Anea syndrome attack mode risk prediction method

Abstract

The invention discloses an apnea syndrome attack mode risk prediction method, which relates to the field of health management, and comprises the following steps: calling user state data of a whole age stage, mining a risk array based on state risk characteristics under different attack modes, and carrying out lightweight monitoring sampling and risk prediction training based on single characteristics and combined characteristics. The method comprises the steps of establishing a risk prediction model, collecting sleep monitoring data of a target user, executing risk prediction and attack mode judgment, determining a risk prediction result and giving a user state alarm, and is used for solving the problem of how to effectively extract and utilize state data of the user and carry out efficient and accurate health supervision in the prior art. The technical problem of providing early risk early warning in daily monitoring is solved, a personalized and low-power-consumption risk prediction and management scheme is provided through multi-feature mining and a lightweight monitoring model, and the efficiency and accuracy of risk prediction can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of health management, and in particular to a method for predicting the risk of sleep apnea syndrome attack patterns. Background Art

[0002] Sleep apnea syndrome is a common sleep breathing disorder, which is mainly characterized by repeated pauses in breathing or hypopnea during sleep, which may lead to decreased oxygen saturation, cardiovascular disease and other serious health problems. In the existing technology, analysis is mainly carried out through polysomnography and other means. Although it is accurate, the equipment is expensive and complicated to operate, and patients need to complete the examination in a medical institution, which is difficult to meet daily monitoring needs. In addition, portable sleep monitoring equipment has gradually become popular, but there are still deficiencies in data collection and risk prediction, especially in the lack of effective methods for fine-grained risk assessment of attack patterns.

[0003] At present, existing technologies often rely on a single feature or simple threshold to determine risks, which makes it difficult to fully capture the personalized risk characteristics of different users. In addition, the monitoring sampling and analysis process consumes a lot of resources, which limits the real-time and low-power requirements of the equipment. Therefore, how to effectively extract and utilize user status data to conduct efficient and accurate health supervision, thereby providing early risk warnings in daily monitoring, is a technical problem that needs to be solved urgently. Summary of the invention

[0004] The present application provides a method for predicting the risk of sleep apnea syndrome attack patterns, which is used to solve the technical problem existing in the prior art of how to effectively extract and utilize user status data to conduct efficient and accurate health supervision, thereby providing early risk warning in daily monitoring.

[0005] In view of the above problems, the present application provides a method for predicting the risk of sleep apnea syndrome attack patterns.

[0006] In the first aspect, the present application provides a method for predicting the risk of sleep apnea syndrome attack patterns, the method comprising: calling user status data of all age groups, mining risk matrixes based on status risk characteristics under different attack modes, wherein the status risk characteristics include common pattern features and specific pattern features, and defining the feature risk boundary by absolute contraction; based on the risk matrix, performing lightweight monitoring sampling and risk prediction training based on single features and combined features, and constructing a risk prediction model, wherein the risk prediction model is built into a sleep monitoring device; connecting the sleep monitoring device, combining the risk prediction model, determining the sleep monitoring data of the target user through staggered sampling, transmitting data back and performing risk prediction and attack pattern determination, and determining the risk prediction result; based on the risk prediction result, issuing a status alarm for the target user, and generating a time series monitoring table for the monitoring time zone.

[0007] In the second aspect, the present application provides a risk prediction system for sleep apnea syndrome attack pattern, the system comprising: a data mining module, the data mining module is used to call user status data of all age stages, and mine risk arrays based on status risk characteristics under different attack modes, wherein the status risk characteristics include pattern common characteristics and pattern specific characteristics, and the feature risk degree boundary is defined by absolute contraction; a training module, the training module is used to perform lightweight monitoring sampling and risk prediction training based on single features and combined features based on the risk array, and construct a risk prediction model, wherein the risk prediction model is built into a sleep monitoring device; a risk prediction module, the risk prediction module is used to connect to the sleep monitoring device, combine with the risk prediction model, determine the sleep monitoring data of the target user through staggered sampling, perform data feedback and execute risk prediction and attack pattern judgment, and determine the risk prediction result; a risk management module, the risk management module is used to issue a status alarm to the target user based on the risk prediction result, and generate a time series monitoring table for the monitoring time zone.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] The embodiment of the present application provides a method for predicting the risk of an apnea syndrome attack pattern, which calls user status data of all age groups, mines risk arrays based on status risk features under different attack modes, wherein the status risk features include common features of the pattern and specific features of the pattern, and define the feature risk boundary by absolute contraction; based on the risk array, lightweight monitoring sampling and risk prediction training based on single features and combined features are performed, and a risk prediction model is constructed and built into a sleep monitoring device; the sleep monitoring device is connected, and the sleep monitoring data of the target user is determined by staggered sampling in combination with the risk prediction model, and data is returned and risk prediction and attack pattern determination are performed to determine the risk prediction result; based on the risk prediction result, a status alarm of the target user is issued, and a time series monitoring table of the monitoring time zone is generated, so as to solve the technical problem of how to effectively extract and utilize the user's status data in the prior art, and perform efficient and accurate health supervision, thereby providing early risk warning in daily monitoring, and providing personalized, low-power risk prediction and management solutions through multi-feature mining and lightweight monitoring models, which can effectively improve the efficiency and accuracy of risk prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic flow chart of a method for predicting the risk of an apnea syndrome attack pattern is provided for this application;

[0011] Figure 2A schematic diagram of a process for determining a monitoring sampling method in a method for predicting the risk of an apnea syndrome attack pattern is provided for this application;

[0012] Figure 3 A schematic diagram of the structure of a sleep apnea syndrome attack pattern risk prediction system is provided for this application.

[0013] Explanation of the accompanying drawings: data mining module 11, training module 12, risk prediction module 13, risk management module 14. DETAILED DESCRIPTION

[0014] The present application provides a method for predicting the risk of sleep apnea syndrome attack patterns, calls user status data of all age groups, mines risk arrays based on status risk characteristics under different attack modes, conducts lightweight monitoring sampling and risk prediction training based on single features and combined features, builds a risk prediction model, collects sleep monitoring data of target users and performs risk prediction and attack pattern determination, determines risk prediction results and issues user status alarms, so as to solve the technical problem of how to effectively extract and utilize user status data in the prior art, conduct efficient and accurate health supervision, and thus provide early risk warnings in daily monitoring.

[0015] Embodiment 1: Figure 1 As shown, the present application provides a method for predicting the risk of sleep apnea syndrome attack pattern, the method comprising:

[0016] S1: Calling user status data of all age groups, mining risk matrix based on status risk characteristics under different attack modes, wherein the status risk characteristics include common characteristics of the mode and specific characteristics of the mode, and defining the characteristic risk boundary by absolute contraction.

[0017] In the embodiment of the present application, to call the user status data of all age stages, it is first necessary to obtain a sleep-related data set covering multi-dimensional characteristics such as different age groups, genders, and physiques. The data should include the user's ventilation, heart rate, respiratory rate, oxygen saturation, movement status and other basic status information to ensure that it is extensive and representative. By mining these data, it can be identified that the attack patterns of apnea syndrome include obstructive sleep apnea hypopnea syndrome, central sleep apnea syndrome, and mixed apnea syndrome. Different patterns differ in etiology and specific manifestations. In more detail, the state of the attack mode in different individuals and situations is also different, specifically including single attacks, periodic attacks, and persistent risks. In order to further refine the analysis, it is necessary to extract the state risk characteristics related to each attack mode.

[0018] The state risk characteristics can be divided into pattern-common characteristics and pattern-specific characteristics. Pattern-common characteristics refer to risk indicators that commonly appear in different attack patterns, such as the amplitude and frequency of oxygen saturation decrease. Pattern-specific characteristics are unique indicators in specific attack patterns, such as the changing trend of respiratory rhythm in a specific time period in periodic attacks. Through the detailed analysis of these characteristics, the inherent laws of the attack pattern can be described more accurately.

[0019] Furthermore, after feature extraction is completed, the above-mentioned state risk features are classified and quantified by mode by constructing a risk matrix. The risk matrix is ​​a multidimensional matrix structure, each dimension of which represents the characteristic risk degree under different attack modes, and is used to describe the risk level of a specific feature. In order to ensure the scientificity and consistency of the definition of risk, the absolute contraction method is introduced to determine the critical value of possible risk. These boundary values ​​not only represent the critical threshold of the risk feature, but also provide a quantitative standard for subsequent mode determination.

[0020] Through the above process, the calling and analysis of user status data of all age stages is completed, which not only lays the foundation for the characteristic identification of different attack patterns, but also provides multi-dimensional support for risk prediction through the establishment of risk arrays, providing a basis for precise risk management of sleep apnea syndrome.

[0021] Further, to mine the risk matrix of state risk characteristics under different attack modes, step S1 of the present application includes:

[0022] Acquire a first hypoventilation feature, wherein the first hypoventilation feature∈the state risk feature; traverse the attack pattern, perform feature mining and clustering based on the user state data, and determine N feature value groups based on the first hypoventilation feature, wherein the attack pattern corresponds one-to-one to the feature value group; based on the feature value group, perform an absolute maximum contraction within the group to determine N risk boundary values ​​of the first hypoventilation feature, wherein the first hypoventilation feature is an extremely small feature that is negatively correlated with the risk; add the N risk boundary values ​​into the risk matrix.

[0023] Among them, for extremely large features, the absolute minimum shrinkage is the absolute shrinkage method; for extremely small features, the absolute maximum shrinkage is the shrinkage method.

[0024] In an embodiment of the present application, the first hypoventilation feature is used to measure the degree of decrease in the user's ventilation volume within a specific time period, and is one of the features in the state risk feature. Hypoventilation feature data related to sleep apnea syndrome is screened from the user state data, and feature mining and clustering analysis are performed according to different attack modes, such as single attacks, periodic attacks, and complex mixed attacks. Specifically, through the statistical and clustering processing of the user state data, the user state data related to the first hypoventilation feature are grouped, and N feature value groups are integrated to determine. Each feature value group corresponds to an attack mode, and the feature values ​​in the group reflect the distribution law of the hypoventilation characteristics under this mode.

[0025] Since the first hypoventilation feature is an extremely small feature, that is, the smaller its value, the higher the risk, the method of absolute maximum shrinkage within the group is used to process the N eigenvalue groups, which can ensure that the boundary value accurately reflects the entire risk interval to determine the N risk boundary values ​​of the first hypoventilation feature. Exemplarily, by absolute maximum shrinkage, that is, shrinking toward the upper limit of the data interval, the boundary of the eigenvalue distribution is obtained. For example, when the value distribution of a eigenvalue group of the first hypoventilation feature is [50,80], after absolute maximum shrinkage, the risk boundary value is 70, that is, 70 is the critical value for possible risk. As long as it reaches 70, there may be a risk. The smaller the data, the greater the probability of risk. 70 is used as the standard for risk determination.

[0026] For extremely large features (i.e., the larger the eigenvalue, the higher the risk), the absolute minimum shrinkage method is adopted according to the determined eigenvalue group, that is, shrinking towards the lower limit of the data interval, so as to obtain the risk boundary of the eigenvalue distribution, that is, the critical value where risk may exist, as the risk boundary value.

[0027] For each state risk feature in the state risk features, according to the above steps, feature mining and clustering based on the user state data are performed respectively to determine the feature type, i.e., extremely large or extremely small, perform absolute shrinkage processing, and determine the corresponding risk boundary value.

[0028] Finally, the risk matrix is ​​generated by integrating the above risk boundary values, which is used as a quantitative risk indicator for the corresponding attack mode. The update and improvement of the risk matrix provides an accurate input basis for subsequent mode determination and risk prediction, and further improves the adaptability and operability of the risk assessment model.

[0029] Further, the N risk boundary values ​​are added into the risk matrix. Step S1 of the present application includes:

[0030] Based on the risk evolution trend of the first hypoventilation feature, a preset tolerance scale is determined, wherein the preset tolerance scale is positively correlated with the risk evolution speed; based on the preset tolerance scale, the N risk boundary values ​​are subjected to value expansion processing to determine N risk precursor characteristic values ​​of the first hypoventilation feature; a mapping between the attack pattern and the N risk precursor characteristic values ​​is established and added into the risk matrix.

[0031] In an embodiment of the present application, the risk evolution trend of the first hypoventilation feature, that is, the law of change over time under different attack modes, that is, how the risk level gradually evolves from low to high. This trend is usually achieved through time series analysis, reflecting the dynamic change rate and direction of the first hypoventilation feature in a specific attack mode. The specific manifestation of the risk evolution trend may include the characteristic value gradually decreasing to a critical point or a sharp drop in a certain time period. The preset tolerance scale refers to the tolerance range set to allow normal fluctuations or avoid misjudgment, and its size is positively correlated with the risk evolution speed. For example, for a mode with a faster risk evolution speed, the tolerance scale needs to be enlarged accordingly in order to capture potential risks in advance.

[0032] Based on the above-mentioned preset tolerance scale, the N risk boundary values ​​are subjected to value expansion processing to determine the N risk precursor characteristic values ​​of the first hypopnea feature. Value expansion processing refers to the use of incremental expansion based on the original boundary value to describe possible early risks. For example, when the risk boundary value of the first hypopnea feature is 70, if the preset tolerance scale is 5, the risk boundary value after the expansion processing is 75, which can effectively enhance the sensitivity to risk precursors, especially in practical applications to avoid missing early risks.

[0033] Similarly, if it is an extremely small feature, it is necessary to expand in a decreasing direction, such as adjusting 70 to 65, as the risk precursor feature value.

[0034] After determining the above N risk precursor characteristic values, a mapping relationship between the attack mode and its corresponding risk precursor characteristic value is further established. This mapping relationship is a one-to-one logical binding, which clearly indicates which precursor characteristic values ​​may indicate the risk of attack in a specific mode. For example, for a periodic attack mode, the range of its risk precursor characteristic value may show a periodic distribution, which is significantly different from the random distribution characteristics of other modes.

[0035] Finally, the above mapping relationship and risk precursor feature values ​​are added to the risk matrix, so that the risk matrix not only contains the risk boundary value, but also has the ability to predict precursor risks, so that it can more comprehensively support the early identification and pattern determination of sleep apnea syndrome.

[0036] S2: Based on the risk matrix, lightweight monitoring sampling and risk prediction training based on single features and combined features are performed to construct a risk prediction model, wherein the risk prediction model is built into the sleep monitoring device.

[0037] First, by analyzing the risk characteristic values ​​and precursor characteristic values ​​of different attack modes in the risk array, and by performing feature grouping and staggered sampling configuration, after determining the sampling method, lightweight monitoring sampling and training are performed. Lightweight means that in a resource-constrained hardware environment, by optimizing the sampling frequency, data volume and computational complexity, low-power and high-efficiency sampling and risk prediction processing training are achieved. Specifically, grouped staggered sampling is adopted, and based on the sample training method, precise control training of feature sampling triggering is performed to construct the first sampling unit.

[0038] Furthermore, based on the single feature determination-attack mode determination-combined feature determination as the benchmark, the risk degree boundary matching determination based on the risk matrix is ​​used as the single feature determination method, and the attack mode determination method is the matching of pattern-specific features. By calling the training samples, that is, the risk analysis records determined by big data retrieval, the sample supervised training method is trained until convergence. If the preset output accuracy is met, the second decision unit is obtained. The sensitivity and reliability of risk decision-making are significantly improved through the multi-level judgment mechanism.

[0039] Then, the second decision unit is post-connected to the first sampling unit to generate the risk prediction model, which is built into the sleep monitoring device, so that the device has real-time data collection and risk prediction capabilities. Through the low-latency operation of the built-in model, users can obtain personalized risk warnings and pattern analysis results without relying on external devices or complex data transmission, providing convenient support for daily health management. At the same time, the construction and deployment of the model not only achieves efficient risk prediction, but also strikes a balance between cost, power consumption and ease of use.

[0040] Further, a risk prediction model is constructed, and step S2 of this application includes:

[0041] Determine the monitoring sampling method, conduct sensor sampling training, and construct a first sampling unit; determine the risk boundary value of a single detection feature and make a collaborative determination with the detection feature group to conduct sample-driven training and construct a second decision unit, wherein the attack mode is determined based on the specific feature; combine the first sampling unit and the second decision unit to generate the risk prediction model.

[0042] In the embodiment of the present application, a specific strategy for sensor sampling is formulated by analyzing the characteristic distribution and sampling requirements in the user status data. The core of the monitoring sampling method is to optimize the sampling frequency and sensor configuration so as to achieve comprehensive capture of risk characteristics with minimum data volume and power consumption. According to the monitoring sampling method, sensor sampling training is performed to form a functional module that focuses on the processing and storage of sampled data, namely the first sampling unit.

[0043] After completing the construction of the first sampling unit, further risk judgment training based on single detection features and detection feature groups is carried out to construct the second decision unit. The risk boundary value judgment of a single detection feature refers to comparing the single feature value obtained by sensor sampling with the boundary value in the risk array one by one to determine whether there is a potential risk. For example, when the hypoventilation feature value is less than its boundary value (such as the critical point 75), the risk flag is triggered immediately. Since there is a certain correlation between the features, the collaborative judgment of the detection feature group analyzes the relationship between multiple features and mines the risk information caused by the collaboration of multiple features in this mode. For example, if the respiratory rate decreases and the blood oxygen saturation decreases rapidly, the collaborative feature points to a higher risk level. Through the sample-driven training method, a large amount of historical data is learned and optimized to ensure that the second decision unit can accurately judge the risk situation of single features and combined features.

[0044] Among them, the second decision unit uses pattern-specific characteristics as key indicators, that is, the difference between different patterns is used as the judgment standard to further realize the accurate judgment of the attack pattern. For example, by identifying the repetitiveness of periodic characteristics or the rapid change trend of sudden characteristics, it is judged which attack mode (such as single attack or mixed attack) the current attack belongs to. This pattern recognition mechanism based on specific characteristics determines the corresponding pattern and conducts analysis under the attack mode, which can help improve the pertinence and practicality of risk prediction.

[0045] Finally, the first sampling unit and the second decision unit are combined to generate a complete risk prediction model. The first sampling unit is responsible for efficient data collection to ensure the integrity and timeliness of the input; the second decision unit achieves high-precision risk assessment and attack pattern determination through intelligent analysis and pattern matching. The organic combination of the two forms a closed-loop monitoring and prediction system that can process the user's sleep data in real time and output accurate risk prediction results, providing reliable technical support for the prevention and control of sleep apnea syndrome.

[0046] Further, such as Figure 2 As shown, the step S2 of determining the monitoring sampling method includes:

[0047] Based on feature correlation, the state risk features are combined to determine M detection feature groups; the M detection feature groups are serialized to determine a risk detection sequence, wherein each detection feature group corresponds to a sequence node; a preset sampling interval is set, and based on the risk detection sequence, a staggered sampling configuration of the state risk features is performed to determine a monitoring sampling method.

[0048] In the embodiment of the present application, based on the feature correlation, the state risk features are combined, and according to the potential correlation and complementarity between different features, the accuracy of risk determination can be effectively improved, and the errors existing in single feature analysis can be avoided. At the same time, by grouping, sampling and monitoring of a feature group is performed at a time, and staggered monitoring and sampling are performed between different feature groups. On the basis of ensuring the accuracy of analysis, a large amount of data analysis of full feature monitoring sampling analysis can be avoided, which can effectively balance the accuracy and efficiency of risk decision-making.

[0049] Specifically, the state risk characteristics refer to key indicators that affect the onset of sleep apnea syndrome, such as hypoventilation characteristics, blood oxygen saturation, respiratory rate, etc. These characteristics can be combined to reveal more complex risk patterns. For example, changes in hypoventilation characteristics and respiratory rate may jointly indicate the occurrence of a risk state in a certain attack mode. Through feature correlation, identify which features have a strong correlation, and combine these highly correlated features into a group to form M detection feature groups. The composition of the detection feature group will be based on the complementarity and correlation between the features, which can further improve the accuracy of risk judgment. For example, hypoventilation and blood oxygen saturation may form a group, while respiratory rate and pulse, heartbeat may form another group, and each detection feature group corresponds to at least two state risk characteristics.

[0050] Next, the M detection feature groups are serialized, and exemplarily, a time series is constructed, and each sequence node corresponds to one of the M detection feature groups, wherein the detection feature group matching based on the time series is randomly assigned, so that each detection feature group can be monitored and analyzed in a certain order. The serialization process arranges the M feature groups in chronological order or according to the priority of their risk assessment to ensure that key risk changes can be captured in time during the monitoring process. Preferably, high-frequency monitoring of strongly correlated features can be performed according to the sleep stage, while appropriately reducing the monitoring frequency of non-strongly correlated features. For example, monitoring of hypoventilation and blood oxygen saturation may need to be performed in the early and middle stages of sleep, while monitoring of respiratory rate may focus on the deep stage of sleep. Through this serialization process, a risk detection sequence is formed, in which each detection feature group corresponds to a specific sequence node, which means that in the actual monitoring process, these feature groups are sampled and analyzed in sequence according to the preset order.

[0051] In order to ensure the rationality of data collection and monitoring efficiency, a preset sampling interval is set, where the preset sampling interval is a sampling interval time interval set by those skilled in the art, for example, three minutes.

[0052] Furthermore, based on the risk detection sequence and the preset sampling interval, staggered sampling configuration of the state risk features is performed. Specifically, according to the risk detection sequence, after the monitoring sampling of the detection feature group of the previous sequence node is performed, the monitoring sampling of the monitoring feature group of the next sequence node is performed based on the preset sampling interval. By performing the sampling of a group of monitoring features at a time and sampling the state risk features in batches, the amount of data processed at a single time can be reduced on the basis of ensuring accuracy, and the accuracy and efficiency of risk decision-making can be reasonably balanced.

[0053] Preferably, the sampling frequency can be appropriately adjusted according to the characteristics of the sleep state. For example, in the deep sleep stage, the change of the hypopnea characteristic may be more obvious, so the sampling frequency of the hypopnea characteristic can be increased during this period; while in the light sleep stage, the sampling frequency can be appropriately reduced to reduce power consumption and reduce interference.

[0054] By grouping features based on feature correlation and configuring staggered sampling of features, a sampling strategy is formed that can efficiently capture risk changes while minimizing energy consumption and data redundancy.

[0055] S3: Connecting the sleep monitoring device, combining the risk prediction model, determining the sleep monitoring data of the target user through staggered sampling, transmitting the data back and performing risk prediction and seizure mode determination, and determining the risk prediction result.

[0056] The sleep monitoring device usually includes a variety of sensors, such as a breathing sensor, a blood oxygen sensor, a heart rate monitoring sensor, etc., for real-time collection of the user's physiological data. By connecting to the sleep monitoring device, the user's sleep state can be monitored in real time for subsequent analysis and processing.

[0057] Based on the first sampling unit and according to the monitoring sampling method, staggered sampling based on the risk detection sequence is performed, that is, sensor collection is performed based on the detection feature group corresponding to the first sequence node. When the preset sampling interval is met, sensor collection is performed based on the detection feature group corresponding to the second sequence node, and the data of a single sampling is used as the sleep monitoring data of the target user.

[0058] The sleep monitoring data is further transmitted to the second decision unit of the risk prediction model to provide real-time input for risk prediction. For example, the device records the user's breathing rate, blood oxygen saturation, heart rate and other data in real time through sensors, and transmits it back to the risk prediction model. At this time, the algorithm built into the model processes the received data, combines the existing characteristic risk boundaries and pattern recognition rules, and performs risk prediction and seizure pattern determination.

[0059] Specifically, risk prediction and attack mode judgment are performed for single feature judgment and feature combination judgment respectively. Specifically, when a certain feature (such as hypoventilation) crosses the preset risk boundary value, the model will identify the potential risk of apnea and determine whether it belongs to a specific attack mode. Further feature combination judgment is performed under this attack mode. For example, when the respiratory rate drops sharply and the blood oxygen saturation drops, the model may determine it as a severe apnea attack. In this case, the risk prediction result will be a high-risk warning and prompt relevant countermeasures. The accuracy of risk analysis can be effectively guaranteed by performing feature complementary judgment.

[0060] S4: Based on the risk prediction result, a status alarm of the target user is issued, and a time series monitoring table of the monitoring time zone is generated.

[0061] Specifically, based on the risk prediction results, the target user's status alarm is issued to ensure that the user can get a timely response when a potential health risk is discovered. For example, if the risk value exceeds the preset safety limit or a possible severe attack mode is predicted (such as worsening apnea, hypoxemia, etc.), the status alarm mechanism is triggered. Status alarms can be notified in a variety of ways, such as sleep monitoring equipment alarms, terminal display warnings, etc. The content of the alarm includes the user's current risk status and possible health impacts. It can help users take appropriate intervention measures as soon as possible to reduce health risks caused by delayed processing.

[0062] The time series monitoring table of the monitoring time zone is a detailed risk change situation and corresponding health status of each time period during the user's sleep. The main purpose of this process is to show the health dynamics of the user in each sleep period through the time series of data. For example, in the table, the system will combine the monitoring data in different time periods with the risk prediction results according to the time axis to generate a clear time series monitoring table. Each monitoring time zone (such as sleep start, light sleep, deep sleep, etc.) will record the corresponding physiological data and risk prediction results, such as changes in hypoventilation characteristics, blood oxygen saturation, respiratory rate and other indicators. The time series monitoring table can show in detail whether there are any characteristics that have crossed the set risk boundary in each time period, or whether there are signs of an attack mode.

[0063] In addition, the time series monitoring table can trace back the data trajectory to understand the health change trends at different stages. The combination of status alarm and time series monitoring table not only improves the accuracy of individual health management, but also effectively promotes disease prevention and early intervention, providing users with scientific and accurate health management services.

[0064] Furthermore, the execution risk prediction and seizure mode determination, step S3 of the present application includes:

[0065] The sleep monitoring data is transmitted back to the second decision unit of the risk prediction model, and the risk array is traversed to perform matching, risk boundary value judgment and trend prediction to determine the first judgment data; the first judgment data is traversed to identify boundary crossing features. If the boundary crossing feature is empty, the risk decision processing is terminated. If the boundary crossing feature is not empty, the attack pattern matching and risk judgment are performed.

[0066] In the embodiment of the present application, the sleep monitoring data is transmitted back to the second decision unit of the risk prediction model, and the second decision unit is used to further analyze and make decisions on the input monitoring data. The risk feature boundary value is matched to help determine the risk status of the user.

[0067] Specifically, the risk matrix contains multi-dimensional risk feature data, where each risk feature has a corresponding risk boundary value. The sleep monitoring data collected in real time is compared based on the risk boundary value. The sleep monitoring data, that is, the sampling value of each feature in the detection feature group, is matched based on the risk matrix, and the corresponding risk boundary value is determined based on the matching result, and a single feature over-limit judgment and trend prediction are performed as the first judgment data.

[0068] For example, if a certain monitoring data (such as respiratory rate) exceeds the preset risk boundary (for example, the respiratory rate is lower than the threshold), it will be marked as a potential risk. In addition to the determination of the risk boundary, the trend of data changes is also predicted. If the data shows a trend of continuous change (such as a gradual decrease in oxygen saturation), its possible development direction (such as possible apnea) is further inferred and a trend prediction result is generated.

[0069] The first judgment data is further traversed to identify boundary crossing features, that is, whether any feature has crossed its corresponding risk boundary. The boundary crossing feature means that when a certain detection feature, that is, a physiological parameter (such as hypoventilation), exceeds the set risk boundary, this feature is marked as a boundary crossing feature. If the boundary crossing feature is empty, it means that all monitoring data are within the safety range. At this time, the system will stop the subsequent risk decision processing, indicating that the user has no immediate health risk at the current moment.

[0070] If the boundary crossing feature is not empty, it means that at least one physiological data point exceeds the preset safety boundary, indicating that there is a potential health risk. At this time, further risk determination is performed under attack pattern matching and feature combination.

[0071] Through the above steps, the accuracy of risk detection can be maximized with minimal data processing, potential health risks can be accurately identified, and users can take timely actions to effectively prevent the occurrence of serious health problems.

[0072] Further, the seizure pattern matching and risk determination are performed, and step S3 of the present application includes:

[0073] If the boundary crossing feature is not empty, the target seizure pattern is determined by matching with the pattern-specific feature; for the target seizure pattern, feature combination judgment and trend prediction are performed to determine the second judgment data; the first judgment data and the second judgment data are fused as the risk prediction result.

[0074] Specifically, if the boundary crossing feature is not empty, it means that during the sleep monitoring process, certain key physiological parameters (such as blood oxygen saturation, respiratory rate, etc.) have exceeded the preset safety threshold, suggesting potential health risks. In this case, the system will determine the target attack mode by matching with the mode-specific feature. The mode-specific feature refers to the distinguishing features under different attack modes.

[0075] By matching the boundary crossing feature with the pattern-specific feature, the current seizure pattern is determined based on the matching result as the target seizure pattern.

[0076] Once the target attack mode is determined, feature combination judgment and trend prediction are performed for the mode. Feature combination judgment means that in the target attack mode, in coordination with the current detection feature group, multiple related physiological characteristics (such as heart rate fluctuations and decreased blood oxygen during apnea) are comprehensively considered to further analyze health risks. For example, if in the apnea attack mode, the system detects that the heart rate gradually increases and the blood oxygen saturation gradually decreases, this may mean that the condition is worsening. Trend prediction is to predict the changing trends of these physiological characteristics in the future (such as a continuous decrease in blood oxygen or aggravated apnea) based on comprehensive multi-dimensional data in order to evaluate the dynamic changes in risks.

[0077] The second judgment data is determined through feature combination judgment and trend prediction. The second judgment data not only reflects the risk situation in the current state, but also takes into account possible health change trends. For example, if it is predicted that the user's blood oxygen will drop further in the next few minutes, a risk warning will be issued in advance based on this trend to ensure timely intervention measures.

[0078] Finally, the first judgment data and the second judgment data are fused as the risk prediction result. The first judgment data comes from the preliminary risk assessment, including the determination of the boundary value of a single feature and the preliminary judgment of whether it exceeds the limit; while the second judgment data provides a more accurate and dynamic risk prediction through a deeper feature combination analysis and trend prediction. After combining the two, the system can generate a comprehensive risk prediction result that accurately reflects the current health status of the target user and predicts possible risks in the future. This result will serve as the basis for subsequent alarms and intervention decisions to ensure that users can receive corresponding health management in a timely manner.

[0079] The present application provides a method for predicting the risk of sleep apnea syndrome attack patterns, which has the following technical effects:

[0080] 1. By combining different attack modes, mining risk arrays based on state risk characteristics, dividing common characteristics and pattern-specific characteristics, and defining the characteristic risk boundary by absolute contraction, it can comprehensively cover the health risks of users of different age groups, fully consider the diversity of attack modes, and improve the accuracy and breadth of risk prediction.

[0081] 2. Taking hypopnea characteristics as an example, determine the mining process of state risk characteristics. Accurately defining the risk boundary of the characteristics can effectively identify risks related to health problems such as apnea and improve the sensitivity of risk determination. By setting a preset tolerance scale, performing value expansion processing, and determining the risk precursor characteristic value, timely warning can be given before the onset of the attack, enhancing the early warning capability and buying time for intervention measures.

[0082] 3. By dividing feature groups and performing staggered sampling configuration, the amount of data processed and the accuracy of analysis can be effectively balanced, lightweight configuration can be achieved, and the accuracy and reliability of prediction can be enhanced by combining feature risk boundary determination and feature group collaborative determination.

[0083] In summary, it can provide comprehensive, real-time and high-precision risk prediction of sleep apnea syndrome attack patterns, significantly improve early warning capabilities and intervention efficiency, and prevent the occurrence of health risks.

[0084] Embodiment 2: Based on the same inventive concept as the method for predicting the risk of apnea syndrome attack pattern in the above embodiment, Figure 3 As shown, the present application provides a risk prediction system for sleep apnea syndrome attack pattern, the system comprising:

[0085] A data mining module 11, which is used to call user status data of all age stages, and mine risk matrixes based on status risk characteristics under different attack modes, wherein the status risk characteristics include common characteristics of the mode and specific characteristics of the mode, and define the characteristic risk degree boundary by absolute contraction;

[0086] A training module 12, wherein the training module 12 is used to perform lightweight monitoring sampling and risk prediction training based on single features and combined features based on the risk matrix to construct a risk prediction model, wherein the risk prediction model is built into the sleep monitoring device;

[0087] The risk prediction module 13 is used to connect to the sleep monitoring device, combine with the risk prediction model, determine the sleep monitoring data of the target user through staggered sampling, transmit the data back and perform risk prediction and seizure mode determination, and determine the risk prediction result;

[0088] The risk management module 14 is used to issue a status alarm to the target user based on the risk prediction result and generate a time series monitoring table for the monitoring time zone.

[0089] Among them, the data mining module 11 is also used to perform the following steps: obtain a first hypoventilation feature, wherein the first hypoventilation feature∈the state risk feature; traverse the attack pattern, perform feature mining and clustering based on the user state data, and determine N feature value groups based on the first hypoventilation feature, wherein the attack pattern corresponds one-to-one to the feature value group; based on the feature value group, perform absolute maximum contraction within the group to determine N risk boundary values ​​of the first hypoventilation feature, wherein the first hypoventilation feature is an extremely small feature that is negatively correlated with the risk; add the N risk boundary values ​​to the risk matrix.

[0090] Among them, for extremely large features, the absolute minimum shrinkage is the absolute shrinkage method; for extremely small features, the absolute maximum shrinkage is the shrinkage method.

[0091] Among them, the data mining module 11 is also used to perform the following steps: based on the risk evolution trend of the first hypoventilation feature, determine a preset tolerance scale, wherein the preset tolerance scale is positively correlated with the risk evolution speed; based on the preset tolerance scale, perform value expansion processing on the N risk boundary values ​​to determine the N risk precursor characteristic values ​​of the first hypoventilation feature; establish a mapping between the attack pattern and the N risk precursor characteristic values, and add them into the risk matrix.

[0092] Among them, the training module 12 is also used to perform the following steps: determine the monitoring sampling method, perform sensor sampling training, and construct a first sampling unit; use the risk boundary value judgment of a single detection feature and the collaborative judgment of the detection feature group to perform sample-driven training and construct a second decision unit, wherein the attack mode is judged based on the specific feature; combine the first sampling unit and the second decision unit to generate the risk prediction model.

[0093] Among them, the training module 12 is also used to perform the following steps: based on feature correlation, the state risk features are combined to determine M detection feature groups; the M detection feature groups are serialized to determine a risk detection sequence, wherein each detection feature group corresponds to a sequence node; a preset sampling interval is set, and based on the risk detection sequence, a staggered sampling configuration of the state risk features is performed to determine a monitoring sampling method.

[0094] Among them, the risk prediction module 13 is also used to perform the following steps: transmitting the sleep monitoring data back to the second decision unit of the risk prediction model, traversing the risk array for matching and risk boundary value judgment and trend prediction, and determining the first judgment data; traversing the first judgment data, identifying boundary crossing features, if the boundary crossing feature is empty, terminating the risk decision processing, if the boundary crossing feature is not empty, performing seizure pattern matching and risk judgment.

[0095] Among them, the risk prediction module 13 is also used to perform the following steps: if the boundary crossing feature is not empty, determine the target seizure pattern by matching with the pattern-specific feature; perform feature combination judgment and trend prediction on the target seizure pattern to determine the second judgment data; and fuse the first judgment data and the second judgment data as the risk prediction result.

[0096] Through the above-mentioned detailed description of a method for predicting the risk of an apnea syndrome attack pattern in this specification, those skilled in the art can clearly know the method for predicting the risk of an apnea syndrome attack pattern in this embodiment. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0097] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting the risk of sleep apnea syndrome attack pattern, characterized in that: The method comprises: Calling user status data of all age groups, mining risk matrix based on status risk characteristics under different attack modes, wherein the status risk characteristics include common characteristics of the mode and specific characteristics of the mode, and defining the characteristic risk boundary by absolute contraction; Based on the risk matrix, lightweight monitoring sampling and risk prediction training based on single features and combined features are performed to build a risk prediction model, wherein the risk prediction model is built into the sleep monitoring device; Connecting to a sleep monitoring device, combining the risk prediction model, determining the sleep monitoring data of the target user through staggered sampling, transmitting the data back and performing risk prediction and seizure mode determination, and determining the risk prediction result; Based on the risk prediction result, a status alarm of the target user is issued, and a time series monitoring table of the monitoring time zone is generated.

2. A method for predicting the risk of sleep apnea syndrome attack pattern according to claim 1, characterized in that: The risk matrix for mining the state risk characteristics under different attack modes includes: Acquire a first hypopnea feature, wherein the first hypopnea feature∈the state risk feature; Traversing the attack patterns, performing feature mining and clustering based on the user status data, and determining N feature value groups based on the first hypopnea feature, wherein the attack patterns correspond to the feature value groups one by one; Based on the characteristic value group, performing an absolute maximum contraction within the group to determine N risk boundary values ​​of the first hypopnea feature, wherein the first hypopnea feature is an extremely small feature that is negatively correlated with the risk; The N risk boundary values ​​are added into the risk matrix.

3. A method for predicting the risk of sleep apnea syndrome attack pattern according to claim 2, characterized in that: For extremely large features, the absolute minimum shrinkage is used as the absolute shrinkage method; for extremely small features, the absolute maximum shrinkage is used as the shrinkage method.

4. A method for predicting the risk of sleep apnea syndrome attack pattern according to claim 2, characterized in that: Adding the N risk boundary values ​​into the risk matrix includes: Determining a preset tolerance scale based on the risk evolution trend of the first hypopnea feature, wherein the preset tolerance scale is positively correlated with the risk evolution speed; Based on the preset tolerance scale, the N risk boundary values ​​are subjected to value expansion processing to determine N risk precursor characteristic values ​​of the first hypopnea characteristic; A mapping between the attack pattern and the N risk precursor feature values ​​is established and added into the risk matrix.

5. The method for predicting the risk of sleep apnea syndrome attack pattern according to claim 1, characterized in that: Construct a risk prediction model, including: Determine the monitoring sampling method, conduct sensor sampling training, and build the first sampling unit; The risk boundary value of a single detection feature is judged, and the detection feature group is judged in coordination, and sample-driven training is performed to construct a second decision unit, wherein the seizure mode is judged by the specific feature; The first sampling unit and the second decision unit are combined to generate the risk prediction model.

6. A method for predicting the risk of sleep apnea syndrome attack pattern according to claim 5, characterized in that: Determining the monitoring sampling method includes: Based on feature correlation, the state risk features are combined to determine M detection feature groups; Sequencing the M detection feature groups to determine a risk detection sequence, wherein each detection feature group corresponds to a sequence node; A preset sampling interval is set, and based on the risk detection sequence, staggered sampling configuration of the state risk characteristics is performed to determine the monitoring sampling method.

7. A method for predicting the risk of sleep apnea syndrome attack pattern according to claim 1, characterized in that: The execution risk prediction and attack mode determination include: The sleep monitoring data is transmitted back to the second decision unit of the risk prediction model, the risk matrix is ​​traversed to perform matching and risk boundary value determination and trend prediction, and the first determination data is determined; The first determination data is traversed to identify boundary crossing features. If the boundary crossing feature is empty, the risk decision process is terminated. If the boundary crossing feature is not empty, the attack pattern matching and risk determination are performed.

8. A method for predicting the risk of sleep apnea syndrome attack pattern according to claim 7, characterized in that: Perform seizure pattern matching and risk assessment, including: If the boundary crossing feature is not empty, determining the target seizure mode by matching with the mode-specific feature; For the target attack pattern, perform feature combination determination and trend prediction to determine second determination data; The first determination data and the second determination data are integrated as the risk prediction result.

9. A sleep apnea syndrome attack pattern risk prediction system, characterized in that: The system is used to execute the method for predicting the risk of apnea syndrome attack pattern according to any one of claims 1 to 8, and the system comprises: A data mining module, which is used to call user status data of all age groups and mine risk matrixes based on status risk characteristics under different attack modes, wherein the status risk characteristics include common characteristics of the mode and specific characteristics of the mode, and define the characteristic risk degree boundary by absolute contraction; A training module, wherein the training module is used to perform lightweight monitoring sampling and risk prediction training based on single features and combined features based on the risk matrix to construct a risk prediction model, wherein the risk prediction model is built into the sleep monitoring device; A risk prediction module, which is used to connect to the sleep monitoring device, combine with the risk prediction model, determine the sleep monitoring data of the target user through staggered sampling, transmit the data back and perform risk prediction and seizure mode determination, and determine the risk prediction result; A risk management module is used to issue a status alarm to the target user based on the risk prediction result and generate a time series monitoring table for the monitoring time zone.

Images