Surgical operation information data management system and method based on mobile internet

Abstract

The invention relates to the technical field of data processing, in particular to a surgical operation information data management system and method based on the mobile internet, and the method comprises the following steps: obtaining first surgical operation information data through a medical high-definition camera, a sound recorder and medical equipment; performing data self-adaption on the first surgical operation information data by utilizing an artificial intelligence algorithm to generate a second surgical operation information data set; performing visual projection on the second surgical operation information data set by using a matrix decomposition method to generate a surgical operation characteristic matrix projection drawing; performing data visualization processing on the surgical operation information matrix decomposition graph by using a deep learning algorithm to generate a surgical operation feature interactive view; performing homomorphic encryption on the surgical operation convolutional feature model by using a homomorphic encryption algorithm; uploading data of the surgical operation homomorphic encryption model to a surgical operation information data management system by using a 5G technology; according to the invention, accurate and orderly management of surgical operation information data is realized.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a surgical operation information data management system and method based on mobile Internet. Background Art

[0002] The traditional method of surgical information data management is to record and store it through a series of handwritten documents, medical records, photos, etc. This method can easily cause confusion in surgical information data and inaccurate information recording, which seriously affects the care and treatment of patients. Moreover, the paper recording method is not conducive to improving the work efficiency of medical staff and information sharing. Therefore, through a surgical information data management system based on the mobile Internet, all information data can be recorded in real time at the operating site and integrated into the data management system, which can greatly simplify the information management and data analysis process of surgical operations. On this basis, combined with artificial intelligence, cloud computing and other technologies, it can provide medical staff with more complete surgical information management and processing services. Medical staff can upload surgical visualization image data for real-time diagnostic monitoring and operation guidance, thereby improving the success rate of surgery; they can also optimize and improve the surgical environment through real-time monitoring and analysis of data, and minimize surgical risks and patient infection rates. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention proposes a surgical operation information data management system and method based on mobile Internet to solve at least one of the above technical problems.

[0004] To achieve the above-mentioned object, the present invention provides a method for managing surgical operation information data based on mobile Internet, comprising the following steps:

[0005] Step S1: Acquire first surgical operation information data using a medical high-definition camera, a recorder, and medical equipment; the medical equipment includes an electrocardiogram (ECG) device and an X-ray detector; and the surgical operation information data includes surgical operation video data, surgical operation audio data, patient vital sign data, anesthesia detection data, and surgical instrument equipment data.

[0006] Step S2: using an artificial intelligence algorithm to perform data adaptive processing on the first surgical operation information data to generate a second surgical operation information data set;

[0007] Step S3: Perform visual projection on the second surgical operation information dataset using a matrix decomposition method to generate a surgical operation feature matrix projection diagram;

[0008] Step S4: Using a deep learning algorithm to perform data visualization on the surgical information matrix decomposition graph to generate an interactive view of surgical features;

[0009] Step S5: using a superpixel convolutional network to perform dilated convolution and multi-scale sampling on the interactive view of the generated surgical information to generate a surgical convolution feature model;

[0010] Step S6: Use the homomorphic encryption algorithm to homomorphically encrypt the surgical convolution feature model to generate a surgical homomorphic encryption model; use 5G technology based on the mobile Internet to upload the surgical homomorphic encryption model data to the surgical information data management system to perform management operations.

[0011] The present invention provides a surgical operation information data management method based on mobile Internet. By using medical high-definition cameras, recorders and medical equipment to obtain first surgical operation information data, most of the information of the entire surgical process can be recorded in detail, including surgical video data, surgical audio data, etc. These data will help improve the accuracy and efficiency of surgical operations, provide valuable data for related research, and provide resources for subsequent surgical reviews. The first surgical operation information data is adaptively processed using an artificial intelligence algorithm, which can better extract and analyze surgical data and generate a second surgical operation information data set with more accurate data analysis and mining capabilities, which will help improve the quality and efficiency of surgical operations. The second surgical operation information data set is visually projected using a matrix decomposition method to generate a surgical feature matrix projection map, which will help to better visualize and understand surgical information, thereby providing doctors with more accurate judgments and decisions. Using deep The degree learning algorithm performs data visualization processing on the surgical information matrix decomposition diagram to generate an interactive view of surgical features. This view can provide sufficient surgical information to help doctors understand the surgical conditions more comprehensively and deeply, thereby improving the safety and accuracy of the operation. By using the superpixel convolutional network to perform dilated convolution and multi-scale sampling on the generated interactive view of surgical information, a surgical convolution feature model is generated, which will help improve the analysis and judgment capabilities of surgical data and further improve the accuracy and efficiency of the operation. The surgical convolution feature model is encrypted using the homomorphic encryption algorithm to generate a surgical homomorphic encryption model, which will help protect the security and privacy of surgical data, thereby protecting patient privacy and doctor's business secrets. At the same time, it will provide better security for the management and analysis of surgical information data. The data is uploaded to the surgical information data management system to ensure data stability and ensure that the surgical information data is non-redundant and stored in an orderly manner.

[0012] Preferably, step S1 includes the following steps:

[0013] Step S11: Using a medical high-definition camera to shoot the surgical process and obtain surgical video data;

[0014] Step S12: using a recorder to record the entire surgical process to obtain surgical audio data;

[0015] Step S13: Using an electrocardiograph to detect the patient's vital signs, obtaining the patient's vital sign data and anesthesia detection data. The patient's vital sign data includes body temperature data, heart rate data, blood oxygen saturation data, blood pressure data, brain wave data, and electrocardiogram data;

[0016] Step S14: Use an X-ray detector to perform laser scanning on the surgical instrument equipment data to obtain the surgical instrument equipment data.

[0017] The present invention uses a medical high-definition camera to shoot the surgical process, which can collect high-quality surgical video data and use the video to record the surgical process in detail. These data will help analyze and study the surgical process, thereby further improving the quality and efficiency of surgical operations. The entire surgical process is recorded using a recorder, which can collect detailed surgical audio data. These data will help doctors better understand the surgical process, thereby better conducting subsequent analysis and research. The patient's vital signs are detected using an electrocardiogram device, which can collect the patient's vital signs data and anesthesia detection data. These data will help doctors comprehensively monitor and record the patient's vital signs and anesthesia status, thereby ensuring the safety and accuracy of the surgical process. The surgical instrument equipment data is laser scanned using an X-ray detector, which can collect detailed surgical instrument equipment data. These data will help doctors better understand the use and operation of surgical instruments and equipment, thereby further improving the quality and efficiency of surgical operations. These data will help doctors have a more comprehensive understanding of the surgical process, thereby improving the quality and efficiency of surgical operations, and also provide important data support for related medical research and practice.

[0018] Preferably, step S2 includes the following steps:

[0019] Step S21: using a machine vision algorithm based on an artificial intelligence algorithm to perform dynamic target tracking on the surgical video data to generate a dynamic visual image of the surgical operation;

[0020] Step S22: using a time series algorithm to perform vectorized sequence conversion on the patient's vital sign data to generate surgical operation vector sequence data;

[0021] Step S23: performing image segmentation on the surgical dynamic visual image using an image segmentation algorithm to generate an organ-tissue segmentation dataset;

[0022] Step S24: Using a feature fusion algorithm to perform heterogeneous data fusion on the surgical vector sequence data and the organ tissue segmentation data set to generate a second surgical data set.

[0023] The present invention uses a machine vision algorithm based on an artificial intelligence algorithm to perform dynamic target tracking on surgical video data, which helps doctors to understand the patient's internal conditions more comprehensively and accurately during the operation, including the location, size and shape of diseased tissues and organs, the location of surgical instruments and related changes, etc. This information can improve the safety and accuracy of the operation, and also help to reduce the operation time and the patient's recovery time. The patient's vital signs data are vectorized and converted into a sequence using a time series algorithm to generate surgical vector sequence data. This data includes the patient's vital signs indicators, such as heart rate, blood pressure, respiratory rate, anesthetic dosage, etc. By systematically recording this data, doctors can better monitor the patient's physiological state and predict possible physiological reactions and complications. The dynamic visual images of the surgical operation are cut using an image cutting algorithm to generate an organ-tissue segmentation dataset. This process uses relevant Algorithms and models segment dynamic visual images into different regions and tissues. These data sets help doctors better understand the location and distribution of disease sites and related pathological tissues, provide better navigation and control for surgery, and help accurately judge the range of resection and the patient's postoperative recovery. Feature fusion algorithms are used to perform heterogeneous data fusion on surgical vector sequence data and organ tissue segmentation data sets to generate a second surgical data set. Feature fusion technology can fuse data from different sources to improve the integrity, richness and accuracy of the data. In the fusion of surgical data sets, patient vital signs data and organ tissue segmentation data sets can be merged with dynamic visual image data to produce more comprehensive and accurate surgical monitoring information. These data sets help doctors to have a deeper understanding of the characteristics and laws of surgery, provide better control and monitoring for surgery, and provide higher quality data support for related medical research and practice.

[0024] Preferably, step S23 includes the following steps:

[0025] Step S231: using a region growing algorithm to divide the surgical dynamic visual image into pixel areas to generate a surgical dynamic visual similarity area;

[0026] Step S232: using an image cutting algorithm to perform image cutting on the surgical dynamic visual similarity area to generate a surgical area cutting area;

[0027] Step S233: using a feature point detection algorithm to mark organ-tissue feature points on the surgical dynamic visual image, and to obtain organ-tissue feature point coordinate data;

[0028] Step S234: Using the organ-tissue feature point coordinate data, the surgical field cutting area is annotated to generate an organ-tissue segmentation data set.

[0029] The present invention divides the dynamic visual images of surgical operations into pixel areas through the region growing algorithm, thereby realizing cluster analysis of relatively similar areas, distinguishing different organs or tissue areas during surgery, allowing doctors to more accurately select and locate the application range and area of surgical instruments, improving the accuracy and safety of surgery, and also providing more targeted data support for subsequent steps, such as the input data of steps such as image cutting. The image cutting algorithm can be used to divide the dynamic visual images of surgical operations into regions, realizing accurate positioning and differentiation of different organs and tissues. According to the segmented organ-tissue regions, the location and nature of the disease during surgery can be better understood, and the surgical path and actions can be better planned as much as possible. In order to effectively reduce damage to non-lesioned tissues and thus improve the effect and safety of surgery, the feature point detection algorithm can mark the feature points of important organs or tissues in dynamic visual images of surgical operations, so that the "key points" of important tissues in the operation can be extracted from the image, providing better data support in the next step of labeling and segmentation. The feature point coordinate data of the organ tissue is annotated with the image cutting area to generate an organ-tissue segmentation dataset. By combining image segmentation and feature point marking, more accurate organ and tissue segmentation results can be obtained, which provides more accurate and powerful data support for subsequent surgical decisions and operations, and also leaves more accurate surgical records and monitoring for patients with impaired consciousness.

[0030] Preferably, step S3 includes the following steps:

[0031] Step S31: performing data preprocessing on the second surgical data set to generate a surgical data preprocessing pipeline, wherein the data preprocessing includes cleaning, integration, and standardization;

[0032] Step S32: performing matrix partitioning on the surgical data preprocessing pipeline to obtain a plurality of surgical data sub-matrices;

[0033] Step S33: performing matrix decomposition on the surgical data submatrix using a non-negative matrix factorization method to generate a principal component matrix and a minimum error matrix;

[0034] Step S34: performing feature extraction based on the principal component matrix and the minimum error matrix to generate a surgical feature weight matrix;

[0035] Step S35: Perform visual projection on the surgical feature weight matrix using a visual projection method to generate a surgical feature matrix projection diagram.

[0036] The present invention can clean, integrate and standardize the surgical data according to the second surgical data set through a data preprocessing pipeline, thereby eliminating outliers and noise in the data, improving data quality, making the data more accurately reflect the characteristics of the surgery, reducing the influence of noise and erroneous data, and better ensuring the accuracy and reliability of subsequent processing. The preprocessed surgical data can be divided into multiple sub-matrices through matrix partitioning, making the data set smaller and easier to process. Dividing a large data set into multiple small sub-data sets makes subsequent calculations more efficient and reduces the overhead of computing resources. The non-negative matrix decomposition method can decompose the surgical data sub-matrix to generate the principal component matrix and the most Small error matrix, thereby extracting the main feature information in the data. By extracting the principal component matrix, the surgical data can be compressed into fewer dimensions, redundant information can be removed, and the main features of the data can be highlighted, making subsequent processing more efficient. Feature extraction can be performed based on the principal component matrix and the minimum error matrix to generate a surgical feature weight matrix, thereby extracting important feature information of the surgical data. Through feature extraction, the key features in the surgical data can be obtained, effectively improving the accuracy and efficiency of tasks such as classification and recognition. Through visual projection, the surgical data can be displayed on a three-dimensional view, allowing doctors to more clearly visualize the feature information of the surgical data and improve the accuracy and efficiency of surgical decision-making.

[0037] Preferably, step S4 includes the following steps:

[0038] Step S41: extracting features from the surgical feature matrix projection image using a principal component algorithm to generate surgical feature vector data;

[0039] Step S42: Performing data visualization processing on the surgical feature vector data using a deep learning algorithm to generate a surgical feature visualization view;

[0040] Step S43: Use the JavaScript library to perform interactive processing on the surgical feature visualization view to generate an interactive view of the surgical feature.

[0041] The present invention uses the principal component algorithm to extract features from the surgical feature matrix projection map to generate surgical feature vector data. The feature vector contains the most important feature information in the surgical data and can be used for further analysis and modeling to improve the accuracy and efficiency of surgical decision-making. The deep learning algorithm is used to visualize the surgical feature vector data, and the features of the surgical data are displayed through a visual view, allowing doctors to understand and analyze the surgical data in a more intuitive way, helping doctors make more accurate decisions. The interactive view allows doctors to more conveniently screen and analyze surgical feature data, achieving more efficient data processing and decision-making. The interactive view can also improve communication between doctors and patients, display surgical data results in a more detailed form, and enhance trust between doctors and patients.

[0042] Preferably, step S5 includes the following steps:

[0043] Step S51: performing convolution preprocessing on the interactive view of surgical features using a superpixel convolutional network to generate a surgical feature sample cluster;

[0044] Step S52: using a superpixel algorithm to perform convolution data segmentation on the surgical feature sample cluster to generate a low-dimensional convolution feature map of the surgical sample;

[0045] Step S53: using a dilated convolution algorithm to perform edge feature enhancement processing on the low-dimensional convolution feature map of the surgical sample to generate a surgical feature network;

[0046] Step S54: performing spatial pyramid pooling multi-layer sampling on the surgical feature network using a multi-scale sampling algorithm to generate a surgical feature sequence;

[0047] Step S55: Based on the combined classifier algorithm and using the combined classifier weighted comprehensive calculation formula, the surgical convolution feature sequence is modeled using the data mining algorithm based on association rules to generate a surgical convolution feature model.

[0048] The present invention uses a superpixel convolutional network to perform convolution preprocessing on interactive views of surgical features to generate surgical feature sample clusters. Convolution preprocessing of interactive feature views can extract more accurate feature information and reduce noise, thereby generating more refined sample clusters, providing a basis for subsequent enhanced feature extraction and processing. The superpixel algorithm is used to perform convolution data segmentation on the sample clusters, and each sample generates a corresponding low-dimensional convolution feature map, which can reduce the dimension of the processed data while maintaining the original data characteristics. The dilated convolution algorithm can enhance the edge features of the low-dimensional convolution feature map of the surgical sample, enhance the edge features in the convolution feature map, thereby generating a more accurate feature network and making the model more refined. The multi-scale sampling method can sample and downsample features at different scales, thereby generating feature sequences suitable for different scales and resolutions, and more comprehensively representing the distribution of features. By using the combined classifier algorithm and the combined classifier weighted comprehensive calculation formula to model the surgical convolution feature sequence based on association rules, the association and interaction between multiple features are identified and analyzed, a more accurate model is generated, and a more reliable decision-making basis is provided.

[0049] The weighted comprehensive calculation formula of the combined classifier in step S55 is specifically:

[0050]

[0051] Among them, f(x) is the weight coefficient of the combined classifier, i is the i-th base classifier, n is the number of base classifiers, t i is the weight of the i-th base classifier in the combined classifier, h i is the predicted value of the weight by the i-th base classifier, x is the sample value input to the initial base classifier, m is the sum of the predicted results of the base classifier for the result value, v i is the prediction result of the i-th base classifier on the result value, g i is the accuracy of the prediction of the i-th base classifier, N is the number of classification results, v jk is the weight of the j-th base classifier to the k-th base classifier, h k is the classification result of the sample by the k-th base classifier, h k (x) is the value of sample (x) in the kth base classifier.

[0052] The present invention Calculate the weights and sums of all samples in each category. For each category of samples, sum the prediction results h of all base classifiers k Then multiply it by the weight of the jth base classifier to get the sum of all the weights of each category sample. By using multiple base classifiers to analyze and make decisions on the samples, the advantages of different classifiers can be integrated to improve the accuracy of the classifier.jk h k Indicates the contribution of the j-th base classifier to the k-th base classifier, lim x→∞ (1+e -x ) n It is an auxiliary parameter that is set. Its value can be regarded as limiting the effect of the activation function within a range to ensure that the adjustment of the weight will not be too drastic. When calculating the weight, it can be regarded as an adjustment coefficient that controls the degree of influence of each base classifier on the prediction result, so that it can make a more reasonable balance adjustment between different base classifiers and different prediction results. This step is to make a more reasonable weight adjustment for the prediction value of the base classifier. It represents the weighted sum of the output results of all base classifiers and their accuracy. The result reflects the proportion of the prediction of each base classifier in the final output among all base classifiers. The role of is to scale the prediction results to ensure that the range of the prediction results is within a stable range, making the output results of the combined classifier more accurate and reliable.

[0053] Preferably, step S5 includes the following steps:

[0054] Step S541: performing spatial pyramid pooling multi-layer sampling on the surgical feature network using a multi-scale sampling algorithm to generate surgical convolution feature data;

[0055] Step S542: performing convolution feature mapping on the surgical feature convolution feature data to generate a surgical feature convolution vector;

[0056] Step S543: performing vector concatenation using the surgical feature convolution feature vectors to generate a surgical feature convolution sequence;

[0057] The present invention performs spatial pyramid pooling multi-layer sampling on the surgical feature network through a multi-scale sampling algorithm. The sampling algorithm can process the multi-scale and multi-resolution of feature data to obtain more comprehensive information, thereby enhancing the feature expression ability of the model. Convolutional feature mapping is performed on the convolution feature data of surgical features. Convolutional feature mapping can reduce the spatial dimension and extract key features, thereby shortening the feature processing time and improving the accuracy of the model. Vector splicing can be used to merge multiple feature vectors into a feature vector of higher dimensionality, thereby improving the feature representation ability and the classification accuracy of the model.

[0058] Preferably, step S6 includes the following steps:

[0059] Step S61: using a homomorphic encryption algorithm to perform data ciphertext conversion on the surgical convolution feature model to generate a surgical homomorphic ciphertext data model;

[0060] Step S62: homomorphically encrypting the surgical operation homomorphic ciphertext data model using the surgical operation information homomorphic encryption calculation formula to generate a surgical operation homomorphic encryption model;

[0061] Step S63: Performing network scheduling slicing on the surgical homomorphic encryption model based on the mobile Internet using 5G technology to generate multiple surgical homomorphic encryption model slices;

[0062] Step S64: Upload the data of multiple surgical homomorphic encryption model slices to the surgical information data management system to perform management operations.

[0063] The present invention can convert the data of the surgical convolution feature model into ciphertext through the homomorphic encryption algorithm, so that the data can still be calculated after encryption, ensuring data privacy while providing a certain data processing capability, so that surgical information can be processed in an encrypted state. The homomorphic encryption calculation formula can be used to perform encryption operations on key data, thereby ensuring data security. At the same time, necessary calculations can be performed in this state, providing the necessary basic conditions for subsequent operations. 5G technology can provide faster network transmission speeds and lower latency to ensure real-time data encryption and decryption. At the same time, network scheduling slicing can cut large data files into smaller data blocks, which is conducive to distributed computing. Uploading data to the surgical information data management system can complete the centralized storage and management of data, realize effective supervision and management of surgical information, reduce the risk of data dispersion and loss, ensure data reliability and consistency, and also improve data utilization and value.

[0064] The surgical information homomorphic encryption calculation formula in step S62 is specifically:

[0065]

[0066] Among them, Enc pk (x) represents the encryption result obtained by homomorphically encrypting the input data (x) using the public key pk, (x) is the data ciphertext of the input model, gi is the generator of the encryption key of the homomorphic encryption algorithm, ri is the cardinality randomly selected by the encryption algorithm, bi is the modulus of a specific power of x, h(x) is the hash function value, r0 is the hash function random number, h(x) r0 is the hash value obtained by inputting the hash function into the model plaintext data, m1 is the first ciphertext obtained by encryption, and gi m1 Construct the generator of the encryption key for the first ciphertext, h ri is a randomly selected hash function value with a base number, m2 is the first ciphertext taken for encryption, (gi m1*m2 ) is the weight coefficient of the generator of the first and second ciphertexts, hr1*m2 The second ciphertext is based on a randomly selected base hash function value.

[0067] The present invention processes (x) data using a hash function through h(x) to obtain a hash value, and uses a random number r0 of the hash function for obfuscation to ensure the security of the ciphertext and prevent attackers from leaking the plaintext or tampering with the data. Bi is a modulus of a specific power of the plaintext data, which can effectively protect the security of the data and prevent attackers from using the modulus to deduce the plaintext data. Gi is used to represent the generator of the encryption key of the homomorphic encryption algorithm, and ri is a cardinality randomly selected by the encryption algorithm. Through (gi ri ) bi *h(x) r0 , using hash functions and random numbers to encrypt data, these random numbers can effectively increase the randomness of encryption, improve the security and difficulty of encryption algorithms, through Divide the encrypted ciphertext into two parts for encryption, (h r1*m2 ) is the second ciphertext based on a randomly selected cardinality hash function value. The formula has the characteristics of randomness, hash confusion, modulus protection and multi-segment encryption. The homomorphic encryption scheme can effectively protect data security and privacy, and support various processing and calculation operations of encrypted data.

[0068] In one embodiment of the present specification, a surgical operation information data management system and method based on mobile Internet is provided, comprising:

[0069] at least one processor;

[0070] a memory communicatively coupled to the at least one processor;

[0071] The memory stores a computer program that can be executed by at least one processor. The computer program is executed by the at least one processor so that the at least one processor can execute the surgical operation information data management method based on the mobile Internet as described above.

[0072] The present invention establishes a surgical operation information data management system based on mobile Internet, and collects data of the entire surgical process by combining medical equipment such as medical cameras, electrocardiogram and X-ray detectors, including surgical video data, surgical audio data, patient vital signs data, anesthesia detection data and surgical instrument equipment data, thereby improving the quality and reliability of surgical process data. The data collection process can ensure the accuracy and reliability of surgical process data through comprehensive monitoring of medical equipment, providing a solid data foundation for subsequent data processing and analysis. By using technologies such as artificial intelligence and deep learning, the data processing process is optimized. The system can automatically eliminate irrelevant data, extract valuable information, and generate standardized and clean data sets, which can greatly reduce the workload of doctors and researchers, improve data processing efficiency, and use visualization technology. The system uses homomorphic encryption to encrypt data. During data upload and transmission, it protects data through network transmission methods such as 5G technology, which greatly improves data security and privacy, realizes centralized storage and management of data, and avoids data dispersion and improper management. At the same time, the use of cloud storage technology helps to reduce storage costs and improve the efficiency of data storage and management. The surgical information data management system based on mobile Internet can give full play to its technological advantages, realize informatization and intelligence according to the needs and characteristics of the medical industry, and provide hospitals with comprehensive and high-quality medical insurance services. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 This is a flowchart of the steps of the surgical operation information data management method based on mobile Internet of the present invention;

[0074] Figure 2 Detailed implementation flow chart of step S1;

[0075] Figure 3 Detailed implementation flow chart of step S2;

[0076] Figure 4 3 is a flowchart of the detailed implementation steps of step S23. DETAILED DESCRIPTION

[0077] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0078] This application provides a mobile internet-based surgical information data management system and method. The execution entities of the mobile internet-based surgical information data management system and method include, but are not limited to, the following: mechanical equipment, data processing platforms, cloud server nodes, network upload devices, etc. that are equipped with the system, which can be regarded as general computing nodes of this application. The data processing platform includes, but is not limited to, at least one of an audio and image management system, an information management system, and a cloud data management system.

[0079] See also Figures 1 to 4 The present invention provides a method for managing surgical operation information data based on mobile Internet, the method comprising the following steps:

[0080] Step S1: Acquire first surgical operation information data using a medical high-definition camera, a recorder, and medical equipment; the medical equipment includes an electrocardiogram (ECG) device and an X-ray detector; and the surgical operation information data includes surgical operation video data, surgical operation audio data, patient vital sign data, anesthesia detection data, and surgical instrument equipment data.

[0081] Step S2: using an artificial intelligence algorithm to perform data adaptive processing on the first surgical operation information data to generate a second surgical operation information data set;

[0082] Step S3: Perform visual projection on the second surgical operation information dataset using a matrix decomposition method to generate a surgical operation feature matrix projection diagram;

[0083] Step S4: Using a deep learning algorithm to perform data visualization on the surgical information matrix decomposition graph to generate an interactive view of surgical features;

[0084] Step S5: using a superpixel convolutional network to perform dilated convolution and multi-scale sampling on the interactive view of the generated surgical information to generate a surgical convolution feature model;

[0085] Step S6: Use the homomorphic encryption algorithm to homomorphically encrypt the surgical convolution feature model to generate a surgical homomorphic encryption model; use 5G technology based on the mobile Internet to upload the surgical homomorphic encryption model data to the surgical information data management system to perform management operations.

[0086] The present invention provides a surgical operation information data management method based on mobile Internet. By using medical high-definition cameras, recorders and medical equipment to obtain first surgical operation information data, most of the information of the entire surgical process can be recorded in detail, including surgical video data, surgical audio data, etc. These data will help improve the accuracy and efficiency of surgical operations, provide valuable data for related research, and provide resources for subsequent surgical reviews. The first surgical operation information data is adaptively processed using an artificial intelligence algorithm, which can better extract and analyze surgical data and generate a second surgical operation information data set with more accurate data analysis and mining capabilities, which will help improve the quality and efficiency of surgical operations. The second surgical operation information data set is visually projected using a matrix decomposition method to generate a surgical feature matrix projection map, which will help to better visualize and understand surgical information, thereby providing doctors with more accurate judgments and decisions. Using deep The degree learning algorithm performs data visualization processing on the surgical information matrix decomposition diagram to generate an interactive view of surgical features. This view can provide sufficient surgical information to help doctors understand the surgical conditions more comprehensively and deeply, thereby improving the safety and accuracy of the operation. By using the superpixel convolutional network to perform dilated convolution and multi-scale sampling on the generated interactive view of surgical information, a surgical convolution feature model is generated, which will help improve the analysis and judgment capabilities of surgical data and further improve the accuracy and efficiency of the operation. The surgical convolution feature model is encrypted using the homomorphic encryption algorithm to generate a surgical homomorphic encryption model, which will help protect the security and privacy of surgical data, thereby protecting patient privacy and doctor's business secrets. At the same time, it will provide better security for the management and analysis of surgical information data. The data is uploaded to the surgical information data management system to ensure data stability and ensure that the surgical information data is non-redundant and stored in an orderly manner.

[0087] In the embodiment of the present invention, reference Figure 1 The above is a flowchart of a surgical operation information data management system and method based on the mobile Internet according to the present invention. In this example, the steps of the surgical operation information data management method based on the mobile Internet include:

[0088] Step S1: Acquire first surgical operation information data using a medical high-definition camera, a recorder, and medical equipment; the medical equipment includes an electrocardiograph and an X-ray detector, and the surgical operation information data includes surgical operation video data, surgical operation audio data, patient vital signs data, anesthesia detection data, and surgical instrument equipment data.

[0089] In an embodiment of the present invention, appropriate medical high-definition cameras and recorders are selected and installed at different positions and angles depending on the size of the surgical area and the type of surgery to ensure comprehensive and accurate data capture throughout the entire surgical procedure. Appropriate medical equipment, such as electrocardiographs and X-ray detectors, is selected based on the patient's condition. Before surgery, these devices are connected to the data acquisition equipment and their output signals are tested to ensure they are correctly outputting data. Before the surgery begins, the high-definition medical camera and recorder are turned on and ensured to meet hygiene requirements. During the surgery, the angle and orientation of the camera and recorder are adjusted as appropriate to ensure valid and representative data is collected. During the surgery, the medical equipment monitors the patient's vital signs and the effectiveness of anesthesia, such as blood pressure, heart rate, respiration, and blood oxygen concentration. It also records surgical instrument information, such as blade usage time and the number of forceps used. The collected data is then transmitted in real time to a surgical information data management system. The system uses a data processing module to preprocess the data, including data denoising, resampling, and data compression, to reduce data storage space and bandwidth usage.

[0090] Step S2: Using an artificial intelligence algorithm to perform data adaptive processing on the first surgical operation information data to generate a second surgical operation information data set.

[0091] In an embodiment of the present invention, a portion of data is selected from a first surgical operation information data set for training an artificial intelligence algorithm model, including surgical video and audio data, vital signs detection data, anesthesia monitoring data, and surgical instrument equipment data. The data set needs to undergo data preprocessing, including data cleaning, format conversion, feature extraction and labeling, etc., to train an artificial intelligence algorithm model to achieve adaptive processing of surgical procedures and data. The algorithm model needs to identify and analyze surgical data, extract key features and information in the surgical process, such as the transition of surgical stages, the sequence and process of surgical operations, the use of instruments, etc. The first surgical operation information data is adaptively processed using the trained artificial intelligence algorithm model to generate a second surgical operation information data set. The generated second surgical operation information data set and the corresponding algorithm model are deployed into the system and used in actual surgery to achieve adaptive processing of surgical procedures and data, so as to assist doctors in providing more accurate and effective treatment and surgery to patients.

[0092] Step S3: Perform visual projection on the second surgical operation information dataset using a matrix decomposition method to generate a surgical operation feature matrix projection diagram.

[0093] In an embodiment of the present invention, data preprocessing is performed on the second surgical data set, and the data set is processed using the matrix decomposition algorithm principal component analysis (PCA). The PCA algorithm can be used to reduce the dimensionality of the data into a two-dimensional or three-dimensional space for visual projection. The data generated in the second step is projected into a two-dimensional or three-dimensional space and can be represented by a scatter plot or a three-dimensional stereogram. The feature projection map is marked and stored, and the surgical feature matrix projection map is verified and evaluated, which includes verifying the accuracy and reliability of the projection map, and comparing and studying the differences between different types of surgeries to improve the accuracy and precision of subsequent surgeries.

[0094] Step S4: Use a deep learning algorithm to perform data visualization on the surgical information matrix decomposition diagram to generate an interactive view of surgical features.

[0095] In an embodiment of the present invention, an appropriate deep learning algorithm is selected for data visualization. Common deep learning algorithms include convolutional neural networks (CNNs), recurrent neural networks (RNNs), and autoencoders. Before visualizing the data, the deep learning model needs to be trained and optimized multiple times to improve the model's accuracy and performance. This process requires the use of a large amount of surgical information data and optimization using techniques such as backpropagation algorithms. The data generated by the deep learning model is converted into an interactive view. This can be accomplished using common data visualization tools such as D3.js and Matplotlib. This visualization process should be as user-friendly and intuitive as possible to simplify the use by doctors and patients.

[0096] Step S5: Use the superpixel convolutional network to perform dilated convolution and multi-scale sampling on the interactive view of surgical information to generate a surgical convolution feature model.

[0097] In an embodiment of the present invention, an appropriate superpixel convolutional network model is selected as a feature extractor. When using the superpixel convolutional network model to extract features from input data, some necessary preprocessing is required for the input data, such as data augmentation and standardization. Then, the preprocessed data is input into the superpixel convolutional network model to extract data features, and dilated convolution and multi-scale sampling are performed. Dilated convolution can increase the receptive field of the convolutional neural network and extract features in a wider range, while multi-scale sampling can capture information at different scales and improve the robustness and accuracy of the model. In order to improve the accuracy and generalization ability of the convolutional neural network, the model parameters need to be optimized, which can be achieved through the backpropagation algorithm and the stochastic gradient descent algorithm. Finally, a surgical convolution feature model is generated.

[0098] Step S6: Use the homomorphic encryption algorithm to homomorphically encrypt the surgical convolution feature model to generate a surgical homomorphic encryption model; use 5G technology based on the mobile Internet to upload the surgical homomorphic encryption model data to the surgical information data management system to perform management operations.

[0099] In an embodiment of the present invention, a homomorphic encryption algorithm is used to encrypt a surgical convolutional feature model to generate a surgical homomorphic encryption model. The homomorphic encryption process can be completed on a local computer. In this algorithm, the encryption process can use a public key for encryption, and the decryption process can use a private key for decryption. This process can ensure the confidentiality and security of the data. Suitable mobile Internet devices and 5G technology are selected to upload the encrypted surgical homomorphic encryption model to the surgical information data management system. The upload process can be completed through mobile Internet devices and 5G technology. It is necessary to use the corresponding protocols and interfaces of the surgical information data management system for communication. The homomorphic encryption model can be used to protect the privacy and data security of surgical information data. At the same time, data analysis and processing can be performed, such as query, statistics, clustering and other operations of surgical data.

[0100] In the embodiment of the present invention, reference Figure 2 The above is a flowchart of the detailed implementation steps of step S1. In one embodiment of this specification, the detailed implementation steps of step S1 include:

[0101] Step S11: Using a medical high-definition camera to shoot the surgical process and obtain surgical video data;

[0102] Step S12: using a recorder to record the entire surgical process to obtain surgical audio data;

[0103] Step S13: Using an electrocardiograph to detect the patient's vital signs, obtaining the patient's vital sign data and anesthesia detection data. The patient's vital sign data includes body temperature data, heart rate data, blood oxygen saturation data, blood pressure data, brain wave data, and electrocardiogram data;

[0104] Step S14: Use an X-ray detector to perform laser scanning on the surgical instrument equipment data to obtain the surgical instrument equipment data.

[0105] The present invention uses a medical high-definition camera to shoot the surgical process, which can collect high-quality surgical video data and use the video to record the surgical process in detail. These data will help analyze and study the surgical process, thereby further improving the quality and efficiency of surgical operations. The entire surgical process is recorded using a recorder, which can collect detailed surgical audio data. These data will help doctors better understand the surgical process, thereby better conducting subsequent analysis and research. The patient's vital signs are detected using an electrocardiogram device, which can collect the patient's vital signs data and anesthesia detection data. These data will help doctors comprehensively monitor and record the patient's vital signs and anesthesia status, thereby ensuring the safety and accuracy of the surgical process. The surgical instrument equipment data is laser scanned using an X-ray detector, which can collect detailed surgical instrument equipment data. These data will help doctors better understand the use and operation of surgical instruments and equipment, thereby further improving the quality and efficiency of surgical operations. These data will help doctors have a more comprehensive understanding of the surgical process, thereby improving the quality and efficiency of surgical operations, and also provide important data support for related medical research and practice.

[0106] In the embodiment of the present invention, by selecting appropriate medical high-definition cameras, recorders and medical equipment, the surgical operation process can be accurately and clearly filmed. At the same time, considering safety factors, it is necessary to ensure that the camera complies with relevant regulations and standards, install the camera in an appropriate location, and perform correct configuration to ensure its normal operation, protect the privacy of patients and doctors, record the surgical operation process, save the recorded video data in a safe and reliable storage device, and make multiple backups to ensure the security and reliability of the data. The recorder is placed in a safe and stable location, and is correctly configured and tested to ensure effective recording. The recorded audio data is saved in a safe and reliable storage device. The electrocardiograph is correctly placed on the patient and correctly configured and tested to ensure accurate measurement of vital signs data. The X-ray detector is correctly placed near the surgical instrument and is correctly configured and tested to ensure accurate detection of surgical instrument data. The detected surgical instrument data is saved in a safe and reliable storage device, and multiple backups are made to ensure the security and reliability of the data.

[0107] In the embodiment of the present invention, reference Figure 3 The above is a flowchart of the detailed implementation steps of step S2. In one embodiment of this specification, the detailed implementation steps of step S2 include:

[0108] Step S21: using a machine vision algorithm based on an artificial intelligence algorithm to perform dynamic target tracking on the surgical video data to generate a dynamic visual image of the surgical operation;

[0109] Step S22: using a time series algorithm to perform vectorized sequence conversion on the patient's vital sign data to generate surgical operation vector sequence data;

[0110] Step S23: performing image segmentation on the surgical dynamic visual image using an image segmentation algorithm to generate an organ-tissue segmentation dataset;

[0111] Step S24: Using a feature fusion algorithm to perform heterogeneous data fusion on the surgical vector sequence data and the organ tissue segmentation data set to generate a second surgical data set.

[0112] The present invention uses a machine vision algorithm based on an artificial intelligence algorithm to perform dynamic target tracking on surgical video data, which helps doctors to understand the patient's internal conditions more comprehensively and accurately during the operation, including the location, size and shape of diseased tissues and organs, the location of surgical instruments and related changes, etc. This information can improve the safety and accuracy of the operation, and also help to reduce the operation time and the patient's recovery time. The patient's vital signs data are vectorized and converted into a sequence using a time series algorithm to generate surgical vector sequence data. This data includes the patient's vital signs indicators, such as heart rate, blood pressure, respiratory rate, anesthetic dosage, etc. By systematically recording this data, doctors can better monitor the patient's physiological state and predict possible physiological reactions and complications. The dynamic visual images of the surgical operation are cut using an image cutting algorithm to generate an organ-tissue segmentation dataset. This process uses relevant Algorithms and models segment dynamic visual images into different regions and tissues. These data sets help doctors better understand the location and distribution of disease sites and related pathological tissues, provide better navigation and control for surgery, and help accurately judge the range of resection and the patient's postoperative recovery. Feature fusion algorithms are used to perform heterogeneous data fusion on surgical vector sequence data and organ tissue segmentation data sets to generate a second surgical data set. Feature fusion technology can fuse data from different sources to improve the integrity, richness and accuracy of the data. In the fusion of surgical data sets, patient vital signs data and organ tissue segmentation data sets can be merged with dynamic visual image data to produce more comprehensive and accurate surgical monitoring information. These data sets help doctors to have a deeper understanding of the characteristics and laws of surgery, provide better control and monitoring for surgery, and provide higher quality data support for related medical research and practice.

[0113] In an embodiment of the present invention, a machine vision algorithm is used to perform dynamic target tracking on surgical video data. By analyzing and comparing key frames in the surgical video data, the algorithm can establish the motion trajectory of the target object and track it. By combining the tracked key frames, a dynamic visual image of the surgical operation can be generated, which can provide useful information for subsequent data processing. The patient's vital signs data is imported into a computer for use by a subsequent time series algorithm. By effectively extracting units or features from the vital signs data at different time points, the algorithm can convert the vital signs data into a vector sequence, and save the vector sequence as surgical vector sequence data, providing useful information for subsequent data processing. By analyzing and processing the image, the algorithm can effectively cut organs and tissues in the dynamic visual image of the surgical operation, segment the organs and tissues, and save them as a segmentation dataset, providing useful information for subsequent data processing. The surgical vector sequence data and the organ and tissue segmentation dataset are imported into a computer, and a feature fusion algorithm is used to perform heterogeneous data fusion on these data, fusing multiple different types of data features to improve the quality and accuracy of the information. The fused data is saved as a second surgical dataset, which can be used for subsequent machine learning and deep learning tasks.

[0114] In the embodiment of the present invention, reference Figure 3 The above is a flowchart of the detailed implementation steps of step S23. In one embodiment of this specification, the detailed implementation steps of step S23 include:

[0115] Step S231: using a region growing algorithm to divide the surgical dynamic visual image into pixel areas to generate a surgical dynamic visual similarity area;

[0116] Step S232: using an image cutting algorithm to perform image cutting on the surgical dynamic visual similarity area to generate a surgical area cutting area;

[0117] Step S233: using a feature point detection algorithm to mark organ-tissue feature points on the surgical dynamic visual image, and to obtain organ-tissue feature point coordinate data;

[0118] Step S234: Using the organ-tissue feature point coordinate data, the surgical field cutting area is annotated to generate an organ-tissue segmentation data set.

[0119] The present invention divides the dynamic visual images of surgical operations into pixel areas through the region growing algorithm, thereby realizing cluster analysis of relatively similar areas, distinguishing different organs or tissue areas during surgery, allowing doctors to more accurately select and locate the application range and area of surgical instruments, improving the accuracy and safety of surgery, and also providing more targeted data support for subsequent steps, such as the input data of steps such as image cutting. The image cutting algorithm can be used to divide the dynamic visual images of surgical operations into regions, realizing accurate positioning and differentiation of different organs and tissues. According to the segmented organ-tissue regions, the location and nature of the disease during surgery can be better understood, and the surgical path and actions can be better planned as much as possible. In order to effectively reduce damage to non-lesioned tissues and thus improve the effect and safety of surgery, the feature point detection algorithm can mark the feature points of important organs or tissues in dynamic visual images of surgical operations, so that the "key points" of important tissues in the operation can be extracted from the image, providing better data support in the next step of labeling and segmentation. The feature point coordinate data of the organ tissue is annotated with the image cutting area to generate an organ-tissue segmentation dataset. By combining image segmentation and feature point marking, more accurate organ and tissue segmentation results can be obtained, which provides more accurate and powerful data support for subsequent surgical decisions and operations, and also leaves more accurate surgical records and monitoring for patients with impaired consciousness.

[0120] In an embodiment of the present invention, a seed pixel is selected as a starting point through a region growing algorithm, and the adjacent pixels of the seed pixel are added to the same region. For the newly added pixel, its similarity with the existing pixels in the region is calculated. If the similarity meets the given conditions, it is added to the same region. The above steps are repeated until no new pixels can be added. The surgical dynamic visual image can be segmented into multiple regions with similar pixel features to generate a surgical dynamic visual similarity field. The surgical dynamic visual similarity field can be segmented using an image cutting algorithm, and a surgical field cutting area can be generated to achieve the target segmentation of specific organs and tissues. The image is segmented into sub-regions with similar features, and the method based on graph theory and minimum cut starts from the perspective of global optimization to find the best segmentation. Segmentation results. In dynamic visual images of surgical operations, feature point detection algorithms can be used to mark points with organ-tissue features and obtain coordinate data to further realize organ-tissue segmentation. Currently, commonly used feature point detection algorithms include Harris corner detection algorithm, SIFT algorithm, SURF algorithm, FAST algorithm, etc. These algorithms can find points with specific features in the image and calculate their feature descriptors for subsequent organ-tissue feature point marking and segmentation. The coordinate data of the organ-tissue feature point marking is used to mark the cutting area in the surgical field, determine which specific organ or tissue each pixel or area belongs to, and verify and correct the labeled data in combination with existing manual annotation data or empirical knowledge to ensure the accuracy and stability of the segmentation results.

[0121] In one embodiment of this specification, the specific steps of step S3 are:

[0122] Step S31: performing data preprocessing on the second surgical data set to generate a surgical data preprocessing pipeline, wherein the data preprocessing includes cleaning, integration, and standardization;

[0123] Step S32: performing matrix partitioning on the surgical data preprocessing pipeline to obtain a plurality of surgical data sub-matrices;

[0124] Step S33: performing matrix decomposition on the surgical data submatrix using a non-negative matrix factorization method to generate a principal component matrix and a minimum error matrix;

[0125] Step S34: performing feature extraction based on the principal component matrix and the minimum error matrix to generate a surgical feature weight matrix;

[0126] Step S35: Perform visual projection on the surgical feature weight matrix using a visual projection method to generate a surgical feature matrix projection diagram.

[0127] The present invention can clean, integrate and standardize the surgical data according to the second surgical data set through a data preprocessing pipeline, thereby eliminating outliers and noise in the data, improving data quality, making the data more accurately reflect the characteristics of the surgery, reducing the influence of noise and erroneous data, and better ensuring the accuracy and reliability of subsequent processing. The preprocessed surgical data can be divided into multiple sub-matrices through matrix partitioning, making the data set smaller and easier to process. Dividing a large data set into multiple small sub-data sets makes subsequent calculations more efficient and reduces the overhead of computing resources. The non-negative matrix decomposition method can decompose the surgical data sub-matrix to generate the principal component matrix and the most Small error matrix, thereby extracting the main feature information in the data. By extracting the principal component matrix, the surgical data can be compressed into fewer dimensions, redundant information can be removed, and the main features of the data can be highlighted, making subsequent processing more efficient. Feature extraction can be performed based on the principal component matrix and the minimum error matrix to generate a surgical feature weight matrix, thereby extracting important feature information of the surgical data. Through feature extraction, the key features in the surgical data can be obtained, effectively improving the accuracy and efficiency of tasks such as classification and recognition. Through visual projection, the surgical data can be displayed on a three-dimensional view, allowing doctors to more clearly visualize the feature information of the surgical data and improve the accuracy and efficiency of surgical decision-making.

[0128] In an embodiment of the present invention, when performing surgical data preprocessing, data cleaning is first required, that is, invalid or duplicate data is removed. Data from multiple different data sources need to be integrated together and standardized to make the data comparable and consistent. In order to achieve matrix decomposition, the data in the surgical data preprocessing pipeline needs to be matrix-divided to obtain multiple surgical data sub-matrices. Matrix decomposition usually requires decomposing a large matrix into multiple small matrices to reduce the difficulty and computational complexity of decomposition, while improving the accuracy and interpretability of the decomposition results. The surgical data sub-matrix is decomposed using non-negative matrix decomposition. Non-negative matrix decomposition is a common matrix decomposition method, which decomposes the matrix into a non-negative principal component matrix and a minimum error matrix, where each column of the principal component matrix represents a main pattern or feature. The non-negative matrix decomposition method can be used to extract the surgical data. Feature information can be quickly discovered to influence and regularity these features have on the surgical process, and provide strong support for subsequent data processing and analysis. After matrix decomposition, the main feature information of surgical data can be extracted through the principal component matrix. These main features can reflect the inherent regularity and personalized characteristics of surgical data. The importance of each principal component and the explanatory power of the data are determined through methods such as covariance matrix and singular value decomposition, thereby determining the importance and influence of each feature in the surgical data. The visual projection method is applied to perform visual projection on the surgical feature weight matrix to generate a surgical feature matrix projection diagram. The eigenvectors of surgical data are presented and displayed in a visual way to better show the characteristics and changing regularity of the data. By applying the visual projection method, a surgical feature matrix projection diagram can be generated to provide more effective support for data analysis and decision-making in the surgical process.

[0129] In one embodiment of this specification, the specific steps of step S4 are:

[0130] Step S41: extracting features from the surgical feature matrix projection image using a principal component algorithm to generate surgical feature vector data;

[0131] Step S42: Performing data visualization processing on the surgical feature vector data using a deep learning algorithm to generate a surgical feature visualization view;

[0132] Step S43: Use the JavaScript library to perform interactive processing on the surgical feature visualization view to generate an interactive view of the surgical feature.

[0133] The present invention uses the principal component algorithm to extract features from the surgical feature matrix projection map to generate surgical feature vector data. The feature vector contains the most important feature information in the surgical data and can be used for further analysis and modeling to improve the accuracy and efficiency of surgical decision-making. The deep learning algorithm is used to visualize the surgical feature vector data, and the features of the surgical data are displayed through a visual view, allowing doctors to understand and analyze the surgical data in a more intuitive way, helping doctors make more accurate decisions. The interactive view allows doctors to more conveniently screen and analyze surgical feature data, achieving more efficient data processing and decision-making. The interactive view can also improve communication between doctors and patients, display surgical data results in a more detailed form, and enhance trust between doctors and patients.

[0134] In an embodiment of the present invention, a principal component algorithm is used to extract features, extract the most critical feature information from the matrix, and generate surgical feature vector data. The principal component algorithm includes steps such as calculating a covariance matrix, calculating eigenvectors and eigenvalues, and selecting principal components. This algorithm can effectively extract important features that affect the surgical process. The generated surgical feature vector data is input into a deep learning algorithm, and data visualization is performed using methods such as convolutional neural networks (CNNs) to generate a surgical feature visualization view. The deep learning algorithm can effectively extract features and patterns and present them in an image format, allowing doctors and clinical medical researchers to more intuitively understand the patient's condition and the characteristics of the surgical process. The surgical feature visualization view is interactively processed using a JavaScript library to implement functions such as dynamic updating, zooming, highlighting, and data querying, generating an interactive view of the surgical features.

[0135] In one embodiment of this specification, the specific steps of step S5 are:

[0136] Step S51: performing convolution preprocessing on the interactive view of surgical features using a superpixel convolutional network to generate a surgical feature sample cluster;

[0137] Step S52: using a superpixel algorithm to perform convolution data segmentation on the surgical feature sample cluster to generate a low-dimensional convolution feature map of the surgical sample;

[0138] Step S53: using a dilated convolution algorithm to perform edge feature enhancement processing on the low-dimensional convolution feature map of the surgical sample to generate a surgical feature network;

[0139] Step S54: performing spatial pyramid pooling multi-layer sampling on the surgical feature network using a multi-scale sampling algorithm to generate a surgical feature sequence;

[0140] Step S55: Based on the combined classifier algorithm and using the combined classifier weighted comprehensive calculation formula, the surgical convolution feature sequence is modeled using the data mining algorithm based on association rules to generate a surgical convolution feature model.

[0141] The present invention uses a superpixel convolutional network to perform convolution preprocessing on interactive views of surgical features to generate surgical feature sample clusters. Convolution preprocessing of interactive feature views can extract more accurate feature information and reduce noise, thereby generating more refined sample clusters, providing a basis for subsequent enhanced feature extraction and processing. The superpixel algorithm is used to perform convolution data segmentation on the sample clusters, and each sample generates a corresponding low-dimensional convolution feature map, which can reduce the dimension of the processed data while maintaining the original data characteristics. The dilated convolution algorithm can enhance the edge features of the low-dimensional convolution feature map of the surgical sample, enhance the edge features in the convolution feature map, thereby generating a more accurate feature network and making the model more refined. The multi-scale sampling method can sample and downsample features at different scales, thereby generating feature sequences suitable for different scales and resolutions, and more comprehensively representing the distribution of features. By using the combined classifier algorithm and the combined classifier weighted comprehensive calculation formula to model the surgical convolution feature sequence based on association rules, the association and interaction between multiple features are identified and analyzed, a more accurate model is generated, and a more reliable decision-making basis is provided.

[0142] In an embodiment of the present invention, a superpixel algorithm is used to segment the interactive view of surgical features and divide it into multiple superpixel clusters. The superpixel convolution network in the convolutional neural network is used to perform convolution preprocessing on each superpixel cluster to extract the key features within the cluster and generate surgical feature sample clusters. These sample clusters contain richer feature information and more detailed image information. The superpixel algorithm is used to perform convolution data cutting on the surgical feature sample clusters and divide them into multiple small blocks to generate low-dimensional convolution feature maps of surgical samples. These convolution feature maps contain more abstract and higher-level feature information and can be better used for feature extraction and model training in subsequent steps. In order to further extract local features and edge features of surgical images, the dilated convolution algorithm is used. The low-dimensional convolution feature map of surgical samples is processed to enhance its edge features and detail information, and a surgical feature network is generated. The multi-scale sampling algorithm is used to perform spatial pyramid pooling and multi-layer sampling operations on the surgical feature network, and its feature information is pooled and downsampled according to different proportions to generate a surgical feature sequence containing feature information of different scales and levels, thereby improving the robustness and generalization of the features. The surgical convolution feature sequence is modeled and trained through the weighted comprehensive calculation formula of the combined classifier based on the combined classifier algorithm to generate a surgical convolution feature model. During the model training process, the data mining algorithm based on association rules can effectively mine the association rules between feature sequences, thereby improving the prediction accuracy and interpretability of the model.

[0143] In one embodiment of the present specification, the formula for calculating the anesthesia drug delivery price in step S55 is specifically:

[0144]

[0145] Among them, f(x) is the weight coefficient of the combined classifier, i is the i-th base classifier, n is the number of base classifiers, t i is the weight of the i-th base classifier in the combined classifier, h i is the predicted value of the weight by the i-th base classifier, x is the sample value input to the initial base classifier, m is the sum of the predicted results of the base classifier for the result value, v i is the prediction result of the i-th base classifier on the result value, g i is the accuracy of the prediction of the i-th base classifier, N is the number of classification results, v jk is the weight of the j-th base classifier to the k-th base classifier, h k is the classification result of the sample by the k-th base classifier, h k (x) is the value of sample (x) in the kth base classifier.

[0146] The present invention Calculate the weights and sums of all samples in each category. For each category of samples, sum the prediction results h of all base classifiers k Then multiply it by the weight of the jth base classifier to get the sum of all the weights of each category sample. By using multiple base classifiers to analyze and make decisions on the samples, the advantages of different classifiers can be integrated to improve the accuracy of the classifier. jk h k Indicates the contribution of the j-th base classifier to the k-th base classifier, lim x→∞ (1+e -x ) n It is an auxiliary parameter that is set. Its value can be regarded as limiting the effect of the activation function within a range to ensure that the adjustment of the weight will not be too drastic. When calculating the weight, it can be regarded as an adjustment coefficient that controls the degree of influence of each base classifier on the prediction result, so that it can make a more reasonable balance adjustment between different base classifiers and different prediction results. This step is to make a more reasonable weight adjustment for the prediction value of the base classifier. It represents the weighted sum of the output results of all base classifiers and their accuracy. The result reflects the proportion of the prediction of each base classifier in the final output among all base classifiers. The role of is to scale the prediction results to ensure that the range of the prediction results is within a stable range, making the output results of the combined classifier more accurate and reliable.

[0147] In one embodiment of this specification, the specific steps of step S54 are:

[0148] Step S541: performing spatial pyramid pooling multi-layer sampling on the surgical feature network using a multi-scale sampling algorithm to generate surgical convolution feature data;

[0149] Step S542: performing convolution feature mapping on the surgical feature convolution feature data to generate a surgical feature convolution vector;

[0150] Step S543: Use the surgical feature convolution feature vector to perform vector splicing to generate a surgical feature convolution sequence.

[0151] The present invention performs spatial pyramid pooling multi-layer sampling on the surgical feature network through a multi-scale sampling algorithm. The sampling algorithm can process the multi-scale and multi-resolution of feature data to obtain more comprehensive information, thereby enhancing the feature expression ability of the model. Convolutional feature mapping is performed on the convolution feature data of surgical features. Convolutional feature mapping can reduce the spatial dimension and extract key features, thereby shortening the feature processing time and improving the accuracy of the model. Vector splicing can be used to merge multiple feature vectors into a feature vector of higher dimensionality, thereby improving the feature representation ability and the classification accuracy of the model.

[0152] In an embodiment of the present invention, first, a multi-scale sampling algorithm is used to perform spatial pyramid pooling multi-layer sampling operations on a surgical feature network obtained through convolution preprocessing, low-dimensional convolution feature map processing, and edge feature enhancement processing. Feature maps of different levels and scales are passed to different pooling layers, pooled and downsampled to obtain convolution feature data of multiple different dimensions. Each convolution feature data is passed to a separate fully connected layer and mapped into a feature vector of fixed dimension. Convolution feature mapping is performed on the convolution feature data of multiple different dimensions, and the multiple feature vectors obtained after convolution feature mapping are spliced to generate a surgical convolution feature sequence. This feature sequence contains feature information of different levels and scales and integrates feature information from different perspectives. This feature sequence can be used in subsequent model training and testing.

[0153] In one embodiment of this specification, the specific steps of step S6 are:

[0154] Step S61: using a homomorphic encryption algorithm to perform data ciphertext conversion on the surgical convolution feature model to generate a surgical homomorphic ciphertext data model;

[0155] Step S62: homomorphically encrypting the surgical operation homomorphic ciphertext data model using the surgical operation information homomorphic encryption calculation formula to generate a surgical operation homomorphic encryption model;

[0156] Step S63: Performing network scheduling slicing on the surgical homomorphic encryption model based on the mobile Internet using 5G technology to generate multiple surgical homomorphic encryption model slices;

[0157] Step S64: Upload the data of multiple surgical homomorphic encryption model slices to the surgical information data management system to perform management operations.

[0158] The present invention can convert the data of the surgical convolution feature model into ciphertext through the homomorphic encryption algorithm, so that the data can still be calculated after encryption, ensuring data privacy while providing a certain data processing capability, so that surgical information can be processed in an encrypted state. The homomorphic encryption calculation formula can be used to perform encryption operations on key data, thereby ensuring data security. At the same time, necessary calculations can be performed in this state, providing the necessary basic conditions for subsequent operations. 5G technology can provide faster network transmission speeds and lower latency to ensure real-time data encryption and decryption. At the same time, network scheduling slicing can cut large data files into smaller data blocks, which is conducive to distributed computing. Uploading data to the surgical information data management system can complete the centralized storage and management of data, realize effective supervision and management of surgical information, reduce the risk of data dispersion and loss, ensure data reliability and consistency, and also improve data utilization and value.

[0159] In an embodiment of the present invention, under the protection of the homomorphic encryption algorithm, the surgical convolution feature model is converted into data ciphertext to generate a surgical homomorphic ciphertext data model, and the surgical information homomorphic encryption calculation formula is used to homomorphically encrypt the surgical homomorphic ciphertext data model. Without destroying the data security, the encrypted data is allowed to be calculated and processed, which can facilitate deeper operations and analysis, generate a surgical homomorphic encryption model, and divide it into multiple small homomorphic model slices, which can facilitate data transmission and processing. Based on mobile Internet and 5G technology, the homomorphic encryption model can be network scheduled and sliced, and divided into multiple small homomorphic model slices, which can facilitate data transmission and processing. Finally, the generated surgical homomorphic encryption model slices are uploaded to the surgical information data management system to perform management operations. During the uploading process, the integrity and security of the data need to be guaranteed to prevent data leakage and tampering.

[0160] In one embodiment of the present specification, the surgical information homomorphic encryption calculation formula in step S62 is specifically:

[0161]

[0162] Among them, Enc pk (x) represents the encryption result obtained by homomorphically encrypting the input data (x) using the public key pk, (x) is the data ciphertext of the input model, gi is the generator of the encryption key of the homomorphic encryption algorithm, ri is the cardinality randomly selected by the encryption algorithm, bi is the modulus of a specific power of x, h(x) is the hash function value, r0 is the hash function random number, h(x) r0is the hash value obtained by inputting the hash function into the model plaintext data, m1 is the first ciphertext obtained by encryption, and gi m1 Construct the generator of the encryption key for the first ciphertext, h ri is a randomly selected hash function value with a base number, m2 is the first ciphertext taken for encryption, (gi m1*m2 ) is the weight coefficient of the generator of the first and second ciphertexts, h r1*m2 The second ciphertext is based on a randomly selected base hash function value.

[0163] The present invention processes (x) data using a hash function through h(x) to obtain a hash value, and uses a random number r0 of the hash function for obfuscation to ensure the security of the ciphertext and prevent attackers from leaking the plaintext or tampering with the data. Bi is a modulus of a specific power of the plaintext data, which can effectively protect the security of the data and prevent attackers from using the modulus to deduce the plaintext data. Gi is used to represent the generator of the encryption key of the homomorphic encryption algorithm, and ri is a cardinality randomly selected by the encryption algorithm. Through (gi ri ) bi *h(x) r0 , using hash functions and random numbers to encrypt data, these random numbers can effectively increase the randomness of encryption, improve the security and difficulty of encryption algorithms, through Divide the encrypted ciphertext into two parts for encryption, (h r1*m2 ) is the second ciphertext based on a randomly selected cardinality hash function value. The formula has the characteristics of randomness, hash confusion, modulus protection and multi-segment encryption. The homomorphic encryption scheme can effectively protect data security and privacy, and support various processing and calculation operations of encrypted data.

[0164] In one embodiment of the present specification, a surgical operation information data management system and method based on mobile Internet is provided, comprising:

[0165] at least one processor;

[0166] at least one memory communicatively coupled to the processor;

[0167] The memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor.

[0168] At least one processor executes so that the at least one processor can execute any one of the above methods for managing surgical information data based on mobile Internet.

[0169] The present invention establishes a surgical operation information data management system based on mobile Internet, and collects data of the entire surgical process by combining medical equipment such as medical cameras, electrocardiogram and X-ray detectors, including surgical video data, surgical audio data, patient vital signs data, anesthesia detection data and surgical instrument equipment data, thereby improving the quality and reliability of surgical process data. The data collection process can ensure the accuracy and reliability of surgical process data through comprehensive monitoring of medical equipment, providing a solid data foundation for subsequent data processing and analysis. By using technologies such as artificial intelligence and deep learning, the data processing process is optimized. The system can automatically eliminate irrelevant data, extract valuable information, and generate standardized and clean data sets, which can greatly reduce the workload of doctors and researchers, improve data processing efficiency, and use visualization technology. The system uses homomorphic encryption to encrypt data. During data upload and transmission, it protects data through network transmission methods such as 5G technology, which greatly improves data security and privacy, realizes centralized storage and management of data, and avoids data dispersion and improper management. At the same time, the use of cloud storage technology helps to reduce storage costs and improve the efficiency of data storage and management. The surgical information data management system based on mobile Internet can give full play to its technological advantages, realize informatization and intelligence according to the needs and characteristics of the medical industry, and provide hospitals with comprehensive and high-quality medical insurance services.

[0170] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0171] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0172] The foregoing description is intended only to provide specific embodiments of the present invention, intended to enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A surgical operation information data management method based on mobile Internet, characterized in that: The following steps are involved: Step S1: Acquire first surgical operation information data using a medical high-definition camera, a recorder, and medical equipment; the medical equipment includes an electrocardiogram (ECG) device and an X-ray detector; and the surgical operation information data includes surgical operation video data, surgical operation audio data, patient vital sign data, anesthesia detection data, and surgical instrument equipment data. Step S2: performing data adaptation on the first surgical operation information data using an artificial intelligence algorithm to generate a second surgical operation information data set; Step S3: Perform visual projection on the second surgical operation information dataset using a matrix decomposition method to generate a surgical operation feature matrix projection diagram; Step S4: Using a deep learning algorithm to perform data visualization on the surgical information matrix decomposition graph to generate an interactive view of surgical features; Step S5: using a superpixel convolutional network to perform dilated convolution and multi-scale sampling on the interactive view of the generated surgical information to generate a surgical convolution feature model; Step S6: Use the homomorphic encryption algorithm to homomorphically encrypt the surgical convolution feature model to generate a surgical homomorphic encryption model; use 5G technology based on the mobile Internet to upload the surgical homomorphic encryption model data to the surgical information data management system to perform management operations.

2. The method according to claim 1, characterized in that The specific steps of step S1 are: Step S11: Using a medical high-definition camera to shoot the surgical process and obtain surgical video data; Step S12: using a recorder to record the entire surgical process to obtain surgical audio data; Step S13: Using an electrocardiograph to detect the patient's vital signs, obtaining the patient's vital sign data and anesthesia detection data. The patient's vital sign data includes body temperature data, heart rate data, blood oxygen saturation data, blood pressure data, brain wave data, and electrocardiogram data; Step S14: Use an X-ray detector to perform laser scanning on the surgical instrument equipment data to obtain the surgical instrument equipment data.

3. The method according to claim 2, characterized in that The specific steps of step S2 are: Step S21: using a machine vision algorithm based on an artificial intelligence algorithm to perform dynamic target tracking on the surgical video data to generate a dynamic visual image of the surgical operation; Step S22: using a time series algorithm to perform vectorized sequence conversion on the patient's vital sign data to generate surgical operation vector sequence data; Step S23: performing image segmentation on the surgical dynamic visual image using an image segmentation algorithm to generate an organ-tissue segmentation dataset; Step S24: Using a feature fusion algorithm to perform heterogeneous data fusion on the surgical vector sequence data and the organ tissue segmentation data set to generate a second surgical data set.

4. The method according to claim 3, characterized in that The specific steps of step S23 are: Step S231: using a region growing algorithm to divide the surgical dynamic visual image into pixel areas to generate a surgical dynamic visual similarity area; Step S232: using an image cutting algorithm to perform image cutting on the surgical dynamic visual similarity area to generate a surgical area cutting area; Step S233: using a feature point detection algorithm to mark organ-tissue feature points on the surgical dynamic visual image, and to obtain organ-tissue feature point coordinate data; Step S234: Using the organ-tissue feature point coordinate data, the surgical field cutting area is annotated to generate an organ-tissue segmentation data set.

5. The method according to claim 4, characterized in that The specific steps of step S3 are: Step S31: performing data preprocessing on the second surgical data set to generate a surgical data preprocessing pipeline, wherein the data preprocessing includes cleaning, integration, and standardization; Step S32: performing matrix partitioning on the surgical data preprocessing pipeline to obtain a plurality of surgical data sub-matrices; Step S33: performing matrix decomposition on the surgical data submatrix using a non-negative matrix factorization method to generate a principal component matrix and a minimum error matrix; Step S34: performing feature extraction based on the principal component matrix and the minimum error matrix to generate a surgical feature weight matrix; Step S35: Perform visual projection on the surgical feature weight matrix using a visual projection method to generate a surgical feature matrix projection diagram.

6. The method according to claim 5, characterized in that The specific steps of step S4 are: Step S41: extracting features from the surgical feature matrix projection image using a principal component algorithm to generate surgical feature vector data; Step S42: Performing data visualization processing on the surgical feature vector data using a deep learning algorithm to generate a surgical feature visualization view; Step S43: Use the JavaScript library to perform interactive processing on the surgical feature visualization view to generate an interactive view of the surgical feature.

7. The method according to claim 6, characterized in that The specific steps of step S5 are: Step S51: performing convolution preprocessing on the interactive view of surgical features using a superpixel convolutional network to generate a surgical feature sample cluster; Step S52: using a superpixel algorithm to perform convolution data segmentation on the surgical feature sample cluster to generate a low-dimensional convolution feature map of the surgical sample; Step S53: using a dilated convolution algorithm to perform edge feature enhancement processing on the low-dimensional convolution feature map of the surgical sample to generate a surgical feature network; Step S54: performing spatial pyramid pooling multi-layer sampling on the surgical feature network using a multi-scale sampling algorithm to generate a surgical feature sequence; Step S55: Based on the combined classifier algorithm and using the combined classifier weighted comprehensive calculation formula, the surgical convolution feature sequence is modeled using the data mining algorithm based on association rules to generate a surgical convolution feature model; The weighted comprehensive calculation formula of the combined classifier in step S55 is specifically: Among them, f(x) is the weight coefficient of the combined classifier, i is the i-th base classifier, n is the number of base classifiers, t i is the weight of the i-th base classifier in the combined classifier, h i is the predicted value of the weight by the i-th base classifier, x is the sample value input to the initial base classifier, m is the sum of the predicted results of the base classifier for the result value, v i is the prediction result of the i-th base classifier on the result value, g i is the accuracy of the prediction of the i-th base classifier, N is the number of classification results, v jk is the weight of the j-th base classifier to the k-th base classifier, h k is the classification result of the sample by the k-th base classifier, h k (x) is the value of sample (x) in the kth base classifier.

8. The method according to claim 7, characterized in that The specific steps of step S54 are: Step S541: performing spatial pyramid pooling multi-layer sampling on the surgical feature network using a multi-scale sampling algorithm to generate surgical convolution feature data; Step S542: performing convolution feature mapping on the surgical feature convolution feature data to generate a surgical feature convolution vector; Step S543: Use the surgical feature convolution feature vector to perform vector splicing to generate a surgical feature convolution sequence.

9. The method according to claim 8, characterized in that The specific steps of step S6 are: Step S61: using a homomorphic encryption algorithm to perform data ciphertext conversion on the surgical convolution feature model to generate a surgical homomorphic ciphertext data model; Step S62: homomorphically encrypting the surgical operation homomorphic ciphertext data model using the surgical operation information homomorphic encryption calculation formula to generate a surgical operation homomorphic encryption model; Step S63: Performing network scheduling slicing on the surgical homomorphic encryption model based on the mobile Internet using 5G technology to generate multiple surgical homomorphic encryption model slices; Step S64: Uploading the data of multiple surgical homomorphic encryption model slices to the surgical information data management system to perform management operations; The surgical information homomorphic encryption calculation formula in step S62 is specifically: Among them, Enc pk (x) represents the encryption result obtained by homomorphically encrypting the input data (x) using the public key pk, (x) is the data ciphertext of the input model, gi is the generator of the encryption key of the homomorphic encryption algorithm, ri is the cardinality randomly selected by the encryption algorithm, bi is the modulus of a specific power, h(x) is the hash function value, r0 is the hash function random number, h(x) r0 is the hash value obtained by inputting the hash function into the model plaintext data, m1 is the first ciphertext obtained by encryption, and gi m1 Construct the generator of the encryption key for the first ciphertext, h ri is a randomly selected hash function value with a base number, m2 is the first ciphertext taken for encryption, (gi m1*m2 ) is the weight coefficient of the generator of the first and second ciphertexts, h r1*m2 The second ciphertext is based on a randomly selected base hash function value.

10. A surgical operation information data management system based on mobile Internet, characterized in that: include: at least one processor; at least one memory communicatively coupled to the processor; The memory stores a computer program that can be executed by at least one processor. The computer program is executed by at least one processor to enable the at least one processor to execute the surgical operation information data management method based on mobile Internet as described in any one of claims 1 to 9.

Images