Fusion analysis method and system for remote diagnosis data

Abstract

The invention relates to a fusion analysis method and system for remote diagnosis data, and the method comprises the following steps: remotely receiving a multi-source heterogeneous physiological signal sent by a patient through a medical data collection unit; carrying out physiological feature extraction on the multi-source heterogeneous physiological signal to obtain a physiological feature sequence; performing data fusion processing on the physiological feature sequence to obtain a fusion feature matrix; performing pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector; and performing diagnosis result generation on the patient based on the pathological feature vector to obtain a patient health status assessment report, thereby solving the technical problem of difficulty in comprehensive assessment of the health status of the patient caused by limited processing capability of a multi-source heterogeneous physiological signal in a traditional medical system.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to a fusion analysis method and system for remote diagnosis data. Background Art

[0002] In today's healthcare landscape, with the advancement of information technology and the widespread adoption of smart devices, telemedicine has become an essential component of medical services. The demand for remote diagnosis has surged, particularly in the wake of global public health emergencies. This requires not only that medical systems be able to effectively process data from diverse devices and platforms, but also that these data be accurately analyzed while protecting patient privacy. However, the traditional medical system's limited ability to process multi-source, heterogeneous physiological signals makes it difficult to comprehensively assess a patient's health status.

[0003] Specifically, existing technologies face challenges in processing the massive amounts of data collected from various medical devices, wearable sensors, and more. This data often comes in different formats, collection frequencies, and accuracy levels, increasing the complexity of data analysis. Furthermore, the lack of effective data fusion methods makes cross-platform or cross-device data integration particularly difficult, further limiting doctors' ability to gain a deeper understanding of the condition. In this scenario, even with advanced medical equipment and abundant data resources, improper data processing may prevent them from fully realizing their potential, affecting the ultimate diagnostic outcome.

[0004] Given these challenges, developing an efficient and reliable method for remote diagnostic data fusion and analysis is crucial. This method must not only overcome the disparities between heterogeneous data sources and effectively integrate them, but also rapidly and accurately extract key information within a big data environment, providing strong support for physicians. Furthermore, considering individual patient differences, improving diagnostic accuracy through personalized analysis is a pressing issue. Therefore, exploring and implementing new data fusion and analysis technologies suitable for telemedicine scenarios is crucial for improving the quality and efficiency of healthcare services. Summary of the Invention

[0005] The main purpose of the present invention is to provide a fusion analysis method and system for remote diagnostic data, which solves the technical problem that the traditional medical system has limited processing capabilities for multi-source heterogeneous physiological signals, resulting in difficulties in comprehensively evaluating the patient's health status.

[0006] To achieve the above object, the present invention provides a method for fusion analysis of remote diagnostic data, comprising the following steps: remotely receiving multi-source heterogeneous physiological signals sent by the patient through a medical data acquisition unit; Extracting physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence; Performing data fusion processing on the physiological feature sequence to obtain a fusion feature matrix; Performing pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector; A diagnosis result is generated for the patient based on the pathological feature vector to obtain a patient health status assessment report.

[0007] Furthermore, extracting physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence includes: Performing time-frequency domain decomposition processing on the multi-source heterogeneous physiological signals to obtain multi-dimensional physiological signal feature components, and performing nonlinear dynamic feature analysis on the multi-dimensional physiological signal feature components to obtain a physiological signal dynamic feature set; Performing physiological rhythm analysis on the multi-dimensional physiological signal feature components based on the physiological signal dynamic feature set to obtain a multi-scale physiological rhythm feature sequence, and performing temporal correlation analysis on the multi-scale physiological rhythm feature sequence to obtain a physiological signal correlation feature map; Multimodal feature decoupling is performed on the physiological signal associated feature map to obtain a decoupled feature component, and physiological feature reconstruction is performed based on the decoupled feature component to obtain the physiological feature sequence, wherein the physiological feature sequence includes heart rate variability features, blood pressure fluctuation features, blood oxygen saturation features and respiratory rhythm features.

[0008] Furthermore, the data fusion processing is performed on the physiological feature sequence to obtain a fusion feature matrix, including: Performing multi-dimensional feature hierarchical processing on the physiological feature sequence to obtain a hierarchical feature group, and performing time series consistency mapping on the hierarchical feature group to obtain a time series feature mapping matrix, wherein the time series feature mapping matrix includes heart rate-blood pressure coupling features, blood oxygen-respiration synchronization features, and multi-source signal synergy features; performing physiological signal complementary enhancement processing on the hierarchical feature group based on the temporal feature mapping matrix to obtain a complementary enhanced feature set, and performing multimodal feature alignment on the complementary enhanced feature set to obtain an aligned feature sequence; Performing physiological rhythm synchronization analysis on the aligned feature sequence to obtain a rhythm synchronization feature map, and performing multi-scale feature fusion on the rhythm synchronization feature map to obtain an initial fusion feature component; A dynamic temporal correlation analysis is performed on the initial fusion feature components to obtain a temporal correlation feature network, and a feature matrix is ​​reorganized based on the temporal correlation feature network to obtain the fusion feature matrix; wherein the fusion feature matrix includes overall cardiovascular function assessment features, respiratory and circulatory system coordination features, and autonomic nervous system regulation features.

[0009] Furthermore, the performing pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector includes: Performing multi-dimensional pathological feature decomposition on the fusion feature matrix to obtain a pathological feature component set, and performing physiological-pathological correlation mapping on the pathological feature component set to obtain a pathological correlation feature group; Performing pathological pattern recognition on the pathological feature component set based on the pathological association feature group to obtain a pathological pattern feature sequence, and performing multi-scale pathological feature extraction on the pathological pattern feature sequence to obtain a pathological feature hierarchical graph; Performing a pathological evolution trajectory analysis on the pathological feature hierarchical graph to obtain a pathological evolution feature network, and performing a multi-system pathological association analysis on the pathological evolution feature network to obtain a system pathological feature set; A pathology feature vector is constructed based on the system pathology feature set to obtain the pathology feature vector.

[0010] Furthermore, the performing pathology pattern recognition on the pathology feature component set based on the pathology association feature group to obtain a pathology pattern feature sequence includes: Performing multimodal feature space mapping on the pathology-related feature group to obtain a pathology feature space mapping matrix, and performing pathology state boundary detection based on the pathology feature space mapping matrix to obtain a pathology state boundary feature sequence; performing a pathological pattern cluster analysis on the pathological feature component set based on the pathological state demarcation feature sequence to obtain a pathological pattern cluster, and performing a physiological-pathological difference evaluation on the pathological pattern cluster to obtain a pathological difference feature set; Performing a pathological feature sensitivity analysis on the pathological difference feature set to obtain a sensitive pathological feature component, and performing a pathological pattern stability assessment based on the sensitive pathological feature component to obtain a stable pathological pattern feature component; Based on the stable pathology pattern feature component, a multi-dimensional pathology pattern association analysis is performed on the sensitive pathology feature component to obtain a pathology pattern association feature network, and pathology pattern key features are extracted from the pathology pattern association feature network to obtain the pathology pattern feature sequence.

[0011] Furthermore, the pathological feature sensitivity analysis is performed on the pathological difference feature set to obtain a sensitive pathological feature component, including: Performing multi-dimensional feature responsiveness calculation on the pathological difference feature set to obtain a pathological feature responsiveness matrix, and performing feature perturbation analysis based on the pathological feature responsiveness matrix to obtain a pathological feature perturbation sequence; Performing feature contribution analysis on the pathological feature response matrix based on the pathological feature perturbation sequence to obtain a pathological feature contribution map, and performing threshold segmentation processing on the pathological feature contribution map to obtain a key pathological feature set; Performing dynamic temporal correlation analysis on the key pathological feature set to obtain a pathological feature correlation sequence, and performing feature significance evaluation based on the pathological feature correlation sequence to obtain a significant pathological feature group; Performing multi-scale feature weight calculation on the pathological feature correlation sequence based on the significant pathological feature group to obtain a pathological feature weight matrix, and performing feature dominance analysis on the pathological feature weight matrix to obtain a dominant pathological feature set; The advantageous pathological feature set is subjected to multi-dimensional pathological feature sensitivity integration to obtain a pathological sensitivity feature network, and key nodes are extracted based on the pathological sensitivity feature network to obtain the sensitive pathological feature component.

[0012] Furthermore, the generating of a diagnosis result of the patient based on the pathological feature vector to obtain a patient health status assessment report includes: Performing time-series dynamic decomposition on the pathological feature vector to obtain a pathological state evolution sequence, and performing multi-dimensional clinical feature mapping on the pathological state evolution sequence to obtain a clinical symptom association matrix; Performing pathophysiological correlation analysis on the clinical symptom correlation matrix to obtain a system function damage degree vector, and performing multi-level health risk quantification on the system function damage degree vector to obtain a health risk assessment feature group; Performing multi-system collaborative diagnosis analysis based on the health risk assessment feature group to obtain a system collaborative diagnosis feature set, and predicting the pathological state evolution of the system collaborative diagnosis feature set to obtain a disease development trend map; Multi-dimensional clinical diagnostic features are extracted from the disease development trend map to obtain a clinical diagnostic feature vector, and a comprehensive health status assessment of the patient is performed based on the clinical diagnostic feature vector to obtain a health status assessment report of the patient.

[0013] The present invention also provides a fusion analysis system for remote diagnostic data, comprising: A sending module, used for remotely receiving multi-source heterogeneous physiological signals sent by a patient through a medical data acquisition unit; An extraction module, configured to extract physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence; A fusion module, configured to perform data fusion processing on the physiological feature sequence to obtain a fusion feature matrix; An analysis module, configured to perform pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector; A generating module is used to generate a diagnosis result for the patient based on the pathological feature vector to obtain a patient health status assessment report.

[0014] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0015] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any of the above methods are implemented.

[0016] The present invention provides a fusion analysis method for remote diagnostic data, comprising the following steps: remotely receiving multi-source heterogeneous physiological signals sent by a patient through a medical data acquisition unit; extracting physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence; performing data fusion processing on the physiological feature sequence to obtain a fusion feature matrix; performing pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector; generating a diagnosis result for the patient based on the pathological feature vector to obtain a patient health status assessment report. This method solves the technical problem of the limited processing capacity of the traditional medical system for multi-source heterogeneous physiological signals, which leads to difficulties in comprehensively assessing the patient's health status. It implements pathological feature analysis of the patient based on the fusion feature matrix to obtain a pathological feature vector, allowing doctors to gain an in-depth understanding of the patient's health status from multiple dimensions. This method goes beyond the insights provided by a single data source or simple data superposition, and helps to discover the beneficial effects of potential health risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of the steps of a fusion analysis method for remote diagnostic data according to an embodiment of the present invention; Figure 2 This is a structural block diagram of a fusion analysis device for remote diagnostic data according to an embodiment of the present invention; Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0018] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. 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.

[0020] like Figure 1 As shown, Figure 1 This is a schematic diagram of the steps of a fusion analysis method for remote diagnostic data in one embodiment of the present invention; In one embodiment of the present invention, a method for fusion analysis of remote diagnostic data is provided, comprising the following steps: Step S1: remotely receiving multi-source heterogeneous physiological signals sent by a patient through a medical data acquisition unit.

[0021] Specifically, in the fusion analysis method for remote diagnostic data, remotely receiving multi-source heterogeneous physiological signals sent by patients via a medical data acquisition unit (MDA) is a fundamental step in the entire process. Here, a "MDA" refers to a device or system capable of acquiring physiological information from diverse sources. It supports collecting data from various medical devices, such as electrocardiographs, blood pressure monitors, and blood glucose meters. This data may have different formats and standards, thus representing "multi-source heterogeneous physiological signals." To achieve this, it is first necessary to ensure that all participating data sources are compatible with the MDA. This typically involves adopting a unified data transmission protocol or middleware to convert different data formats for centralized processing. For example, in a specific remote health monitoring application scenario, a patient with hypertension and diabetes regularly measures their physiological parameters at home using smart wearable devices (such as smartwatches) and home medical devices (such as electronic blood pressure monitors and blood glucose meters). These devices transmit recorded physiological signals, such as heart rate, blood pressure, and blood glucose levels, in their own unique formats at preset intervals to the patient's smartphone, which acts as the MDA. Subsequently, the mobile app's built-in data integration capabilities automatically identify and parse data streams from various devices, converting them into a unified format before uploading them to the telemedicine provider's server. This ensures that even though the raw data originates from different device types and brands, it is ultimately accurately aggregated, laying a solid foundation for subsequent physiological feature extraction and other analysis steps. This process not only ensures data integrity and accuracy but also greatly improves the efficiency and convenience of medical services, enabling remote doctors to obtain necessary diagnostic information in a timely manner for effective health management.

[0022] Step S2: extracting physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence.

[0023] Specifically, in remote diagnostic data fusion analysis methods, extracting physiological features from multi-source heterogeneous physiological signals to generate physiological feature sequences is crucial, as it directly impacts the effectiveness of subsequent data processing and analysis. Specifically, this step first requires the use of specific algorithms and techniques to identify and extract key physiological features from previously collected multi-source heterogeneous physiological signals. These physiological features may include, but are not limited to, heart rate variability, blood pressure fluctuation range, and blood glucose level trends. These physiological features are medically significant information points extracted from the raw data. To achieve this, machine learning or deep learning methods are often employed. By learning from large amounts of labeled data, models are built to automatically extract relevant physiological features from new data. For example, in the aforementioned scenario of monitoring patients with hypertension and diabetes, assume that a smartwatch records the patient's heart rate, activity level, and sleep quality for a week, while an electronic blood pressure monitor and blood glucose meter provide blood pressure and blood glucose readings three times daily, morning, noon, and evening, respectively. In this context, the physiological feature extraction process involves screening these diverse data streams for indicators closely related to cardiovascular health and blood glucose control. For example, heart rate variability can be calculated as a marker of heart health by analyzing heart rate data; mean arterial pressure and its circadian rhythm changes can be determined based on blood pressure measurement results; and the trend of glycated hemoglobin levels can be assessed based on the values ​​provided by a blood glucose meter. In this way, the physiological feature sequence generated after processing not only contains specific numerical information, but also reflects the change pattern of various physiological parameters over time, providing rich material for subsequent data fusion processing. At the same time, it also provides medical experts with a more intuitive and easy-to-understand overview of the patient's health status, helping to formulate more accurate and effective treatment plans.

[0024] Step S3: performing data fusion processing on the physiological feature sequence to obtain a fusion feature matrix.

[0025] Specifically, in remote diagnostic data fusion analysis methods, data fusion processing of physiological feature sequences to generate a fused feature matrix is ​​a key step in achieving precision medicine. This process first requires integrating various physiological feature sequences extracted from heterogeneous physiological signals from multiple sources according to specific rules. This aims to overcome the limited information provided by a single data source and to uncover the inherent connections between different physiological features. Specifically, this often involves applying mathematical models or algorithms, such as principal component analysis (PCA) and factor analysis, to reduce data dimensionality while retaining the most important information. Furthermore, machine learning techniques such as neural networks can be used for deeper data fusion, thereby constructing a fused feature matrix that comprehensively reflects the patient's health status. Continuing with the aforementioned scenario of monitoring patients with hypertension and diabetes, assume that multiple physiological feature sequences, including heart rate variability, mean arterial pressure, and glycated hemoglobin levels, have been acquired from a smartwatch, an electronic blood pressure monitor, and a blood glucose meter. Next, during the data fusion phase, these sequences are fed into a pre-trained model that can identify and quantify the interactions between these physiological parameters. For example, analysis revealed a significant correlation between heart rate variability and mean arterial pressure, which in turn may be affected by glycated hemoglobin levels. In this way, the originally dispersed physiological feature sequences are fused to form a new fused feature matrix, in which each element not only represents a specific physiological feature value but also implies the interaction between them. This fused feature matrix provides richer and more accurate data support for subsequent pathological feature analysis of patients, enabling doctors to gain a comprehensive perspective on their patients' health status and develop more personalized and effective treatment plans. This process greatly improves the quality and efficiency of medical services, allowing doctors in remote locations to make accurate judgments based on detailed data.

[0026] Step S4: Analyze the pathological characteristics of the patient based on the fusion feature matrix to obtain a pathological feature vector.

[0027] Specifically, pathological feature analysis of patients based on a fused feature matrix to generate a pathological feature vector is a core component of remote diagnostic data fusion analysis methods. It aims to identify potential health issues by deeply mining and analyzing patients' physiological data. First, the fused feature matrix obtained in the previous step contains key physiological features extracted and integrated from multiple heterogeneous physiological signals, such as heart rate variability, blood pressure fluctuation range, and blood glucose level trends. Next, advanced data analysis techniques, such as machine learning algorithms or statistical models, are used to process the fused feature matrix to discover the complex relationships between different physiological features and their associations with specific disease states. For example, algorithms such as support vector machines (SVMs) or random forests can be used to train a model capable of predicting potential pathological features based on the input fused feature matrix. In the aforementioned scenario of monitoring patients with hypertension and diabetes, assume that a fused feature matrix containing information on the interactions between heart rate variability, mean arterial pressure, and glycated hemoglobin levels has been obtained. In this stage, the pathological feature analysis phase evaluates this matrix using a pre-built disease prediction model. The model might analyze whether reduced heart rate variability is directly linked to elevated blood pressure, and further explore whether this link is affected by poor blood sugar control. In this way, the model can identify key combinations of indicators that indicate worsening hypertension or an increased risk of diabetic complications, thereby generating a pathological feature vector. This vector not only quantifies the extent to which each physiological parameter affects the disease state, but also provides a scientific basis for the subsequent formulation of personalized treatment plans. For example, if the pathological feature vector shows that a patient's cardiovascular risk is significantly increased, the doctor can adjust the treatment plan accordingly, strengthen cardiovascular protective measures, and closely monitor blood sugar levels to prevent the occurrence of complications. In this way, through accurate pathological feature analysis, medical resources can be allocated more efficiently, and patient health management can be more personalized and effective.

[0028] Step S5: Generate a diagnosis result for the patient based on the pathological feature vector to obtain a patient health status assessment report.

[0029] Specifically, generating a patient diagnosis based on pathological feature vectors to produce a patient health status assessment report is the final step in the remote diagnostic data fusion analysis method. Its goal is to transform complex physiological and pathological data analysis into easily understandable and actionable medical recommendations. First, pathological feature vectors are extracted from the fused feature matrix. These vectors quantify the impact of key physiological parameters on the disease state, providing physicians with in-depth insights into the patient's health status. Next, by integrating clinical guidelines, expert knowledge, and the latest medical research findings, a specific algorithm or rule system is used to interpret these pathological feature vectors, generating detailed diagnostic results and treatment recommendations. For example, a rule-based reasoning system or machine learning model can be used to automatically match each indicator in the pathological feature vector to the corresponding disease pattern, while also considering individual patient differences such as age and gender to develop a personalized health management plan. Continuing with the aforementioned monitoring scenario for patients with hypertension and diabetes, assume that the pathological feature vectors indicate significantly reduced heart rate variability, elevated mean arterial pressure, and a glycated hemoglobin level within the risk range. On this basis, the diagnostic result generation step will integrate this information and refer to relevant clinical guidelines, such as best practice guidelines for hypertension and diabetes, to determine the specific diagnostic conclusion. If the analysis shows that the patient has an increased risk of cardiovascular disease, the system will automatically generate a health status assessment report that includes strengthening blood pressure control measures and adjusting diet structure to improve blood sugar management. In addition, the report will also list in detail lifestyle adjustment recommendations, drug treatment options and their potential side effects, so that patients and their families can understand and implement them. This not only improves the efficiency and accuracy of medical services, but also promotes effective communication between doctors and patients, making personalized medicine possible, which in turn helps to improve the overall treatment effect and quality of life of patients.

[0030] In a specific embodiment, extracting physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence includes: Performing time-frequency domain decomposition processing on the multi-source heterogeneous physiological signals to obtain multi-dimensional physiological signal feature components, and performing nonlinear dynamic feature analysis on the multi-dimensional physiological signal feature components to obtain a physiological signal dynamic feature set; Performing physiological rhythm analysis on the multi-dimensional physiological signal feature components based on the physiological signal dynamic feature set to obtain a multi-scale physiological rhythm feature sequence, and performing temporal correlation analysis on the multi-scale physiological rhythm feature sequence to obtain a physiological signal correlation feature map; Multimodal feature decoupling is performed on the physiological signal associated feature map to obtain a decoupled feature component, and physiological feature reconstruction is performed based on the decoupled feature component to obtain the physiological feature sequence, wherein the physiological feature sequence includes heart rate variability features, blood pressure fluctuation features, blood oxygen saturation features and respiratory rhythm features.

[0031] Specifically, in the fusion analysis method for remote diagnostic data, extracting physiological features from multi-source heterogeneous physiological signals to obtain physiological feature sequences is a key step in achieving precision medicine. This process first requires time-frequency domain decomposition of the multi-source heterogeneous physiological signals. This method decomposes the original physiological signals into multi-dimensional feature components, which contain information at different time scales and frequency components. For example, when processing heart rate, blood pressure, and blood oxygen saturation data recorded by a smartwatch for a patient with hypertension and diabetes, time-frequency domain decomposition can reveal heart rate variations across different time periods (such as daytime and nighttime) and how blood oxygen saturation fluctuates with the respiratory cycle. This stage yields dozens or even hundreds of multidimensional physiological signal feature components reflecting the patient's physiological status. Next, nonlinear dynamic feature analysis is performed on these multidimensional physiological signal feature components to explore the complex relationships between the feature components and their temporal evolution patterns. This step helps identify dynamic feature sets that are important for health status. For example, in the aforementioned case, nonlinear analysis of nocturnal heart rate variability and blood pressure fluctuations revealed a specific correlation pattern between the two, which may indicate potential cardiovascular problems. This analysis is not limited to a single parameter but rather involves the interaction between multiple physiological indicators, providing a more comprehensive perspective than traditional methods. Based on this analysis, we construct a set of dozens of dynamic features that reflect subtle changes in the patient's physiological state. Subsequently, based on this set of physiological signal dynamic features, we perform circadian rhythm analysis on the multidimensional physiological signal feature components to obtain a multi-scale circadian rhythm feature sequence. "Multi-scale" here refers to rhythmic changes over time scales from seconds to days, and even longer. For example, when analyzing data from the same patient, we can observe multiple fluctuations in heart rate variability throughout the day, and these fluctuations are closely correlated with daily activity patterns. Furthermore, over the long term, blood pressure levels also show certain trends with seasonal changes. This approach yields a multi-scale circadian rhythm feature sequence containing hundreds of records, each representing the rhythmic characteristics of a physiological parameter within a specific time window. Next, we perform temporal correlation analysis on the multi-scale physiological rhythm feature sequences to construct a physiological signal correlation feature map. This step aims to identify the interrelationships between various physiological rhythm features to better understand how they work together to affect health. For example, when studying the relationship between heart rate variability and respiratory rhythm, we may find that when heart rate variability increases, respiratory rate also shows a corresponding adjustment trend, indicating that there is some inherent connection between the two.By analyzing a large number of data points (e.g., thousands of continuous measurements), we can construct a detailed correlation feature map that clearly demonstrates the complex network structure between different physiological signals. The final step is to perform multimodal feature decoupling on this physiological signal correlation feature map, thereby obtaining decoupled feature components. Based on these components, physiological features are reconstructed, ultimately forming a physiological feature sequence. In this process, we strive to isolate each individual physiological feature, such as heart rate variability, blood pressure fluctuation, blood oxygen saturation, and respiratory rhythm, so that each feature accurately reflects its corresponding physiological phenomenon. For example, after processing the data of the aforementioned patient, we can determine that there are approximately 24 significant peaks and valleys in blood pressure per day, while heart rate variability fluctuates within the normal range of approximately 60 to 100 beats per minute, and blood oxygen saturation remains above 95%. This approach not only enables us to accurately quantify each physiological feature but also integrates it with other relevant information to develop personalized health management plans, significantly improving the quality and efficiency of medical services. The entire process demonstrates the powerful capabilities of modern medical technology in data analysis and provides a solid foundation for the development of personalized medicine.

[0032] In a specific embodiment, the data fusion processing is performed on the physiological feature sequence to obtain a fusion feature matrix, including: Performing multi-dimensional feature hierarchical processing on the physiological feature sequence to obtain a hierarchical feature group, and performing time series consistency mapping on the hierarchical feature group to obtain a time series feature mapping matrix, wherein the time series feature mapping matrix includes heart rate-blood pressure coupling features, blood oxygen-respiration synchronization features, and multi-source signal synergy features; performing physiological signal complementary enhancement processing on the hierarchical feature group based on the temporal feature mapping matrix to obtain a complementary enhanced feature set, and performing multimodal feature alignment on the complementary enhanced feature set to obtain an aligned feature sequence; Performing physiological rhythm synchronization analysis on the aligned feature sequence to obtain a rhythm synchronization feature map, and performing multi-scale feature fusion on the rhythm synchronization feature map to obtain an initial fusion feature component; A dynamic temporal correlation analysis is performed on the initial fusion feature components to obtain a temporal correlation feature network, and a feature matrix is ​​reorganized based on the temporal correlation feature network to obtain the fusion feature matrix; wherein the fusion feature matrix includes overall cardiovascular function assessment features, respiratory and circulatory system coordination features, and autonomic nervous system regulation features.

[0033] Specifically, in the fusion analysis method for remote diagnostic data, the process of fusing physiological feature sequences to generate a fused feature matrix is ​​a crucial step in achieving precision medicine. First, this process requires multi-dimensional feature stratification of the physiological feature sequences to extract meaningful information from complex data. For example, in monitoring a patient with hypertension and diabetes, features such as heart rate variability, blood pressure fluctuations, blood oxygen saturation, and respiratory rhythm can be categorized according to their source, timescale, or physiological significance. This allows for better understanding and management of this information, making subsequent analysis more focused and efficient. This stratification process yields multiple hierarchical feature groups, each containing a specific type of physiological feature, such as changes in heart rate variability over a specific time period or blood pressure fluctuations under different activity states. Next, temporal consistency mapping is performed on these hierarchical feature groups to generate a temporal feature mapping matrix. The key to this step is identifying and quantifying the temporal relationships between individual physiological features, particularly those exhibiting coupled or synchronized characteristics. For example, we may find that a patient's heart rate and blood pressure exhibit a regular pattern of change—when heart rate increases, blood pressure also increases. Similarly, blood oxygen saturation and respiratory rhythm may exhibit a certain degree of synchronization. By analyzing large amounts of data (e.g., weeks or even months of continuous monitoring data), we can construct a mapping matrix containing multiple time-series features. This includes not only heart rate-blood pressure coupling features, blood oxygen-respiration synchronization features, but also multi-source signal synergy features that reflect the synergistic effects of multiple signal sources. This matrix provides a comprehensive view of the patient's physiological state, helping doctors more accurately assess their condition. Based on this time-series feature mapping matrix, the hierarchical feature groups are further subjected to physiological signal complementary enhancement processing to improve the quality and reliability of the original data. This stage involves leveraging the complementarity between different features to fill data gaps or correct outliers. For example, in the above example, if heart rate data is missing for a certain period of time, but blood pressure and respiration data are normal during the same period, the possible heart rate value can be inferred by analyzing the correlation between these available data. This processing results in a complementary enhanced feature set that is more complete and accurate than the original data. This collection is then subjected to multimodal feature alignment, ensuring that data from different sources or at different time points can be compared and analyzed within the same framework, thereby generating an aligned feature sequence. This process is crucial for integrating heterogeneous data from multiple sources, as it eliminates data bias caused by differences in acquisition equipment or time asynchrony. This aligned feature sequence is then analyzed for circadian rhythm synchrony to generate a rhythm synchronization feature map. In this process, we will focus on the synchronization phenomenon between different circadian rhythms and its potential physiological significance.For example, studies have found a relationship between heart rate variability and respiratory rhythm called respiratory sinus arrhythmia (RSA), a normal physiological phenomenon that indicates a healthy functioning of the autonomic nervous system. By analyzing thousands of continuous measurements, we can construct a detailed rhythm synchronization feature map that not only shows the synchronization patterns between various physiological rhythms but also reveals their temporal trends. This feature map is then subjected to multi-scale feature fusion to produce an initial fused feature component. This step allows us to combine features at different time scales to form a comprehensive, holistic view of the patient's physiological state. The final step is to perform dynamic temporal correlation analysis on these initial fused feature components to build a temporal correlation feature network. Based on this network, we reconstruct the feature matrix to ultimately obtain the fused feature matrix. The key here is to understand how different physiological features interact over time and how these interactions influence overall health. For example, when analyzing long-term monitoring data from patients with hypertension and diabetes, we may observe complex interactions between cardiovascular function and autonomic nervous system regulation. By constructing a network that incorporates holistic cardiovascular function assessment features, respiratory and circulatory system synergy features, and autonomic nervous system regulation features, we can gain a deeper understanding of disease progression mechanisms and provide a basis for personalized treatment. The entire process reflects the powerful capabilities of modern medical technology in data analysis, and also lays the foundation for achieving more accurate and effective medical services.

[0034] In a specific embodiment, the performing pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector includes: Performing multi-dimensional pathological feature decomposition on the fusion feature matrix to obtain a pathological feature component set, and performing physiological-pathological correlation mapping on the pathological feature component set to obtain a pathological correlation feature group; Performing pathological pattern recognition on the pathological feature component set based on the pathological association feature group to obtain a pathological pattern feature sequence, and performing multi-scale pathological feature extraction on the pathological pattern feature sequence to obtain a pathological feature hierarchical graph; Performing a pathological evolution trajectory analysis on the pathological feature hierarchical graph to obtain a pathological evolution feature network, and performing a multi-system pathological association analysis on the pathological evolution feature network to obtain a system pathological feature set; A pathology feature vector is constructed based on the system pathology feature set to obtain the pathology feature vector.

[0035] Specifically, in the fusion analysis method for remote diagnostic data, analyzing a patient's pathological characteristics based on a fused feature matrix to obtain a pathological feature vector is a complex and sophisticated process. First, the fused feature matrix must undergo multi-dimensional pathological feature decomposition. This process further refines the information in the fused feature matrix into a set of independent but interrelated pathological feature components. For example, in the case of monitoring patients with hypertension and diabetes, specific feature components related to cardiovascular health, blood sugar control, and autonomic nervous system function can be extracted from the fused feature matrix. These components not only reflect the specific values ​​of physiological parameters but also reveal the inherent connections between them, providing a foundation for subsequent pathological analysis. Next, physiological-pathological correlation mapping is performed on this set of pathological feature components to obtain a pathological correlation feature group. This step aims to establish a direct link between physiological features and underlying pathological conditions. For example, we may find a significant correlation between reduced heart rate variability and increased blood pressure fluctuations, and further explore whether this relationship is related to diabetic microvascular disease. By studying and comparing large amounts of case data (e.g., data from thousands of patients with similar conditions), a detailed set of pathological association features can be constructed. This set not only quantifies the impact of individual physiological features on the disease but also reveals how they work together to lead to specific pathological conditions. Based on this set of pathological association features, pathological pattern recognition is performed on the set of pathological feature components to identify feature sequences that represent specific disease patterns. For example, in the aforementioned example, if analysis reveals that a patient's heart rate variability and blood pressure fluctuations exhibit a specific pattern of changes that aligns with known typical manifestations of early-stage cardiovascular disease, this can be labeled as a pathological pattern feature sequence. This sequence may contain dozens or even hundreds of feature points reflecting different pathological patterns, each corresponding to a specific pathological phenomenon or trend. Multi-scale pathological feature extraction is then performed on this pathological pattern feature sequence to generate a pathological feature hierarchy. "Multi-scale" here refers to pathological features at different levels, from microscopic to macroscopic, including changes at the cellular level, the degree of tissue damage, and even the decline of overall organ function. This analysis provides a comprehensive understanding of the disease progression and its broad impact on the patient's overall health. Next, pathological evolution trajectory analysis is performed on the pathological feature hierarchical graph to depict a pathological evolution feature network. This step focuses on exploring the dynamic process of disease development over time, attempting to identify key turning points and acceleration periods. For example, when studying the development of hypertension and diabetes, we may find that a deterioration in certain physiological indicators (such as blood oxygen saturation) indicates that the disease has entered a more serious stage.Through in-depth analysis of continuous monitoring data over months or even years, a detailed pathological evolution network can be mapped. This network clearly illustrates the various pathways of disease progression and their interrelationships from initial symptoms to final diagnosis. The final step is to construct a pathological feature vector based on this systemic pathological feature set. In this step, we integrate the results of all previous analyses to form a vector that comprehensively describes the patient's pathological status. This vector not only contains information about specific pathological features, such as the degree of cardiovascular damage and blood sugar control, but also reflects the interactions between these features and their impact on overall health. For example, for a patient with hypertension and diabetes, their pathological feature vector may indicate a high cardiovascular risk score accompanied by significant signs of metabolic disorders. This vector provides a scientific basis for physicians to develop personalized treatment plans, helps patients better understand their health status, and promotes effective communication between doctors and patients. The entire process demonstrates the tremendous potential of modern medical technology in data analysis and disease management, and lays a solid foundation for the realization of precision medicine.

[0036] In a specific embodiment, performing pathology pattern recognition on the pathology feature component set based on the pathology association feature group to obtain a pathology pattern feature sequence includes: Performing multimodal feature space mapping on the pathology-related feature group to obtain a pathology feature space mapping matrix, and performing pathology state boundary detection based on the pathology feature space mapping matrix to obtain a pathology state boundary feature sequence; performing a pathological pattern cluster analysis on the pathological feature component set based on the pathological state demarcation feature sequence to obtain a pathological pattern cluster, and performing a physiological-pathological difference evaluation on the pathological pattern cluster to obtain a pathological difference feature set; Performing a pathological feature sensitivity analysis on the pathological difference feature set to obtain a sensitive pathological feature component, and performing a pathological pattern stability assessment based on the sensitive pathological feature component to obtain a stable pathological pattern feature component; Based on the stable pathology pattern feature component, a multi-dimensional pathology pattern association analysis is performed on the sensitive pathology feature component to obtain a pathology pattern association feature network, and pathology pattern key features are extracted from the pathology pattern association feature network to obtain the pathology pattern feature sequence.

[0037] Specifically, in the fusion analysis method for remote diagnostic data, pathological pattern recognition of a set of pathological feature components based on a pathological correlation feature group to obtain a pathological pattern feature sequence is a complex and sophisticated process. First, the pathological correlation feature group must be mapped to a multimodal feature space, transforming pathological features from different sources and types into a unified space to form a pathological feature space mapping matrix. This process enables data from multiple physiological parameters, such as heart rate variability, blood pressure fluctuations, and blood glucose levels, to be compared and analyzed within a common framework, facilitating subsequent processing. For example, in monitoring a patient with hypertension and diabetes, multimodal feature space mapping allows us to place seemingly unrelated features, heart rate variability and blood pressure fluctuations, into the same coordinate system, revealing their potential connection. Furthermore, pathological state boundary detection is performed based on the pathological feature space mapping matrix to determine the boundaries between different pathological states, thereby generating a pathological state boundary feature sequence. These boundary features help identify the critical point between the normal range and early warning signs of disease, which is crucial for early disease detection. Next, pathology pattern cluster analysis is performed on the pathology feature component set based on the pathology state demarcation feature sequences. The goal is to identify patient groups with similar pathology features and divide them into distinct pathology pattern clusters. For example, in the aforementioned example, some patients may exhibit a clear trend of increased cardiovascular risk, while others primarily face metabolic disorders. By performing cluster analysis on the data of thousands of patients with similar conditions, several representative pathology pattern clusters can be identified, each representing a specific pathology state or disease progression pattern. These pathology pattern clusters are then evaluated for physiological-pathology differences to quantify the degree of variation between different pathology patterns and generate a pathology difference feature set. This set not only demonstrates the unique characteristics of each pathology pattern but also provides a basis for understanding the diversity and complexity of disease progression. Subsequently, pathology feature sensitivity analysis is performed on this pathology difference feature set to identify key features that are particularly sensitive to disease progression, thereby forming a sensitive pathology feature component. For example, in studying the progression of hypertension and diabetes, we may find that even small changes in blood oxygen saturation often indicate significant disease transitions. By analyzing large amounts of case data (e.g., tens of thousands of consecutive measurements), we can identify several key sensitive pathological features that are valuable for determining disease progression. Based on these sensitive pathological feature components, we conduct a pathological pattern stability assessment to verify their stability across time and environmental conditions, ultimately generating a stable pathological pattern feature component. This step ensures that the selected features accurately reflect pathological conditions in a variety of situations, enhancing the reliability of diagnostic results.Finally, based on the stable pathological pattern feature components, a multi-dimensional pathological pattern correlation analysis is performed on the sensitive pathological feature components to construct a pathological pattern correlation feature network. This network demonstrates the complex interactions between different pathological features and their impact on overall health. For example, when analyzing long-term monitoring data from patients with hypertension and diabetes, we may find a close synergistic effect between heart rate variability and blood oxygen saturation, which may increase the burden on the cardiovascular system. Through in-depth analysis of these correlation features, a series of key pathological pattern features, namely pathological pattern feature sequences, can be extracted. These features not only reveal the essential characteristics and development patterns of the disease but also provide a scientific basis for developing personalized treatment plans. This entire process demonstrates the powerful capabilities of modern medical technology in data analysis and disease management, and lays the foundation for more precise and effective medical services. This approach enables doctors to make accurate judgments based on detailed data, improving diagnosis and treatment effectiveness, while also helping patients better understand and manage their health.

[0038] In a specific embodiment, performing pathological feature sensitivity analysis on the pathological difference feature set to obtain a sensitive pathological feature component includes: Performing multi-dimensional feature responsiveness calculation on the pathological difference feature set to obtain a pathological feature responsiveness matrix, and performing feature perturbation analysis based on the pathological feature responsiveness matrix to obtain a pathological feature perturbation sequence; Performing feature contribution analysis on the pathological feature response matrix based on the pathological feature perturbation sequence to obtain a pathological feature contribution map, and performing threshold segmentation processing on the pathological feature contribution map to obtain a key pathological feature set; Performing dynamic temporal correlation analysis on the key pathological feature set to obtain a pathological feature correlation sequence, and performing feature significance evaluation based on the pathological feature correlation sequence to obtain a significant pathological feature group; Performing multi-scale feature weight calculation on the pathological feature correlation sequence based on the significant pathological feature group to obtain a pathological feature weight matrix, and performing feature dominance analysis on the pathological feature weight matrix to obtain a dominant pathological feature set; The advantageous pathological feature set is subjected to multi-dimensional pathological feature sensitivity integration to obtain a pathological sensitivity feature network, and key nodes are extracted based on the pathological sensitivity feature network to obtain the sensitive pathological feature component.

[0039] Specifically, in the fusion analysis method for remote diagnostic data, performing pathology feature sensitivity analysis on a set of pathology feature differences to obtain sensitive pathology feature components is a key step. First, a multi-dimensional feature responsiveness calculation is performed on the pathology feature difference set to generate a pathology feature responsiveness matrix. This step aims to quantify the responsiveness of each pathology feature under different conditions. For example, in monitoring patients with hypertension and diabetes, we can assess how features such as heart rate variability, blood pressure fluctuations, and blood glucose levels respond to therapeutic interventions or lifestyle changes. By analyzing data from thousands of patients with similar conditions, a detailed pathology feature responsiveness matrix can be constructed, in which each element represents the potential magnitude of change in a specific pathology feature under certain conditions. Based on this pathology feature responsiveness matrix, feature perturbation analysis is further performed to determine how each feature behaves when subjected to external factors, thereby generating a pathology feature perturbation sequence. This process simulates various possible changes in real-world environments (such as changes in dietary habits or the effects of medications) and observes how these changes affect pathology features. For example, we may find that changes in a patient's diet can lead to significant fluctuations in their blood glucose levels, which in turn have a ripple effect on their blood pressure. In this way, we can identify features that are particularly sensitive to external interference and record them as a pathological feature perturbation sequence, which contains a wealth of information about the dynamic behavior of the pathological features. Next, based on this pathological feature perturbation sequence, we perform feature contribution analysis on the pathological feature response matrix to identify which features have the greatest influence on changes in disease status. This step generates a pathological feature contribution map, which visually displays the importance ranking of each pathological feature. For example, in the above example, we might find that changes in blood oxygen saturation have a high contribution to predicting cardiovascular events, while blood glucose levels primarily affect metabolic function. This pathological feature contribution map is then thresholded to identify the features with the highest contribution, forming a set of key pathological features. This set of features not only serves as core indicators of disease progression but also provides key targets for subsequent analysis. Subsequently, dynamic temporal correlation analysis is performed on this key pathological feature set to reveal patterns in the temporal evolution of these features and generate a pathological feature correlation sequence. For example, when studying long-term monitoring data from patients with hypertension and diabetes, we might discover a close temporal correlation between heart rate variability and respiratory rhythm, which may reflect the health of the autonomic nervous system. By analyzing continuous monitoring data over months or even years, a detailed pathological feature correlation sequence can be established, which not only describes the interactions between different pathological features but also reveals their changing trends over time.Feature significance assessment is performed based on the pathological feature correlation sequence to identify feature groups that are crucial for disease progression, known as the significant pathological feature group. For example, in the aforementioned example, if the analysis shows that a deterioration in certain physiological indicators (such as blood oxygen saturation) indicates a more severe stage of the disease, these can be labeled as significant pathological feature groups. Then, based on the significant pathological feature groups, multi-scale feature weighting is performed on the pathological feature correlation sequence to determine the relative importance of each feature throughout the disease progression, thereby generating a pathological feature weight matrix. This matrix not only quantifies the importance of each feature but also reveals the strength of their effects at different time scales. Next, feature dominance analysis is performed on the pathological feature weight matrix to identify key features that perform well across multiple scenarios, ultimately forming a dominant pathological feature set. For example, in analyzing the progression of hypertension and diabetes, we may find that certain features (such as heart rate variability) show high consistency and reliability in both short-term and long-term monitoring. These features constitute the core of the dominant pathological feature set. Finally, the dominant pathological feature set is integrated with multi-dimensional pathological feature sensitivity to construct a pathological sensitivity feature network. This network comprehensively considers the sensitivity of all features and their interrelationships, forming a model that comprehensively reflects the disease state. Key nodes are extracted based on the pathology sensitivity feature network to identify those features that have the greatest impact on overall health, ultimately yielding a sensitive pathology feature component. These features not only reveal the essential characteristics and progression of the disease but also provide a scientific basis for developing personalized treatment plans. This entire process demonstrates the powerful capabilities of modern medical technology in data analysis and disease management, and lays the foundation for more precise and effective medical services. This approach enables doctors to make accurate judgments based on detailed data, improving diagnosis and treatment outcomes while also helping patients better understand and manage their health.

[0040] In a specific embodiment, the generating of the diagnosis result of the patient based on the pathological feature vector to obtain the patient health status assessment report includes: Performing time-series dynamic decomposition on the pathological feature vector to obtain a pathological state evolution sequence, and performing multi-dimensional clinical feature mapping on the pathological state evolution sequence to obtain a clinical symptom association matrix; Performing pathophysiological correlation analysis on the clinical symptom correlation matrix to obtain a system function damage degree vector, and performing multi-level health risk quantification on the system function damage degree vector to obtain a health risk assessment feature group; Performing multi-system collaborative diagnosis analysis based on the health risk assessment feature group to obtain a system collaborative diagnosis feature set, and predicting the pathological state evolution of the system collaborative diagnosis feature set to obtain a disease development trend map; Multi-dimensional clinical diagnostic features are extracted from the disease development trend map to obtain a clinical diagnostic feature vector, and a comprehensive health status assessment of the patient is performed based on the clinical diagnostic feature vector to obtain a health status assessment report of the patient.

[0041] Specifically, in the fusion analysis method for remote diagnostic data, generating patient diagnostic results based on pathological feature vectors to produce a patient health status assessment report is a systematic and comprehensive process. First, the pathological feature vectors need to be subjected to a time-series dynamic decomposition. This process breaks down the information in the pathological feature vectors into chronological order, revealing the dynamic patterns of disease progression over time and forming a pathological state evolution sequence. For example, in monitoring patients with hypertension and diabetes, we can observe the temporal trends of key indicators such as heart rate variability, blood pressure fluctuations, and blood glucose levels. By conducting in-depth analysis of continuous monitoring data over months or even years, we can identify how these physiological parameters gradually deteriorate or improve over time, thereby constructing a detailed pathological state evolution sequence. Next, this pathological state evolution sequence is subjected to multi-dimensional clinical feature mapping, aiming to link changes in pathological states with specific clinical symptoms, thereby generating a clinical symptom correlation matrix. This step not only considers changes in the physiological parameters themselves, but also focuses on how they impact patients' daily experiences and quality of life. For example, we may find that when heart rate variability decreases, patients experience increased fatigue, while persistently elevated blood pressure may lead to symptoms such as headaches. By comparing and analyzing the data of thousands of patients with similar conditions, a detailed clinical symptom correlation matrix can be established. This matrix not only illustrates the specific symptom manifestations of different pathological conditions but also reveals the underlying connections between them. Subsequently, pathophysiological correlation analysis is conducted based on this clinical symptom correlation matrix, aiming to quantify the degree of impairment in each system's function and generate a system impairment vector. This step focuses on understanding the specific impact of the disease on various body systems (such as the cardiovascular, respiratory, and endocrine systems). For example, in the above example, if the analysis reveals a significant increase in the burden on the cardiovascular system and a decrease in the regulatory capacity of the autonomic nervous system, this information can be integrated into a system impairment vector. This system impairment vector is then subjected to multi-level health risk quantification to determine the patient's overall health risk level and generate a health risk assessment feature set. For example, by analyzing heart rate variability and blood pressure fluctuations, a patient's risk level for cardiovascular events can be assessed. This assessment not only considers current symptoms but also predicts possible future developments. Multi-system collaborative diagnosis analysis, based on this health risk assessment feature set, aims to comprehensively consider the interactions between different systems and their impact on overall health, thereby generating a system collaborative diagnosis feature set. For example, when studying the development of hypertension and diabetes, we may find that the health of the cardiovascular and metabolic systems is closely linked, and deterioration in either system can exacerbate problems in the other. By analyzing large amounts of case data, we can identify key factors that serve as bridges between multiple systems and incorporate them into the system's collaborative diagnostic feature set.Next, the system's collaborative diagnostic feature set is used to predict the pathological state evolution, creating a disease development trend map. This map not only displays the current state of the disease but also predicts its likely path over the coming months or even years, which is crucial for developing long-term treatment plans. Finally, multidimensional clinical diagnostic feature extraction is performed on the disease development trend map to refine the most representative clinical diagnostic feature vectors. These features not only reflect the core issue of the disease but also take into account individual patient differences and living environments. For example, in the above example, the final diagnostic recommendation may be adjusted based on the patient's age, gender, lifestyle, and other information. Based on the clinical diagnostic feature vectors, a comprehensive health assessment of the patient's health status is performed, ultimately generating a patient health status assessment report. This report not only provides a detailed description of the patient's current health status but also includes preventive measures and treatment recommendations for potential future issues. The entire process demonstrates the powerful capabilities of modern medical technology in data analysis and disease management, and lays the foundation for more precise and effective medical services. This approach enables doctors to make accurate judgments based on detailed data, improving diagnosis and treatment effectiveness, while also helping patients better understand and manage their health. The application of this method not only improves the quality of medical services, but also enhances the efficiency of communication between doctors and patients and promotes the development of personalized medicine.

[0042] The above describes the fusion analysis method of remote diagnosis data in the embodiment of the present invention. The following describes the fusion analysis system of remote diagnosis data in the embodiment of the present invention. Figure 2 An embodiment of a remote diagnostic data fusion analysis system according to an embodiment of the present invention includes: A sending module 21 is used to remotely receive multi-source heterogeneous physiological signals sent by a patient through a medical data acquisition unit; An extraction module 22 is configured to extract physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence; A fusion module 23 is used to perform data fusion processing on the physiological feature sequence to obtain a fusion feature matrix; An analysis module 24 is configured to perform pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector; The generating module 25 is configured to generate a diagnosis result for the patient based on the pathological feature vector to obtain a patient health status assessment report.

[0043] In this embodiment, for the specific implementation of each unit in the above system embodiment, please refer to the above method embodiment, which will not be repeated here.

[0044] Reference Figure 3In an embodiment of the present invention, a computer device is also provided, wherein the internal structure of the computer device can be as follows: Figure 3 As shown. The computer device includes a processor, memory, display screen, input device, network interface and database connected via a system bus. The processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the above method is implemented.

[0045] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0046] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which implements the above-described method when executed by a processor. It is understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0047] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware using a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media provided herein and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM.

[0048] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, apparatus, article, or method comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, apparatus, article, or method. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, apparatus, article, or method comprising the element.

[0049] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A fusion analysis method for remote diagnostic data, characterized in that: The following steps are involved: remotely receiving multi-source heterogeneous physiological signals sent by the patient through a medical data acquisition unit; Extracting physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence; Performing data fusion processing on the physiological feature sequence to obtain a fusion feature matrix; Performing pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector; A diagnosis result is generated for the patient based on the pathological feature vector to obtain a patient health status assessment report.

2. The fusion analysis method of remote diagnosis data according to claim 1, characterized in that: The step of extracting physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence includes: Performing time-frequency domain decomposition processing on the multi-source heterogeneous physiological signals to obtain multi-dimensional physiological signal feature components, and performing nonlinear dynamic feature analysis on the multi-dimensional physiological signal feature components to obtain a physiological signal dynamic feature set; Performing physiological rhythm analysis on the multi-dimensional physiological signal feature components based on the physiological signal dynamic feature set to obtain a multi-scale physiological rhythm feature sequence, and performing temporal correlation analysis on the multi-scale physiological rhythm feature sequence to obtain a physiological signal correlation feature map; Multimodal feature decoupling is performed on the physiological signal associated feature map to obtain a decoupled feature component, and physiological feature reconstruction is performed based on the decoupled feature component to obtain the physiological feature sequence, wherein the physiological feature sequence includes heart rate variability features, blood pressure fluctuation features, blood oxygen saturation features and respiratory rhythm features.

3. The fusion analysis method of remote diagnosis data according to claim 1, characterized in that: The performing data fusion processing on the physiological feature sequence to obtain a fusion feature matrix includes: Performing multi-dimensional feature hierarchical processing on the physiological feature sequence to obtain a hierarchical feature group, and performing time series consistency mapping on the hierarchical feature group to obtain a time series feature mapping matrix, wherein the time series feature mapping matrix includes heart rate-blood pressure coupling features, blood oxygen-respiration synchronization features, and multi-source signal synergy features; performing physiological signal complementary enhancement processing on the hierarchical feature group based on the temporal feature mapping matrix to obtain a complementary enhanced feature set, and performing multimodal feature alignment on the complementary enhanced feature set to obtain an aligned feature sequence; Performing physiological rhythm synchronization analysis on the aligned feature sequence to obtain a rhythm synchronization feature map, and performing multi-scale feature fusion on the rhythm synchronization feature map to obtain an initial fusion feature component; A dynamic temporal correlation analysis is performed on the initial fusion feature components to obtain a temporal correlation feature network, and a feature matrix is ​​reorganized based on the temporal correlation feature network to obtain the fusion feature matrix; wherein the fusion feature matrix includes overall cardiovascular function assessment features, respiratory and circulatory system coordination features, and autonomic nervous system regulation features.

4. The fusion analysis method of remote diagnosis data according to claim 1, characterized in that: The performing pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector includes: Performing multi-dimensional pathological feature decomposition on the fusion feature matrix to obtain a pathological feature component set, and performing physiological-pathological correlation mapping on the pathological feature component set to obtain a pathological correlation feature group; Performing pathological pattern recognition on the pathological feature component set based on the pathological association feature group to obtain a pathological pattern feature sequence, and performing multi-scale pathological feature extraction on the pathological pattern feature sequence to obtain a pathological feature hierarchical graph; Performing a pathological evolution trajectory analysis on the pathological feature hierarchical graph to obtain a pathological evolution feature network, and performing a multi-system pathological association analysis on the pathological evolution feature network to obtain a system pathological feature set; A pathology feature vector is constructed based on the system pathology feature set to obtain the pathology feature vector.

5. The fusion analysis method of remote diagnosis data according to claim 4, characterized in that: The performing pathology pattern recognition on the pathology feature component set based on the pathology association feature group to obtain a pathology pattern feature sequence includes: Performing multimodal feature space mapping on the pathology-related feature group to obtain a pathology feature space mapping matrix, and performing pathology state boundary detection based on the pathology feature space mapping matrix to obtain a pathology state boundary feature sequence; performing a pathological pattern cluster analysis on the pathological feature component set based on the pathological state demarcation feature sequence to obtain a pathological pattern cluster, and performing a physiological-pathological difference evaluation on the pathological pattern cluster to obtain a pathological difference feature set; Performing a pathological feature sensitivity analysis on the pathological difference feature set to obtain a sensitive pathological feature component, and performing a pathological pattern stability assessment based on the sensitive pathological feature component to obtain a stable pathological pattern feature component; Based on the stable pathology pattern feature component, a multi-dimensional pathology pattern association analysis is performed on the sensitive pathology feature component to obtain a pathology pattern association feature network, and pathology pattern key features are extracted from the pathology pattern association feature network to obtain the pathology pattern feature sequence.

6. The fusion analysis method of remote diagnosis data according to claim 5, characterized in that: The performing pathological feature sensitivity analysis on the pathological difference feature set to obtain a sensitive pathological feature component includes: Performing multi-dimensional feature responsiveness calculation on the pathological difference feature set to obtain a pathological feature responsiveness matrix, and performing feature perturbation analysis based on the pathological feature responsiveness matrix to obtain a pathological feature perturbation sequence; Performing feature contribution analysis on the pathological feature response matrix based on the pathological feature perturbation sequence to obtain a pathological feature contribution map, and performing threshold segmentation processing on the pathological feature contribution map to obtain a key pathological feature set; Performing dynamic temporal correlation analysis on the key pathological feature set to obtain a pathological feature correlation sequence, and performing feature significance evaluation based on the pathological feature correlation sequence to obtain a significant pathological feature group; Performing multi-scale feature weight calculation on the pathological feature correlation sequence based on the significant pathological feature group to obtain a pathological feature weight matrix, and performing feature dominance analysis on the pathological feature weight matrix to obtain a dominant pathological feature set; The advantageous pathological feature set is subjected to multi-dimensional pathological feature sensitivity integration to obtain a pathological sensitivity feature network, and key nodes are extracted based on the pathological sensitivity feature network to obtain the sensitive pathological feature component.

7. The fusion analysis method of remote diagnosis data according to claim 1, characterized in that: Generating a diagnosis result for the patient based on the pathological feature vector to obtain a patient health status assessment report includes: Performing time-series dynamic decomposition on the pathological feature vector to obtain a pathological state evolution sequence, and performing multi-dimensional clinical feature mapping on the pathological state evolution sequence to obtain a clinical symptom association matrix; Performing pathophysiological correlation analysis on the clinical symptom correlation matrix to obtain a system function damage degree vector, and performing multi-level health risk quantification on the system function damage degree vector to obtain a health risk assessment feature group; Performing multi-system collaborative diagnosis analysis based on the health risk assessment feature group to obtain a system collaborative diagnosis feature set, and predicting the pathological state evolution of the system collaborative diagnosis feature set to obtain a disease development trend map; Multi-dimensional clinical diagnostic features are extracted from the disease development trend map to obtain a clinical diagnostic feature vector, and a comprehensive health status assessment of the patient is performed based on the clinical diagnostic feature vector to obtain a health status assessment report of the patient.

8. A fusion analysis system for remote diagnostic data, characterized in that: include: A sending module, used for remotely receiving multi-source heterogeneous physiological signals sent by a patient through a medical data acquisition unit; An extraction module, configured to extract physiological features from the multi-source heterogeneous physiological signals to obtain a physiological feature sequence; A fusion module, configured to perform data fusion processing on the physiological feature sequence to obtain a fusion feature matrix; An analysis module, configured to perform pathological feature analysis on the patient based on the fusion feature matrix to obtain a pathological feature vector; A generating module is used to generate a diagnosis result for the patient based on the pathological feature vector to obtain a patient health status assessment report.

9. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images