery prediction curve. If the data is highly concentrated near the mean value and the kurtosis exceeds the preset critical value, then the 85% quantile is used instead of the 95% quantile to reduce the threshold and avoid interference from a few extreme tail values in the determination.

[0051] For example, the 90% quantile of a certain recombinant gastrointestinal disease vector in the stable maintenance period is 3.5, the standard deviation is 1.2, and the kurtosis of the stable maintenance period does not exceed the preset critical value. Then the dynamic threshold T corresponding to this recovery stage is T = 3.5 + 0.2×1.2 = 3.74.

[0052] Step S5: Match the real-time data vector of the patient to be rehabilitated with the corresponding gastrointestinal disease rehabilitation prediction curve, and compare the real-time data vector of the patient to be rehabilitated with the corresponding time period within the gastrointestinal disease rehabilitation prediction curve.

[0053] Specifically, calculate the Euclidean distance between the real-time data vector of the patient to be rehabilitated and the centroid of each gastrointestinal disease candidate set in the geometric space. The closer the distance between the two, the higher the similarity. Map the Euclidean distance between the two to the interval of 0-1 to generate a normalized similarity value. The system automatically selects the candidate set with the highest similarity as the main matching object, and uses its pre-stored prediction curve as the individualized rehabilitation benchmark. For example, if the Euclidean distance between the real-time data vector of a patient to be rehabilitated and the gastrointestinal disease candidate set corresponding to the disease type of acute infectious enteritis - antibiotic treatment type is the shortest, and the Euclidean distance between the two is 2.0, then the similarity between the two is 1 / (1 + 2) = 0.333.

[0054] Furthermore, for patients with complex conditions, if the similarity difference between the real-time data vector of the patient to be rehabilitated and multiple gastrointestinal disease candidate sets is less than 0.05, then start the multi-curve parallel monitoring mechanism to synchronously track the evolution paths of different gastrointestinal disease subtypes, such as hormone-sensitive type and biologic agent-responsive type, to avoid missing potential risks.

[0055] Step S6: If the real-time data vector of the patient to be rehabilitated exceeds the gastrointestinal disease threshold for the corresponding time period, then perform a gastrointestinal disease deterioration judgment to obtain a gastrointestinal disease deterioration judgment result, and the gastrointestinal disease deterioration judgment result can assist external physicians in generating corresponding gastrointestinal disease deterioration treatment plans.

[0056] Specifically, quantify the multi-dimensional absolute deviation between the real-time data vector of the patient to be rehabilitated and the gastrointestinal disease rehabilitation prediction curve at each time point through the Manhattan distance, and calculate the sum of the absolute values of the differences in the three core dimensions of symptom score, biochemical index, and imaging feature. For example, a patient is in the inflammation control period, with a deviation of 3.2 in the biochemical index CRP and a deviation of 0.8 in the symptom score, then the total deviation of this patient is 4.0.

[0057] If the total deviation of the patient exceeds the preset gastrointestinal disease threshold three times in a row, such as 4.2 > 4.0, 4.5 > 4.0, 4.8 > 4.0, or the single deviation exceeds 150% of the preset gastrointestinal disease threshold, such as the single deviation exceeds 6.0 when the preset gastrointestinal disease threshold is 4.0, then the judgment of gastrointestinal disease deterioration is triggered. The judgment of gastrointestinal disease deterioration means judging and screening the factors that may cause the disease deterioration during the treatment period, and decomposing the abnormal dimension contribution degree based on the judgment result, such as CRP accounting for 65% and imaging accounting for 25%, to assist the doctor to quickly locate the main cause of deterioration, such as recurrence of intestinal inflammation or structural damage, and recommend a targeted plan based on the located main cause of deterioration, such as upgrading anti-inflammatory drugs or increasing the frequency of imaging reexamination.

[0058] Figure 2 The figure shows a schematic diagram of a postoperative rehabilitation system for gastrointestinal diseases based on big data provided by some embodiments of the present application, which can implement the idea of the present application.

[0059] Specifically, a postoperative rehabilitation system for gastrointestinal diseases based on big data includes: An acquisition module, which is used to acquire historical rehabilitation patient data, real-time data vectors of patients to be rehabilitated and preset gastrointestinal disease standard vectors, and cluster the historical rehabilitation patient data to obtain the clustering result of historical rehabilitation patients; A generation module, which is used to calculate the disease cluster similarity corresponding to the gastrointestinal disease cluster and generate a candidate set of gastrointestinal diseases based on the disease cluster similarity; A construction module, which is used to perform constrained fusion of the preset gastrointestinal disease standard vector and the candidate set of gastrointestinal diseases to generate a gastrointestinal disease rehabilitation prediction curve; A recombination module, which is used to perform guided recombination on the candidate set of gastrointestinal diseases, obtain multiple groups of recombined gastrointestinal disease vectors, construct a recombined candidate set of gastrointestinal diseases based on the multiple groups of recombined gastrointestinal disease vectors, and calculate the gastrointestinal disease threshold to generate the gastrointestinal disease threshold; A comparison module, which is used to match the real-time data vector of the patient to be rehabilitated with the corresponding gastrointestinal disease rehabilitation prediction curve, and perform a comparison operation on the real-time data vector of the patient to be rehabilitated and the corresponding time period in the gastrointestinal disease rehabilitation prediction curve; A judgment module, which is used to perform a judgment of gastrointestinal disease deterioration on the real-time data vector of the patient to be rehabilitated that exceeds the gastrointestinal disease threshold in the corresponding time period, and obtain the judgment result of gastrointestinal disease deterioration.

[0060] The following describes the specific usage and functions of this embodiment: First, obtain the historical rehabilitation patient data and the real-time data vector of the patient to be rehabilitated, cluster the historical rehabilitation patient data, and obtain the clustering results of the historical rehabilitation patients. The clustering is expressed as dividing the historical rehabilitation patient data into different types of gastrointestinal disease clusters according to the types of gastrointestinal diseases.

[0061] Then, obtain the disease cluster similarities corresponding to different types of gastrointestinal disease clusters, and generate a candidate set of gastrointestinal diseases based on the disease cluster similarities.

[0062] Next, obtain the preset gastrointestinal disease standard vector and perform constraint fusion on the preset gastrointestinal disease standard vector and the candidate set of gastrointestinal diseases based on the gastrointestinal constraint fusion mechanism to generate a gastrointestinal disease rehabilitation prediction curve.

[0063] Immediately afterwards, perform guided recombination on the candidate set of gastrointestinal diseases to obtain multiple groups of recombined gastrointestinal disease vectors, construct a candidate set of recombined gastrointestinal diseases based on the multiple groups of recombined gastrointestinal disease vectors, and calculate the gastrointestinal disease threshold based on the candidate set of recombined gastrointestinal diseases to generate the gastrointestinal disease threshold.

[0064] Finally, match the real-time data vector of the patient to be rehabilitated with the corresponding gastrointestinal disease rehabilitation prediction curve, and compare the real-time data vector of the patient to be rehabilitated with the corresponding time period in the gastrointestinal disease rehabilitation prediction curve. If the real-time data vector of the patient to be rehabilitated exceeds the gastrointestinal disease threshold in the corresponding time period, perform a deterioration judgment of the gastrointestinal disease to obtain the deterioration judgment result of the gastrointestinal disease. The deterioration judgment result of the gastrointestinal disease can assist external physicians in generating corresponding gastrointestinal disease deterioration treatment plans.

[0065] In this embodiment, by performing constraint fusion on the historical rehabilitation patient data vector and the preset gastrointestinal disease standard vector, a gastrointestinal disease rehabilitation prediction curve is generated, multi-dimensional rehabilitation index data during the patient's rehabilitation period is integrated, and the analysis and evaluation process of the multi-dimensional rehabilitation index data is simplified in the form of a generated prediction curve. At the same time, by performing guided recombination on the candidate set of gastrointestinal diseases, the boundary of the multi-dimensional rehabilitation index data is expanded, making the generated dynamic threshold more accurate and comprehensive, providing assistance for physicians to generate more accurate and patient-compliant postoperative rehabilitation treatment plans.

[0066] In addition, an embodiment of the present invention also provides an electronic device, including: At least one processor; and a memory communicatively connected to at least one processor; wherein, the memory stores instructions executable by at least one processor, and the instructions are executed by at least one processor so that at least one processor can execute the method proposed in Embodiment 1 of the present invention.

[0067] Next, specific introductions will be made to the various components of the electronic device: Among them, the processor is the control center of the electronic device, which can be a single processor or a collective term for multiple processing elements. For example, the processor can be one or more central processing units (CPUs), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement Embodiment 1 of the present invention. For example, one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs).

[0068] Among them, the processor can execute various functions of the electronic device by running or executing software programs stored in the memory and invoking data stored in the memory.

[0069] The memory is used to store the software program for implementing the solution of the present invention and is controlled by the processor for execution. The specific implementation manner can refer to the above method embodiments and will not be elaborated here.

[0070] The memory can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only (CD-ROM), or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory can be integrated with the processor or exist independently and be coupled to the processor through the interface circuit of the electronic device. The embodiments of the present invention do not make specific limitations on this.

[0071] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that contains one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0072] It should be understood that the term "and / or" in this document is merely a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone, where A and B can be singular or plural. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood by referring to the context before and after.

[0073] It should be understood that in the embodiments of the present invention, the magnitude of the sequence numbers of the above processes does not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0074] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A postoperative rehabilitation method for gastrointestinal diseases based on big data, characterized in that, The method includes: Obtain historical rehabilitation patient data and real-time data vectors of patients to be rehabilitated, cluster the historical rehabilitation patient data, and obtain the clustering results of historical rehabilitation patients. The clustering is expressed as dividing the historical rehabilitation patient data into different types of gastrointestinal disease clusters according to the types of gastrointestinal diseases; Obtain the disease cluster similarities corresponding to different types of gastrointestinal disease clusters, and generate a gastrointestinal disease candidate set based on the disease cluster similarities; Obtain a preset gastrointestinal disease standard vector and perform constraint fusion on the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set based on the gastrointestinal constraint fusion mechanism to generate a gastrointestinal disease rehabilitation prediction curve; Perform guided recombination on the gastrointestinal disease candidate set, obtain multiple groups of recombined gastrointestinal disease vectors, construct a recombined gastrointestinal disease candidate set based on the multiple groups of recombined gastrointestinal disease vectors, and calculate the gastrointestinal disease threshold based on the recombined gastrointestinal disease candidate set to generate a gastrointestinal disease threshold; Match the real-time data vector of the patient to be rehabilitated with the corresponding gastrointestinal disease rehabilitation prediction curve, and compare the real-time data vector of the patient to be rehabilitated with the corresponding time period in the gastrointestinal disease rehabilitation prediction curve; If the real-time data vector of the patient to be rehabilitated exceeds the gastrointestinal disease threshold in the corresponding time period, perform a gastrointestinal disease deterioration judgment to obtain a gastrointestinal disease deterioration judgment result. The gastrointestinal disease deterioration judgment result can assist external physicians in generating corresponding gastrointestinal disease deterioration treatment plans.

2. The postoperative rehabilitation method for gastrointestinal diseases based on big data according to claim 1, wherein The obtaining of the disease cluster similarities corresponding to different types of gastrointestinal disease clusters and generating a gastrointestinal disease candidate set based on the disease cluster similarities includes: Calculate the maximum distance between data within the gastrointestinal disease cluster, generate a cluster density neighborhood based on the maximum distance, calculate the proportion of the data within the gastrointestinal disease cluster included in the corresponding cluster density neighborhood of its adjacent gastrointestinal disease cluster, obtain the inter-cluster data coverage, calculate the inter-cluster similarity between any two gastrointestinal disease clusters based on the inter-cluster data coverage, compare the inter-cluster similarity with a preset threshold to obtain an inter-cluster similarity judgment result, and generate a gastrointestinal disease candidate set based on the gastrointestinal disease iteration mechanism and the inter-cluster similarity judgment result.

3. The postoperative rehabilitation method for gastrointestinal diseases based on big data according to claim 2, characterized in that The generating of a gastrointestinal disease candidate set based on the gastrointestinal disease iteration mechanism and the inter-cluster similarity judgment result includes: Construct a gastrointestinal disease network with the gastrointestinal disease clusters as nodes and the inter-cluster similarities as edge weights, merge the gastrointestinal disease clusters through an iterative update rule to obtain a gastrointestinal disease candidate set. The iterative update rule is expressed as assigning a disease label to each gastrointestinal disease cluster during initialization, and during the iteration process, the disease label of each gastrointestinal disease cluster will be replaced by the disease label of its strongest associated neighbor. The strongest associated neighbor is expressed as the gastrointestinal disease cluster with the largest inter-cluster similarity to the gastrointestinal disease cluster with the replaced disease label. The iteration process continues until the disease label distribution is stable, and finally, the gastrointestinal disease clusters with the same disease label are merged into a gastrointestinal disease candidate set.

4. A method for postoperative rehabilitation of gastrointestinal diseases based on big data according to claim 1, characterized in that, The obtaining of a preset gastrointestinal disease standard vector and performing constraint fusion on the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set based on the gastrointestinal constraint fusion mechanism to generate a gastrointestinal disease rehabilitation prediction curve includes: Obtain the stage benchmark data of the historical recovered patient data vectors included in the gastrointestinal disease candidate set at each recovery stage. The stage benchmark data of each recovery stage is represented as the median of the features corresponding to all historical recovered patient data vectors in the gastrointestinal disease candidate set at each recovery stage. Calculate the fusion similarity between the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set through the fusion similarity function, and obtain the fusion similarity result; Generate a fusion factor for the gastrointestinal disease candidate set based on the fusion similarity result. According to the fusion function and the fusion factor, fuse the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set to obtain a fusion result, and generate a time-continuous gastrointestinal disease recovery prediction curve for the fusion results corresponding to each recovery stage based on multi-dimensional constraints.

5. A postoperative rehabilitation method for gastrointestinal diseases based on big data according to claim 4, characterized in that, The generating a fusion factor for the gastrointestinal disease candidate set based on the fusion similarity result includes: When the fusion similarity result is not less than 0.8, the generated fusion factor is 0.

2. At this time, it indicates that the gastrointestinal disease candidate set conforms to the preset gastrointestinal disease standard vector, and the gastrointestinal disease candidate set is mainly used for fusion; When the fusion similarity result is in the interval [0.5, 0.8), the generated fusion factor is 0.5, and the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set are evenly fused during fusion; When the fusion similarity result is less than 0.5, the generated fusion factor is 0.

8. At this time, it indicates that the gastrointestinal disease candidate set deviates from the preset gastrointestinal disease standard vector, and the preset gastrointestinal disease standard vector is mainly used for fusion.

6. The postoperative rehabilitation method for gastrointestinal diseases based on big data according to claim 1, characterized in that The guiding and reorganizing the gastrointestinal disease candidate set to obtain multiple groups of reorganized gastrointestinal disease vectors, constructing a reorganized gastrointestinal disease candidate set based on the multiple groups of reorganized gastrointestinal disease vectors, and calculating the gastrointestinal disease threshold based on the reorganized gastrointestinal disease candidate set to generate the gastrointestinal disease threshold includes: Guiding and reorganizing the gastrointestinal disease candidate set based on the gastrointestinal disease reorganization strategy. The gastrointestinal disease reorganization strategy is expressed as generating multiple groups of reorganized gastrointestinal disease vectors by introducing gradient perturbation and injecting noise, and constructing a reorganized gastrointestinal disease candidate set based on the multiple groups of reorganized gastrointestinal disease vectors; Performing dynamic threshold calculation on the reorganized gastrointestinal disease candidate set based on the Manhattan distance calculation and the quantile threshold selection mechanism to generate the gastrointestinal disease threshold.

7. A postoperative rehabilitation method for gastrointestinal diseases based on big data according to claim 1, characterized in that The matching the real-time data vector of the patient to be recovered with the corresponding gastrointestinal disease recovery prediction curve, and comparing the real-time data vector of the patient to be recovered with the corresponding time period in the gastrointestinal disease recovery prediction curve includes: Calculate the centroid distance between the real-time data vector of the patient to be recovered and each gastrointestinal disease candidate set, select the gastrointestinal disease candidate set with the smallest linear centroid distance as the main matching object, use the gastrointestinal disease recovery prediction curve corresponding to the main matching object as the individualized recovery benchmark of the patient to be recovered, and compare the real-time data vector of the patient to be recovered with the individualized recovery benchmark.

8. A method for postoperative rehabilitation of gastrointestinal diseases based on big data according to claim 1, characterized in that, The clustering the historical recovered patient data to obtain the historical recovered patient clustering result includes: Cluster the historical rehabilitation patient data vectors based on a clustering algorithm, cluster the historical rehabilitation patient data vectors into multiple different types of gastrointestinal disease clusters according to the types of gastrointestinal diseases, and generate a historical rehabilitation patient clustering result based on the gastrointestinal disease clusters.

9. A postoperative rehabilitation method for gastrointestinal diseases based on big data according to claim 1, characterized in that, If the real-time data vector of the patient to be rehabilitated exceeds the gastrointestinal disease threshold for the corresponding period, then perform a gastrointestinal disease deterioration judgment to obtain a gastrointestinal disease deterioration judgment result, including: Calculate the multi-dimensional absolute deviation of the real-time data vector of the patient to be rehabilitated and the gastrointestinal disease rehabilitation prediction curve at each time point based on the Manhattan distance, and perform a threshold judgment based on the multi-dimensional absolute deviation. If the multi-dimensional absolute deviation of the patient exceeds the preset gastrointestinal disease threshold three times continuously or the single deviation exceeds 150% of the preset gastrointestinal disease threshold, then trigger a gastrointestinal disease deterioration judgment to obtain a gastrointestinal disease deterioration judgment result.

10. A postoperative rehabilitation system for gastrointestinal diseases based on big data, characterized in that, Including: An acquisition module, which is used to acquire historical rehabilitation patient data, the real-time data vector of the patient to be rehabilitated, and a preset gastrointestinal disease standard vector, and cluster the historical rehabilitation patient data to obtain a historical rehabilitation patient clustering result; A generation module, which is used to calculate the disease cluster similarity corresponding to the gastrointestinal disease cluster and generate a gastrointestinal disease candidate set based on the disease cluster similarity; A construction module, which is used to perform constrained fusion of the preset gastrointestinal disease standard vector and the gastrointestinal disease candidate set to generate a gastrointestinal disease rehabilitation prediction curve; A recombination module, which is used to perform guided recombination on the gastrointestinal disease candidate set to obtain multiple groups of recombined gastrointestinal disease vectors, construct a recombined gastrointestinal disease candidate set based on the multiple groups of recombined gastrointestinal disease vectors, and calculate the gastrointestinal disease threshold to generate a gastrointestinal disease threshold; A comparison module, which is used to match the real-time data vector of the patient to be rehabilitated with the corresponding gastrointestinal disease rehabilitation prediction curve, and perform a comparison operation on the real-time data vector of the patient to be rehabilitated and the corresponding period in the gastrointestinal disease rehabilitation prediction curve; A judgment module, which is used to perform a gastrointestinal disease deterioration judgment on the real-time data vector of the patient to be rehabilitated that exceeds the gastrointestinal disease threshold for the corresponding period to obtain a gastrointestinal disease deterioration judgment result.

Images