Medical software intelligent health data analysis method based on cloud computing

By combining wolf pack algorithms and variational autocoding models, optimizing the subset of health data features and using cloud computing platforms, multiple technical bottlenecks in the existing medical and health data analysis methods are solved, and efficient and accurate health risk assessment and personalized management are achieved.

CN120432160AInactive Publication Date: 2025-08-05FUZHOU QUANCHUANG NETWORK TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510505313.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-04-22
Publication Date
2025-08-05
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing medical and health data analysis methods have obvious shortcomings in high-dimensional feature extraction capabilities, multi-source data fusion mechanism, dynamic adaptability of model, group intelligence optimization effect, and personalized suggestions generation strategies, resulting in poor accuracy of health risk assessment and disease prediction and personalized management effects.

Method used

Combining the wolf pack algorithm and variational autocoding model, through feature compression and dynamic modeling, we optimize the subset of health data features, use the cloud computing platform to achieve real-time calculation and feedback, and generate personalized health management suggestions and early warning information.

Benefits of technology

It improves the accuracy of health risk assessment and disease prediction, realizes the flexibility and timeliness of personalized health management, and improves the user's health management experience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120432160A_ABST
    Figure CN120432160A_ABST
Patent Text Reader

Abstract

The invention discloses a medical software intelligent health data analysis method based on cloud computing, and the method comprises the following steps: S1, collecting and preprocessing health data of a user, and uploading the health data to a cloud computing platform; s2, performing feature learning and dimension reduction processing on the health data by using a variational self-encoding model, and extracting health data features; s3, initializing positions and speeds of wolf pack individuals by using a variational auto-encoder; s4, evaluating the fitness of each wolf pack individual in the feature space, and searching an optimal health data feature subset; s5, evaluating the health risk of the user, and predicting possible diseases; s6, generating personalized health management suggestions, and providing health early warning information; and S7, providing a dynamic health report for the user. According to the invention, a wolf pack algorithm, a variational self-encoding model and a data analysis and management technology of a cloud computing platform are combined, and intelligent health data analysis of medical software based on cloud computing is realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of medical health technology, and in particular to a cloud computing-based medical software intelligent health data analysis method. Background Art

[0002] In the current healthcare management landscape, the advancement of sensor technology, mobile devices, wearable devices, and electronic health record systems has made it easier for users to access their personal health data, including multi-dimensional health indicators such as heart rate, blood pressure, blood sugar, weight, physical activity, sleep duration, and emotional state. The accumulation of this health data provides a rich foundation for building personalized health analysis and interventions. Simultaneously, the development of emerging technologies such as cloud computing, big data, and artificial intelligence has injected a powerful technological impetus into the healthcare industry. In particular, cloud computing, with its support for real-time transmission of large-scale data, distributed storage, and parallel computing, enables centralized remote processing and dynamic analysis of health data. Consequently, an increasing number of medical software applications are experimenting with AI algorithms to model, evaluate, and predict user health data, thereby assisting users in daily health management and chronic disease prevention.

[0003] In existing research and practical applications, deep learning models, particularly multi-layer neural networks, convolutional neural networks, and long-short-term memory networks, have been widely used in health data analysis tasks. By learning from historical health data, these models can uncover underlying temporal patterns and individual differences, constructing health risk scoring models or disease prediction models. At the same time, some studies have also introduced traditional machine learning methods such as ensemble learning, support vector machines, and logistic regression to screen and classify key features in health data. Although these methods have improved the efficiency and accuracy of health risk assessment to a certain extent, several technical bottlenecks remain that need to be overcome.

[0004] First, existing methods often rely on expert knowledge or static feature engineering during feature extraction, lacking the ability to deeply explore the underlying structure in high-dimensional health data. For example, traditional methods struggle to extract discriminative low-dimensional features from multidimensional, redundant, and noisy sensor data, resulting in unstable subsequent modeling and weak generalization capabilities. Furthermore, most methods lack fusion mechanisms when processing multi-source health data, making it impossible to fully explore the coupling relationships between various types of data, limiting the overall expressive power of health assessment models.

[0005] Secondly, current health data modeling methods mostly use static models. This means that once the model parameters are determined during the training phase, they remain static, making them unable to adapt to the dynamic changes in a user's health status over time. For example, a user's health status may continuously adjust with changes in factors such as dietary habits, exercise intensity, and lifestyle. However, static models lack a real-time update mechanism, resulting in lagging prediction results and disconnected recommendations, which affects the effectiveness of personalized health management. Furthermore, traditional algorithms have poor adaptability to new data and lack the ability to continuously learn, making it impossible to optimize and adjust model strategies based on the latest data.

[0006] Thirdly, data imbalance is a significant issue in health risk assessment and disease prediction tasks. Most users are relatively healthy, while data on high-risk groups or potential patients is relatively scarce. This results in a bias in model training towards the healthy category, making it less able to identify users truly in need of intervention. Furthermore, current models often output scores or classification labels, lacking fine-grained interpretability. This makes it difficult for doctors and users to understand the underlying rationale behind the model's decisions, impacting its usability and trustworthiness in clinical scenarios.

[0007] In terms of model optimization, although a small number of existing studies have introduced swarm intelligence algorithms for health data feature selection and parameter adjustment, most have failed to effectively integrate them with deep learning models. In particular, efficient search and optimization capabilities in multi-dimensional feature spaces remain insufficient, often facing problems such as slow convergence and local optimality. Furthermore, most optimization algorithms fail to fully utilize the advantages of deep models in feature representation, resulting in a disconnect between the model optimization process and the data representation process, making it difficult to achieve end-to-end collaborative optimization.

[0008] Furthermore, the existing systems’ mechanisms for generating health management recommendations are generally crude, relying primarily on fixed rules or templated recommendations based on rating levels, and are unable to dynamically adjust recommendations based on real-time changes in individual health data. This static recommendation method is prone to repeated, redundant, or outdated information push during actual use, seriously impacting the user experience. For users with chronic disease management needs, an intelligent health management mechanism is needed that can continuously optimize intervention plans based on changes in health status, provide real-time feedback on health trends, and warn of risks.

[0009] In summary, the existing medical and health data analysis systems have obvious technical defects in terms of high-dimensional feature extraction capabilities, multi-source data fusion mechanisms, model dynamic adaptability, group intelligence optimization effects, and personalized recommendation generation strategies. Therefore, there is an urgent need to build a health data analysis method that can integrate deep feature learning and intelligent optimization search, has real-time adaptability, and relies on cloud computing platforms to realize the processing, evaluation, prediction and feedback of large-scale health data, so as to achieve truly intelligent, dynamic and personalized health management goals. The present invention is proposed in response to the above problems, and aims to provide a cloud computing-based medical software intelligent health data analysis method, combining variational autoencoders and wolf pack algorithms, by optimizing health data feature subsets, dynamically evaluating health risks, predicting potential diseases, and generating personalized health management recommendations and early warning information, ultimately realizing a health intelligent decision-making system that integrates data-driven, model updating, and strategy feedback.

[0010] Therefore, how to provide an intelligent health data analysis method for medical software based on cloud computing is an urgent problem that technicians in this field need to solve. Summary of the Invention

[0011] One purpose of the present invention is to propose an intelligent health data analysis method for medical software based on cloud computing. The present invention combines the wolf pack algorithm, variational autoencoder model and data analysis and management technology of the cloud computing platform. Through feature compression, feature subset optimization and dynamic modeling of high-dimensional health data, it accurately extracts the user's potential health status and realizes the assessment and prediction of health risks. The system uses variational autoencoders to construct a low-dimensional potential feature space and performs global optimization search of feature subsets through the wolf pack algorithm to improve the accuracy and stability of health modeling. Relying on the real-time computing and dynamic feedback capabilities of the cloud computing platform, the system realizes the continuous tracking of health status, automatic determination of risk level and personalized generation of health management recommendations, with high precision, high scalability and adaptive optimization capabilities.

[0012] According to an embodiment of the present invention, a cloud computing-based medical software intelligent health data analysis method includes the following steps:

[0013] S1, collect and pre-process the user's health data, and upload the health data to the cloud computing platform;

[0014] S2. Use the variational autoencoder model to perform feature learning and dimensionality reduction on the health data. The encoder maps the health data from a high-dimensional space to a low-dimensional latent space, extracts the health data features, and uses the decoder to map the health data features back to the high-dimensional data space for reconstruction.

[0015] S3. Use a variational autoencoder to initialize the position and speed of each wolf. The position of each wolf is the potential feature vector of the health data, and the speed of each wolf is the adjustment amplitude of the health data feature.

[0016] S4, based on the position and speed of individual wolves, evaluate the fitness of each individual wolf in the feature space, automatically adjust the feature weights, and find the best subset of health data features;

[0017] S5. Use the wolf pack algorithm to search and optimize the best subset of health data features, assess the user's health risks, and predict possible diseases;

[0018] S6. Generate personalized health management recommendations based on the assessed health risk results, provide health warning information, and dynamically adjust the recommendations based on changes in the user's health status;

[0019] S7. Upload the health data analysis results and health management suggestions to the cloud computing platform. The cloud platform analyzes and compares the health data of multiple users in real time and provides users with dynamic health reports.

[0020] Optionally, the health data specifically includes blood sugar, blood pressure, heart rate, body temperature, exercise volume and sleep quality.

[0021] Optionally, the S2 specifically includes:

[0022] S21. Use the encoder part of the variational autoencoder model to map the health data from a high-dimensional space to a low-dimensional latent space. Input the health data to the encoder, which gradually compresses the input high-dimensional data into a low-dimensional latent space representation through multiple neural network layers.

[0023] S22. The encoder of the variational autoencoder automatically extracts features from the health data by learning the distribution of the input health data:

[0024]

[0025] Among them, Z is the health data feature output by the variational autoencoder model, X i is the i-th health data, μ i is the mean of the i-th health data, α i is the weighting coefficient of the health data, β is the adjustment index of the health data, γ is the adjustment factor, δ is the nonlinear transformation parameter for enhancing the health data representation, ‖·‖2 is the L2 norm, i is the health data index, N is the total number of health data, W i is the transformation matrix of the i-th health data feature;

[0026] S23, mapping the health data features back to the high-dimensional data space for reconstruction through the decoder of the variational autoencoder. The decoder accepts the health data features as input and transforms them layer by layer through the neural network structure to gradually restore the high-dimensional representation of the data;

[0027] S24. The goal of the decoder is to map the health data features to a space of the same dimension as the original health data through network structures such as activation functions and fully connected layers to generate reconstructed health data; during the reconstruction process, the difference between the reconstructed data and the original health data is calculated and the reconstruction error is used for optimization.

[0028] Optionally, the S3 specifically includes:

[0029] S31. Health data features are used as individual positions in the wolf pack algorithm. The initial position of each wolf pack individual is the health data feature generated by the variational autoencoder, which represents the relationship and dependency between various features in the health data and captures the inherent structure and pattern of the health data.

[0030] S32. Since the goal of each wolf pack individual is to find the optimal solution in the latent space, the position of the wolf pack individual is a representation of the health data;

[0031] S33, the speed of the individual wolves is initialized to the adjustment amplitude of the health data feature, which represents the moving step length of the individual wolves in the latent space. The method for initializing the speed of the individual wolves can be set according to a specific strategy;

[0032] S34. During initialization, the positions and velocities of all wolf individuals are randomly distributed in the latent space to form an initial population;

[0033] S35. The initial position of each wolf pack individual can be randomly generated, and the positions of all individuals will be evenly distributed in the latent space. The initialization of the individual speed of the wolf pack is based on the difference between the current health data characteristics of each individual and the target health data characteristics.

[0034] Optionally, the S4 specifically includes:

[0035] S41. Evaluate the fitness of each wolf individual in the feature space. For each wolf individual, the wolf individual position is the health data feature. The fitness of each wolf individual is calculated by evaluating the quality of the health data feature and the validity of the health data feature subset:

[0036]

[0037] in, For the The fitness of individual wolves, For the Individuals in The potential feature value of each health data feature, is the ideal value of the target health data feature, For the Health data feature weights, is the index of the individual wolf pack, is the index of health data features, For the The weight coefficient of each health data feature, m is the number of health data features, and β is the regularization factor;

[0038] S42. After evaluating the fitness, automatically adjust the health data feature weight of each wolf individual. A fitness-driven feature weight adjustment mechanism is used. Each health data feature weight is dynamically adjusted based on fitness feedback to achieve the optimal health data feature weight. The adjustment of the health data feature weight is determined based on the fitness value of each wolf individual. The higher the fitness value, the larger the health data feature weight.

[0039] S43. The goal of the wolf pack algorithm is to automatically find the best subset of health data features. Through multiple iterative optimization processes, the wolf pack individuals continuously optimize the search of the feature space by adjusting their positions and feature weights, and gradually find the best subset of health data features.

[0040] Optionally, the S5 specifically includes:

[0041] S51. Optimize the best health data feature subset found using the wolf pack algorithm. During the optimization process, update the individual positions and speeds of the wolf pack to further optimize the feature subset:

[0042]

[0043] in, For the The position of each wolf pack individual at time t, where t is the time index, For the The position of each wolf at time t, For the The adjustment factor for each feature, W l is the transformation matrix of the lth health data feature, is the adjusted power index of the health data feature, is the adjustment factor, S il (t) The position of each wolf pack individual on the lth feature dimension, ∈ is a constant to prevent division by zero errors, is the adjustment factor, is the index of the health data feature, the random factor at time t, m is the number of health data features, N is the total number of health data, is the index of the wolf pack individual, l is the dimension that identifies the health data feature, For the The fitness of individual wolves;

[0044] S52. Through the wolf pack algorithm, each individual in the wolf pack adjusts the health data features according to fitness during the optimization process. Fitness automatically feeds back and adjusts feature weights by evaluating the accuracy of the individual's health risk assessment under the current feature subset, so that the optimal feature subset can better represent the key information in the health data. As the iteration proceeds, the feature weights and positions of the individuals will continue to adjust;

[0045] S53. Evaluate the user's health risks and predict possible diseases through the optimized health data feature subset.

[0046] Optionally, the S53 specifically includes:

[0047] S531. Assess the user's health risk based on the optimized health data feature subset. The health data feature subset is optimized using a fitness function to calculate the user's health risk score. The health risk score is used to represent the user's health risk level. The health risk is accurately assessed based on the actual performance of the health data and the health data features previously extracted by the variational autoencoder.

[0048] S532. Based on the health risk score, predict possible diseases. Use the optimized health data feature subset and health risk score to predict the type and probability of the disease that the user may develop. The prediction process is based on the user's health risk score and combines the user's health data features to output the disease prediction result.

[0049] Optionally, the S6 specifically includes:

[0050] S61. Generate personalized health management recommendations based on the assessed health risk results, and calculate the health risk level based on each user's health risk assessment results;

[0051] S62. Generate personalized health management recommendations based on health risk levels. Personalized health management recommendations include dietary adjustments, exercise recommendations, health check frequency, medication recommendations, and lifestyle changes.

[0052] S63: Automatically generate health warning information based on the health risk score and the user's health data characteristics. When the health risk score exceeds the preset T threshold When the user is in a critical condition, a corresponding health warning message will be generated to remind the user of the possible risks of their current health condition and provide relevant preventive measures and warnings. The health warning message also includes possible disease risks and the need to take necessary preventive actions;

[0053] S64. Dynamically adjust the content of health management recommendations based on changes in the user's health status. As the user's health data continues to be updated, the health management recommendations will be continuously adjusted. Each time the user's health data changes, the system will recalculate the health risk assessment results and adjust the health management recommendations based on the latest health status.

[0054] Optionally, the S7 specifically includes:

[0055] S71. Upload the user's health data analysis results, health risk assessment scores, disease prediction results, health management recommendations, and health warning information data to the cloud for storage and management via the cloud computing platform;

[0056] S72. The cloud platform analyzes and compares the health data of multiple users in real time. The cloud computing platform aggregates and analyzes all uploaded health data in real time. By comparing data from multiple users, the platform can identify health trends, changes in group health status, and potential health risks.

[0057] S73. During the analysis process, the platform will generate a comprehensive health analysis report based on each user's health risk score and disease prediction results, comparing the differences between groups;

[0058] S74. The cloud platform generates a health report for each user, which includes health risk assessment results, disease prediction probability, personalized health management suggestions, and health warning information;

[0059] S75. As the user's health data changes, the health report will be updated in real time and provided to the user through the user interface of the cloud platform. The user can view the changing trends of his or her health status and adjust the health management strategy through the cloud platform.

[0060] The beneficial effects of the present invention are:

[0061] This invention combines the wolf pack algorithm, variational autoencoder model, and cloud computing platform technologies to successfully address multiple technical bottlenecks in existing medical and health data analysis methods. By combining the optimization of health data features with deep learning models, the invention achieves efficient feature extraction and feature optimization, enabling more accurate representation of the multidimensional features of health data, thereby improving the accuracy of health risk assessment and disease prediction. The application of variational autoencoders in feature learning and dimensionality reduction not only reduces redundant data but also effectively retains key information in health data, significantly improving the performance of subsequent analysis models.

[0062] Furthermore, the application of the Wolf Pack algorithm to health data feature selection enables the model to dynamically adapt to changes in the user's health status, avoiding the limitations of static feature selection in traditional methods. By globally optimizing subsets of health data features, the Wolf Pack algorithm can effectively find the optimal feature subset, thereby improving the accuracy of health risk assessment and the reliability of disease prediction.

[0063] This invention also enables real-time analysis and dynamic feedback of health data through a cloud computing platform, providing users with a flexible and personalized health management solution. The cloud platform not only generates real-time health reports and health management recommendations, but also automatically adjusts health management strategies based on changes in the user's health status, ensuring that users receive the latest and most appropriate health intervention plans. This intelligent and dynamic health management model not only enhances the user's health management experience but also provides new solutions for the prevention and management of chronic diseases.

[0064] In summary, this invention significantly improves the efficiency and accuracy of health data analysis by combining the wolfpack algorithm with deep learning and cloud computing technologies. The system enables more accurate health risk assessments, more efficient disease predictions, and personalized health management recommendations, offering significant technical advantages and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0066] Figure 1 This is a flowchart of a cloud computing-based medical software intelligent health data analysis method proposed by the present invention;

[0067] Figure 2 This is a schematic diagram of a cloud computing-based intelligent health data analysis method for medical software proposed by the present invention;

[0068] Figure 3 This is a data flow diagram of the cloud computing-based medical software intelligent health data analysis method proposed by the present invention. DETAILED DESCRIPTION

[0069] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0070] refer to Figure 1-3 , a cloud computing-based medical software intelligent health data analysis method, comprising the following steps:

[0071] S1, collect and pre-process the user's health data, and upload the health data to the cloud computing platform;

[0072] S2. Use the variational autoencoder model to perform feature learning and dimensionality reduction on the health data. The encoder maps the health data from a high-dimensional space to a low-dimensional latent space, extracts the health data features, and uses the decoder to map the health data features back to the high-dimensional data space for reconstruction.

[0073] S3. Use a variational autoencoder to initialize the position and speed of each wolf. The position of each wolf is the potential feature vector of the health data, and the speed of each wolf is the adjustment amplitude of the health data feature.

[0074] S4, based on the position and speed of individual wolves, evaluate the fitness of each individual wolf in the feature space, automatically adjust the feature weights, and find the best subset of health data features;

[0075] S5. Use the wolf pack algorithm to search and optimize the best subset of health data features, assess the user's health risks, and predict possible diseases;

[0076] S6. Generate personalized health management recommendations based on the assessed health risk results, provide health warning information, and dynamically adjust the recommendations based on changes in the user's health status;

[0077] S7. Upload the health data analysis results and health management suggestions to the cloud computing platform. The cloud platform analyzes and compares the health data of multiple users in real time and provides users with dynamic health reports.

[0078] The present invention combines the variational autoencoder model with the wolf pack algorithm, adopts the dimensionality reduction and optimization strategy of health data features, accurately extracts key information from health data, and determines the optimal feature subset through global optimization search. In the feature space, by evaluating the fitness of individual wolf pack members, the system can automatically adjust the feature weights to improve the accuracy of health risk assessment. By dynamically adjusting health management recommendations, the present invention can generate personalized health management plans in a timely manner based on the user's real-time health status changes, and analyze the health data of multiple users in real time through a cloud computing platform to provide users with dynamic health reports. This method not only improves the accuracy of health risk prediction, but also ensures the flexibility and timeliness of health management.

[0079] In this embodiment, the health data specifically includes blood sugar, blood pressure, heart rate, body temperature, exercise volume and sleep quality.

[0080] This invention integrates multidimensional health data, including blood sugar, blood pressure, heart rate, temperature, physical activity, and sleep quality, and uses a cloud computing platform to analyze and evaluate this data in real time. Based on a comprehensive analysis of various health indicators, the system accurately assesses a user's health risks and dynamically adjusts personalized management strategies based on health changes. By optimizing data processing and predictive models, this invention achieves more comprehensive and accurate health management, improving the effectiveness of health warnings and personalized interventions.

[0081] In this embodiment, S2 specifically includes:

[0082] S21. Use the encoder part of the variational autoencoder model to map the health data from a high-dimensional space to a low-dimensional latent space. Input the health data to the encoder, which gradually compresses the input high-dimensional data into a low-dimensional latent space representation through multiple neural network layers.

[0083] S22. The encoder of the variational autoencoder automatically extracts features from the health data by learning the distribution of the input health data:

[0084]

[0085] Among them, Z is the health data feature output by the variational autoencoder model, X i is the i-th health data, μ i is the mean of the i-th health data, α i is the weighting coefficient of the health data, β is the adjustment index of the health data, γ is the adjustment factor, δ is the nonlinear transformation parameter for enhancing the health data representation, ‖·‖2 is the L2 norm, i is the health data index, N is the total number of health data, W i is the transformation matrix of the i-th health data feature;

[0086] S23, mapping the health data features back to the high-dimensional data space for reconstruction through the decoder of the variational autoencoder. The decoder accepts the health data features as input and transforms them layer by layer through the neural network structure to gradually restore the high-dimensional representation of the data;

[0087] S24. The goal of the decoder is to map the health data features to a space of the same dimension as the original health data through network structures such as activation functions and fully connected layers to generate reconstructed health data; during the reconstruction process, the difference between the reconstructed data and the original health data is calculated and the reconstruction error is used for optimization.

[0088] The present invention uses a variational autoencoder to perform feature learning and dimensionality reduction of health data. First, the input high-dimensional health data is gradually compressed into a low-dimensional latent space representation through the encoder. This process automatically learns the distribution of health data and extracts key information, making the representation of health data more compact and having higher explanatory power. By optimizing the encoder and decoder structure, the system can effectively restore and reconstruct the data, and minimize the difference between the original data and the reconstructed data, thereby achieving efficient feature compression and recovery. In this way, the system can not only retain the important features in the health data, but also eliminate redundant information, and improve the efficiency and accuracy of data processing. This method can provide more accurate and concise data feature support for subsequent health risk assessment and disease prediction, and enhance the adaptability and stability of the system in real-time health monitoring.

[0089] In this embodiment, S3 specifically includes:

[0090] S31. Health data features are used as individual positions in the wolf pack algorithm. The initial position of each wolf pack individual is the health data feature generated by the variational autoencoder, which represents the relationship and dependency between various features in the health data and captures the inherent structure and pattern of the health data.

[0091] S32. Since the goal of each wolf pack individual is to find the optimal solution in the latent space, the position of the wolf pack individual is a representation of the health data;

[0092] S33, the speed of the individual wolves is initialized to the adjustment amplitude of the health data feature, which represents the moving step length of the individual wolves in the latent space. The method for initializing the speed of the individual wolves can be set according to a specific strategy;

[0093] S34. During initialization, the positions and velocities of all wolf individuals are randomly distributed in the latent space to form an initial population;

[0094] S35. The initial position of each wolf pack individual can be randomly generated, and the positions of all individuals will be evenly distributed in the latent space. The initialization of the individual speed of the wolf pack is based on the difference between the current health data characteristics of each individual and the target health data characteristics.

[0095] The present invention combines health data features with the wolf pack algorithm, and uses the low-dimensional feature representation generated by the variational autoencoder as the individual position of the wolf pack algorithm to achieve an optimized search for health data features. Through the uniform distribution of individual positions and adjustment amplitudes in the latent space, the wolf pack algorithm can explore the deep relationship between health data features and dynamically optimize the feature subset of health data. This process ensures that each individual can effectively find the optimal solution in the latent space by setting appropriate individual speed and position update strategies. The system can not only accurately capture the inherent patterns in health data, but also adjust the search process in real time according to changes in user health data, making the optimization of health data features more flexible and accurate, and providing more accurate feature support for subsequent health risk assessment and personalized health management. The dynamic adjustment capability of this method enables the system to adapt to changes in the health status of different users, thereby improving the adaptability and accuracy of health management.

[0096] In this embodiment, the S4 specifically includes:

[0097] S41. Evaluate the fitness of each wolf individual in the feature space. For each wolf individual, the wolf individual position is the health data feature. The fitness of each wolf individual is calculated by evaluating the quality of the health data feature and the validity of the health data feature subset:

[0098]

[0099] in, For the The fitness of individual wolves, For the Individuals in The potential feature value of each health data feature, is the ideal value of the target health data feature, For the Health data feature weights, is the index of the individual wolf pack, is the index of health data features, For the The weight coefficient of each health data feature, m is the number of health data features, and β is the regularization factor;

[0100] S42. After evaluating the fitness, automatically adjust the health data feature weight of each wolf individual. A fitness-driven feature weight adjustment mechanism is used. Each health data feature weight is dynamically adjusted based on fitness feedback to achieve the optimal health data feature weight. The adjustment of the health data feature weight is determined based on the fitness value of each wolf individual. The higher the fitness value, the larger the health data feature weight.

[0101] S43. The goal of the wolf pack algorithm is to automatically find the best subset of health data features. Through multiple iterative optimization processes, the wolf pack individuals continuously optimize the search of the feature space by adjusting their positions and feature weights, and gradually find the best subset of health data features.

[0102] The present invention combines the wolf pack algorithm with health data feature optimization and adopts a feature weight adjustment mechanism based on fitness feedback, which can accurately evaluate the fitness of each wolf pack individual in the feature space. The system automatically adjusts the weights of health data features by calculating the quality of health data features and the effectiveness of feature subsets to ensure that high-quality features receive greater weight support during the optimization process. This process achieves the precise selection and optimization of health data features. Through multiple iterations and continuous adjustment of individual positions and feature weights, the system can quickly converge to the optimal feature subset. In this way, the system effectively improves the accuracy and reliability of health data analysis, and can perform personalized feature optimization based on the specific health status of each user, providing more accurate data support for subsequent health risk assessments and disease predictions, while ensuring that health management programs are more scientific and efficient.

[0103] In this embodiment, the S5 specifically includes:

[0104] S51. Optimize the best health data feature subset found using the wolf pack algorithm. During the optimization process, update the individual positions and speeds of the wolf pack to further optimize the feature subset:

[0105]

[0106] in, For the The position of each wolf pack individual at time t, where t is the time index, For the The position of each wolf at time t, For the The adjustment factor for each feature, W l is the transformation matrix of the lth health data feature, is the adjusted power index of the health data feature, is the adjustment factor, S il (t) The position of each wolf pack individual on the lth feature dimension, ∈ is a constant to prevent division by zero errors, is the adjustment factor, is the index of the health data feature, the random factor at time t, m is the number of health data features, N is the total number of health data, is the index of the wolf pack individual, l is the dimension that identifies the health data feature, For the The fitness of individual wolves;

[0107] S52. Through the wolf pack algorithm, each individual in the wolf pack adjusts the health data features according to fitness during the optimization process. Fitness automatically feeds back and adjusts feature weights by evaluating the accuracy of the individual's health risk assessment under the current feature subset, so that the optimal feature subset can better represent the key information in the health data. As the iteration proceeds, the feature weights and positions of the individuals will continue to adjust;

[0108] S53. Evaluate the user's health risks and predict possible diseases through the optimized health data feature subset.

[0109] The present invention further optimizes the best health data feature subset that has been found through the wolf pack algorithm, utilizes the individual position and speed update strategy, and continuously refines the selection of health data features by adjusting the transformation matrix and adjustment factor of the feature subset. During the optimization process, the system dynamically adjusts the feature weights based on the fitness feedback of each wolf pack individual to ensure that the optimal feature subset can better represent the key information in the health data. As the iterative process proceeds, the system continuously adjusts the individual position and feature weights according to the accuracy of the health risk assessment, thereby improving the accuracy and stability of the health risk prediction. In this way, the present invention can accurately optimize the feature subset so that each health data feature can reflect the user's health status to the greatest extent, provide more accurate data support for subsequent health assessment and management, and ensure adaptability and flexibility in different health management scenarios.

[0110] In this embodiment, the S53 specifically includes:

[0111] S531. Assess the user's health risk based on the optimized health data feature subset. The health data feature subset is optimized using a fitness function to calculate the user's health risk score. The health risk score is used to represent the user's health risk level. The health risk is accurately assessed based on the actual performance of the health data and the health data features previously extracted by the variational autoencoder.

[0112] S532. Based on the health risk score, predict possible diseases. Use the optimized health data feature subset and health risk score to predict the type and probability of the disease that the user may develop. The prediction process is based on the user's health risk score and combines the user's health data features to output the disease prediction result.

[0113] The present invention accurately assesses the user's health risk through the optimized health data feature subset and health risk score, and predicts the type and probability of possible diseases based on this assessment. The health data feature subset is optimized through the fitness function to ensure that the extracted features accurately reflect the user's health status. Combined with the health features extracted by the variational autoencoder, the system can comprehensively assess health risks, generate a health risk score for the user, and predict possible disease risks based on this score. The prediction process not only takes into account the immediate performance of health data, but also combines the user's historical health records to provide each user with personalized and accurate disease predictions. Through this intelligent analysis process, the present invention ensures the efficiency and foresight of health management, can dynamically adjust health intervention strategies based on real-time health data, and provide a scientific basis for disease prevention and early intervention.

[0114] In this embodiment, S6 specifically includes:

[0115] S61. Generate personalized health management recommendations based on the assessed health risk results, and calculate the health risk level based on each user's health risk assessment results;

[0116] S62. Generate personalized health management recommendations based on health risk levels. Personalized health management recommendations include dietary adjustments, exercise recommendations, health check frequency, medication recommendations, and lifestyle changes.

[0117] S63: Automatically generate health warning information based on the health risk score and the user's health data characteristics. When the health risk score exceeds the preset T threshold When the user is in a critical condition, a corresponding health warning message will be generated to remind the user of the possible risks of their current health condition and provide relevant preventive measures and warnings. The health warning message also includes possible disease risks and the need to take necessary preventive actions;

[0118] S64. Dynamically adjust the content of health management recommendations based on changes in the user's health status. As the user's health data continues to be updated, the health management recommendations will be continuously adjusted. Each time the user's health data changes, the system will recalculate the health risk assessment results and adjust the health management recommendations based on the latest health status.

[0119] The present invention generates personalized health management recommendations based on the results of health risk assessments, and combines the user's health risk level to accurately provide personalized intervention measures such as diet adjustments, exercise recommendations, and health check frequency. The system can monitor the user's health status in real time and dynamically generate health warning information based on the health risk score, prompting the user of possible health risks and providing specific preventive measures. As health data is continuously updated, the system will automatically adjust health management recommendations to ensure that each user's health management plan is always in line with their current health status. In this way, the present invention can not only accurately assess health risks, but also flexibly adjust health management strategies, so that users can always maintain the best health status and effectively prevent the occurrence of potential diseases.

[0120] In this embodiment, the S7 specifically includes:

[0121] S71. Upload the user's health data analysis results, health risk assessment scores, disease prediction results, health management recommendations, and health warning information data to the cloud for storage and management via the cloud computing platform;

[0122] S72. The cloud platform analyzes and compares the health data of multiple users in real time. The cloud computing platform aggregates and analyzes all uploaded health data in real time. By comparing data from multiple users, the platform can identify health trends, changes in group health status, and potential health risks.

[0123] S73. During the analysis process, the platform will generate a comprehensive health analysis report based on each user's health risk score and disease prediction results, comparing the differences between groups;

[0124] S74. The cloud platform generates a health report for each user, which includes health risk assessment results, disease prediction probability, personalized health management suggestions, and health warning information;

[0125] S75. As the user's health data changes, the health report will be updated in real time and provided to the user through the user interface of the cloud platform. The user can view the changing trends of his or her health status and adjust the health management strategy through the cloud platform.

[0126] This invention uses a cloud computing platform to achieve centralized storage and intelligent analysis of health data, ensuring efficient management and real-time processing of health data. The system uploads users' health data analysis results, health risk assessment scores, disease prediction results, health management recommendations, and health warning information to the cloud, utilizing the cloud computing platform to aggregate and analyze the health data of multiple users in real time. By comparing group health data, the platform can identify health trends and potential risks and provide users with group health status analysis reports. Simultaneously, the cloud platform generates personalized health reports based on each user's health data. The reports include health risk assessments, disease predictions, personalized management recommendations, and warning information, enabling each user to obtain real-time and accurate health management plans. As user health data is updated, the platform automatically updates the health report and provides it to the user, allowing them to view changes in their health status in real time and adjust their health management strategies based on the latest information. This approach not only improves the accuracy and efficiency of health data analysis but also makes health management more flexible and personalized.

[0127] Example 1:

[0128] In order to verify the feasibility of the present invention in implementation, the present invention was applied to the health management center of a general hospital. The patient group included 50 patients with a history of chronic diseases such as hypertension, hyperglycemia and heart disease, and the test period was three months. These patients upload their health data to the cloud computing platform every week through the health monitoring equipment provided by the hospital (including blood sugar, blood pressure, weight, heart rate monitors, etc.). The system uses a variational autoencoder to reduce the dimensionality of the health data, extract important health features, and combine the wolf pack algorithm to optimize the health data feature subset. Finally, the user's health risk is evaluated through a deep learning model, and personalized health management recommendations are generated.

[0129] A hospital's health management center is assessing the health risks of chronic disease patients. Each patient has varying degrees of high blood pressure, high blood sugar, and other conditions, and many experience progressively worsening conditions due to irregular lifestyles and a lack of effective health management. Fluctuations in patient health are often influenced by a variety of factors, including dietary habits, exercise levels, and sleep patterns. Therefore, a precise, up-to-date health management plan is essential to help patients manage their conditions and prevent further deterioration.

[0130] By combining a cloud computing platform with health monitoring devices, we achieve real-time collection, transmission, and processing of health data. Patients upload their health data to the platform weekly using smart wearable devices. This data includes health indicators such as blood pressure, blood sugar, heart rate, weight, and physical activity. The platform then stores, analyzes, and evaluates this data. Using a variational autoencoder, the system reduces the dimensionality of the collected high-dimensional data and extracts the most critical features for health risk assessment.

[0131] Next, a wolfpack algorithm is used to optimize these features, automatically selecting the most representative subset of health features. These optimized health features are then fed into a deep learning model to assess each patient's health risk. The model calculates a health risk score based on the patient's health data and, combined with information such as their lifestyle and medical history, generates personalized health management recommendations, such as dietary control, exercise recommendations, and medication adjustments. Furthermore, the system dynamically adjusts recommendations based on changes in health data, ensuring patients receive real-time health guidance.

[0132] Table 1 Comparison of optimization effects of intelligent health data analysis methods for medical software based on cloud computing

[0133]

[0134]

[0135] Table 1 demonstrates that the system successfully identified each patient's health risk and promptly adjusted health management recommendations based on changes in health data. By using key features extracted by a variational autoencoder and feature subsets optimized by the wolfpack algorithm, the system accurately captures the patient's health status and provides personalized, scientific health management recommendations for each patient. In practical applications, patients' health risk scores decreased significantly, with blood sugar and blood pressure levels being effectively controlled, and some patients' health conditions significantly improved.

[0136] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A cloud computing-based medical software intelligent health data analysis method, characterized in that: The steps include: S1, collect and pre-process the user's health data, and upload the health data to the cloud computing platform; S2. Use the variational autoencoder model to perform feature learning and dimensionality reduction on the health data. The encoder maps the health data from a high-dimensional space to a low-dimensional latent space, extracts the health data features, and uses the decoder to map the health data features back to the high-dimensional data space for reconstruction. S3. Use a variational autoencoder to initialize the position and speed of each wolf. The position of each wolf is the potential feature vector of the health data, and the speed of each wolf is the adjustment amplitude of the health data feature. S4, based on the position and speed of individual wolves, evaluate the fitness of each individual wolf in the feature space, automatically adjust the feature weights, and find the best subset of health data features; S5. Use the wolf pack algorithm to search and optimize the best subset of health data features, assess the user's health risks, and predict possible diseases; S6. Generate personalized health management recommendations based on the assessed health risk results, provide health warning information, and dynamically adjust the recommendations based on changes in the user's health status; S7. Upload the health data analysis results and health management suggestions to the cloud computing platform. The cloud platform analyzes and compares the health data of multiple users in real time and provides users with dynamic health reports.

2. The cloud computing-based medical software intelligent health data analysis method according to claim 1, characterized in that: The health data specifically includes blood sugar, blood pressure, heart rate, body temperature, exercise volume and sleep quality.

3. The cloud computing-based medical software intelligent health data analysis method according to claim 1, characterized in that: The S2 specifically includes: S21. Use the encoder part of the variational autoencoder model to map the health data from a high-dimensional space to a low-dimensional latent space. Input the health data to the encoder, which gradually compresses the input high-dimensional data into a low-dimensional latent space representation through multiple neural network layers. S22. The encoder of the variational autoencoder automatically extracts features from the health data by learning the distribution of the input health data: Among them, Z is the health data feature output by the variational autoencoder model, X i is the i-th health data, μ i is the mean of the i-th health data, α i is the weighting coefficient of the health data, β is the adjustment index of the health data, γ is the adjustment factor, δ is the nonlinear transformation parameter for enhancing the health data representation, ‖·‖2 is the L2 norm, i is the health data index, N is the total number of health data, W i is the transformation matrix of the i-th health data feature; S23, mapping the health data features back to the high-dimensional data space for reconstruction through the decoder of the variational autoencoder. The decoder accepts the health data features as input and transforms them layer by layer through the neural network structure to gradually restore the high-dimensional representation of the data; S24. The goal of the decoder is to map the health data features to a space of the same dimension as the original health data through network structures such as activation functions and fully connected layers to generate reconstructed health data; during the reconstruction process, the difference between the reconstructed data and the original health data is calculated and the reconstruction error is used for optimization.

4. The method for intelligent health data analysis of medical software based on cloud computing according to claim 1, characterized in that: The health data characteristics specifically include blood sugar fluctuations, heart rate changes, exercise intensity, and sleep quality.

5. The method for intelligent health data analysis of medical software based on cloud computing according to claim 1, characterized in that: The S3 specifically includes: S31. Health data features are used as individual positions in the wolf pack algorithm. The initial position of each wolf pack individual is the health data feature generated by the variational autoencoder, which represents the relationship and dependency between various features in the health data and captures the inherent structure and pattern of the health data. S32. Since the goal of each wolf pack individual is to find the optimal solution in the latent space, the position of the wolf pack individual is a representation of the health data; S33, the speed of the individual wolves is initialized to the adjustment amplitude of the health data feature, which represents the moving step length of the individual wolves in the latent space. The method for initializing the speed of the individual wolves can be set according to a specific strategy; S34. During initialization, the positions and velocities of all wolf individuals are randomly distributed in the latent space to form an initial population; S35. The initial position of each wolf pack individual can be randomly generated, and the positions of all individuals will be evenly distributed in the latent space. The initialization of the individual speed of the wolf pack is based on the difference between the current health data characteristics of each individual and the target health data characteristics.

6. The method for intelligent health data analysis of medical software based on cloud computing according to claim 1, characterized in that: The S4 specifically includes: S41. Evaluate the fitness of each wolf individual in the feature space. For each wolf individual, the wolf individual position is the health data feature. The fitness of each wolf individual is calculated by evaluating the quality of the health data feature and the validity of the health data feature subset: Among them, F i is the fitness of the i-th wolf individual, Z ij is the potential eigenvalue of the i-th individual on the j-th health data feature, is the ideal value of the target health data feature, α j is the jth health data feature weight, i is the index of the wolf pack individual, j is the index of the health data feature, is the weight coefficient of the jth health data feature, m is the number of health data features, and β is the regularization factor; S42. After evaluating the fitness, automatically adjust the health data feature weight of each wolf individual. A fitness-driven feature weight adjustment mechanism is used. Each health data feature weight is dynamically adjusted based on fitness feedback to achieve the optimal health data feature weight. The adjustment of the health data feature weight is determined based on the fitness value of each wolf individual. The higher the fitness value, the larger the health data feature weight. S43. The goal of the wolf pack algorithm is to automatically find the best subset of health data features. Through multiple iterative optimization processes, the wolf pack individuals continuously optimize the search of the feature space by adjusting their positions and feature weights, and gradually find the best subset of health data features.

7. The method for intelligent health data analysis of medical software based on cloud computing according to claim 1, characterized in that: The S5 specifically includes: S51. Optimize the best health data feature subset found using the wolf pack algorithm. During the optimization process, update the individual positions and speeds of the wolf pack to further optimize the feature subset: Among them, S i (t+1) is the position of the i-th wolf pack individual at time t, t is the time index, S i (t) is the position of the i-th wolf pack individual at time t, α j is the adjustment factor of the jth feature, W l is the transformation matrix of the lth health data feature, u is the adjustment power index of the health data feature, v is the adjustment factor, S il (t) The position of the i-th wolf pack individual on the l-th feature dimension, ∈ is a constant to prevent division by zero errors, v is an adjustment factor, j is the index of the health data feature, the random factor at time t, m is the number of health data features, N is the total number of health data, i is the index of the wolf pack individual, l is the dimension that identifies the health data feature, F i is the fitness of the i-th wolf individual; S52. Through the wolf pack algorithm, each individual in the wolf pack adjusts the health data features according to fitness during the optimization process. Fitness automatically feeds back and adjusts feature weights by evaluating the accuracy of the individual's health risk assessment under the current feature subset, so that the optimal feature subset can better represent the key information in the health data. As the iteration proceeds, the feature weights and positions of the individuals will continue to adjust; S53. Evaluate the user's health risks and predict possible diseases through the optimized health data feature subset.

8. The method for intelligent health data analysis of medical software based on cloud computing according to claim 1, characterized in that: The S53 specifically includes: S531. Assess the user's health risk based on the optimized health data feature subset. The health data feature subset is optimized using a fitness function to calculate the user's health risk score. The health risk score is used to represent the user's health risk level. The health risk is accurately assessed based on the actual performance of the health data and the health data features previously extracted by the variational autoencoder. S532. Based on the health risk score, predict possible diseases. Use the optimized health data feature subset and health risk score to predict the type and probability of the disease that the user may develop. The prediction process is based on the user's health risk score and combines the user's health data features to output the disease prediction result.

9. The cloud computing-based medical software intelligent health data analysis method according to claim 1, characterized in that: The S6 specifically includes: S61. Generate personalized health management recommendations based on the assessed health risk results, and calculate the health risk level based on each user's health risk assessment results; S62. Generate personalized health management recommendations based on health risk levels. Personalized health management recommendations include dietary adjustments, exercise recommendations, health check frequency, medication recommendations, and lifestyle changes. S63: Automatically generate health warning information based on the health risk score and the user's health data characteristics. When the health risk score exceeds the preset T threshold When the user is in a critical condition, a corresponding health warning message will be generated to remind the user of the possible risks of their current health condition and provide relevant preventive measures and warnings. The health warning message also includes possible disease risks and the need to take necessary preventive actions; S64. Dynamically adjust the content of health management recommendations based on changes in the user's health status. As the user's health data continues to be updated, the health management recommendations will be continuously adjusted. Each time the user's health data changes, the system will recalculate the health risk assessment results and adjust the health management recommendations based on the latest health status.

10. The cloud computing-based medical software intelligent health data analysis method according to claim 1, characterized in that: The S7 specifically includes: S71. Upload the user's health data analysis results, health risk assessment scores, disease prediction results, health management recommendations, and health warning information data to the cloud for storage and management via the cloud computing platform; S72. The cloud platform analyzes and compares the health data of multiple users in real time. The cloud computing platform aggregates and analyzes all uploaded health data in real time. By comparing data from multiple users, the platform can identify health trends, changes in group health status, and potential health risks. S73. During the analysis process, the platform will generate a comprehensive health analysis report based on each user's health risk score and disease prediction results, comparing differences between groups; S74. The cloud platform generates a health report for each user, which includes health risk assessment results, disease prediction probability, personalized health management suggestions, and health warning information; S75. As the user's health data changes, the health report will be updated in real time and provided to the user through the user interface of the cloud platform. The user can view the changing trends of his or her health status and adjust the health management strategy through the cloud platform.