Big data driven emotional fluctuation recognition and determination system for blood pressure of the elderly

Abstract

The application discloses a big data driven old people blood pressure emotional fluctuation recognition and determination system, relates to the technical field of medical health monitoring, and comprises the following steps: detecting environmental stimulation and recording trigger nodes through an environment monitoring module, acquiring blood pressure data and physiological stress data of patients after trigger nodes through a data acquisition module; a lag feature extraction module determines a first delay from environmental stimulation to physiological stress reaction and a second delay from physiological stress reaction to blood pressure peak; a probability calculation module matches the two delays with an old people group reference delay distribution to obtain a matching degree; an association score module calculates a cause-effect association score based on the matching degree; and a determination module determines whether blood pressure fluctuation is emotional or non-emotional according to the relationship between the score and a threshold value and in combination with blood pressure fall values after a preset time length.

Description

Technical Field

[0001] This invention relates to the field of medical and health monitoring technology, and in particular to a big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly. Background Technology

[0002] The elderly are a high-risk group for cardiovascular diseases. Compared with ordinary adults, the elderly have decreased vascular elasticity, weakened autonomic nervous system regulation, and reduced tolerance to environmental stimuli, making their vital signs more susceptible to fluctuations caused by non-disease factors. In addition, the elderly recover from psychological stress much more slowly and are more prone to anxiety due to unfamiliar environments or medical procedures. These age-related physiological and psychological characteristics make it difficult for traditional vital sign monitoring methods to directly reflect the true cardiovascular health status of the elderly.

[0003] In hospitals, this difference is particularly pronounced. For ordinary adults, the tension caused by hospital examinations is usually short-lived and limited in magnitude, and vital signs can recover to near baseline levels in a short time. However, due to the slowed autonomic nervous system regulation, the tension response of the elderly is not only more easily triggered, but also lasts longer, resulting in drastic but not pathological short-term fluctuations in data such as blood pressure and heart rate. Therefore, even if the elderly do not actually have a worsening of cardiovascular disease, the vital signs data recorded by monitoring equipment may show a trend similar to the worsening of the disease.

[0004] Existing big data cardiovascular disease prediction systems typically assess risk based on patterns of vital sign fluctuations, such as the rate of blood pressure rise, the magnitude of heart rate changes, or abnormal morphologies of electrocardiogram signals. However, most of these systems assume that the monitoring data can objectively reflect the physiological state during model building, without modeling the "stress-induced fluctuations" unique to the elderly population, nor considering that the hospital setting amplifies this characteristic in the elderly. This can easily lead to an inability to distinguish between short-term fluctuations caused by emotions such as tension and anxiety and non-emotional blood pressure fluctuations, thus affecting the accuracy of the prediction results.

[0005] Therefore, a big data analysis-based system for predicting cardiovascular diseases in the elderly is proposed. Summary of the Invention

[0006] In view of the above-mentioned prior art, this application is hereby made. Embodiments of this application provide a big data-driven system for identifying and determining emotional fluctuations in blood pressure in the elderly, which can distinguish between emotional and non-emotional blood pressure fluctuations in the elderly, providing higher interpretability and reliability for cardiovascular disease prediction results.

[0007] According to one aspect of this application, a big data-driven system for identifying and determining blood pressure-emotional fluctuations in the elderly is provided, comprising: an environmental monitoring module configured to continuously detect environmental stimuli and record the occurrence time of the environmental stimuli as trigger nodes; a data acquisition module configured to acquire blood pressure data and physiological stress data of a target patient after the trigger nodes; a lag feature extraction module configured to determine a first delay and a second delay based on the trigger nodes, the blood pressure data, and the physiological stress data, wherein the first delay is the delay from the environmental stimulus to the physiological stress response, and the second delay is the delay from the physiological stress response to the blood pressure peak; and a probability calculation module configured to respectively compare the first delay and the second delay with data based on elderly population statistics. The system matches the corresponding reference delay distribution to obtain a first matching degree and a second matching degree. The correlation scoring module is configured to calculate a causal correlation score representing the strength of the correlation between blood pressure fluctuations and emotions based on the first and second matching degrees. The judgment module is configured to: determine blood pressure fluctuations as emotional fluctuations when the causal correlation score is greater than a first threshold; determine blood pressure fluctuations as non-emotional fluctuations when the causal correlation score is less than a second threshold; and execute the following steps when the causal correlation score is between the first and second thresholds: reacquire blood pressure data after waiting for a preset time and calculate the drop value of the blood pressure data; determine blood pressure fluctuations as emotional fluctuations when the drop value is greater than a preset drop threshold, otherwise determine them as non-emotional fluctuations.

[0008] Compared with existing technologies, the big data-driven blood pressure emotional fluctuation identification and judgment system for the elderly according to the embodiments of this application can distinguish between emotional and non-emotional blood pressure fluctuations in the elderly, providing higher interpretability and reliability for cardiovascular disease prediction results. Attached Figure Description

[0009] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 This is a system architecture diagram of the big data-driven system for recognizing and judging blood pressure and emotional fluctuations in the elderly, based on the present invention. Detailed Implementation

[0011] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0012] Exemplary system:

[0013] Figure 1 The illustration shows a big data-driven system for identifying and judging blood pressure and emotional fluctuations in the elderly according to an embodiment of this application, including an environmental monitoring module, a data acquisition module, a lag feature extraction module, a probability calculation module, a correlation scoring module, and a judgment module.

[0014] like Figure 1 As shown, the environmental monitoring module is configured to continuously detect environmental stimuli and record the time of occurrence of the environmental stimuli as a trigger node.

[0015] Traditional methods often overlook the impact of external environmental factors on blood pressure in the elderly. The hospital environment contains numerous stimuli that can trigger stress responses, such as sudden calls for service, medical equipment alarms, and the sudden approach of medical staff. These stimuli can cause a temporary increase in blood pressure through the physiological pathway of "stimulus perception → autonomic nervous system activation → cardiovascular system response." By capturing the timing of these environmental stimuli, a crucial time baseline can be provided for subsequent causal association analysis.

[0016] Among these, environmental stimuli include sound events and interactive events.

[0017] Sound events include at least one of the following: queuing call events, medical equipment alarm sounds, and environmental noise intensity exceeding a preset noise threshold. Queuing call events are the most common stimulus in hospital waiting environments; patients typically experience immediate psychological stress upon hearing their numbers called. Medical equipment alarm sounds are high-frequency and sharp, such as the beeping of an electrocardiogram monitor, which strongly stimulates the auditory system of the elderly. Sudden increases in environmental noise intensity, such as noise in the corridor or the movement of equipment, can also trigger stress responses. The preset noise threshold is typically set at a 15 dB increase relative to the baseline noise level. This threshold is based on acoustic physiology research and can identify sound events that pose a real stimulus to the elderly.

[0018] An interactive event is defined as the entry of medical personnel into a region centered on the target patient and within a predetermined radius. The approach of medical personnel triggers heightened alertness and social stress in patients, particularly when the personnel are carrying medical equipment or wearing white coats; this "white coat effect" is more pronounced. The predetermined distance is typically set at 1.5-2 meters, based on the theory of "personal space" in interpersonal communication, where the entry of others into this distance triggers a physiological alert response.

[0019] Specifically, the environmental monitoring module includes a sound detection unit and an image detection unit.

[0020] The sound detection unit is configured to detect sound events. In implementation, the sound detection unit can use a room-level microphone, placed in a corner of the waiting room or examination room, to collect acoustic signals in real time. It should be noted that, to protect patient privacy, the system only extracts acoustic features, such as sound intensity, frequency band energy distribution, and temporal envelope, and does not save the speech content. Sound event detection is based on the following technologies:

[0021] For queuing events, Voice Activity Detection (VAD) combined with keyword recognition technology is used. When a voice segment matching the queuing is detected, it is recorded as a queuing event. For medical device alarm sounds, specific frequency characteristics are identified through spectrum analysis. For environmental noise events, short-time sound intensity within a 100ms window is calculated in real time. When the sound intensity increases more than the baseline value of the past 30 seconds, it is triggered to record.

[0022] The image detection unit is configured to acquire video streams associated with the target patient and detect interactive events based on video analysis. In implementation, the image detection unit can use a non-contact camera and be implemented using computer vision technology. The specific detection process is as follows: First, the target patient's position in the video frame is located using a target tracking algorithm; then, human detection algorithms such as YOLO and Faster R-CNN are used to detect the positions of all personnel in the frame in real time; finally, the Euclidean distance between other personnel and the target patient is calculated, and when the distance is less than a preset distance threshold, it is recorded as an interactive event. To further improve detection accuracy, clothing color recognition (white coat detection) and behavioral pattern analysis (walking trajectory characteristics of medical personnel) can be combined for auxiliary judgment.

[0023] It should be noted that in actual implementation, all image features are extracted and the original image data is discarded immediately, retaining only numerical information such as spatial coordinates and distances, thereby fully protecting patient privacy while achieving technical implementation.

[0024] Through the aforementioned environmental monitoring module, the system can capture external stimuli that may trigger stress responses in elderly patients and record the time of their occurrence as trigger nodes, which provides an accurate time reference for subsequent time lag analysis.

[0025] return Figure 1 The data acquisition module is configured to acquire the target patient's blood pressure and physiological stress data after the trigger node.

[0026] After detecting an environmental stimulus event, it is necessary to collect the patient's physiological response data for causal correlation analysis. The data acquisition module adopts an event-driven windowed acquisition strategy, that is, starting from the trigger node, it continuously collects blood pressure data and physiological stress data within a preset time window. Compared with the traditional continuous full-time acquisition method, this strategy can capture key response processes while reducing data storage and computational burden.

[0027] Blood pressure data can be obtained using a cuff-type blood pressure monitor, with measurement intervals typically set to 1-3 minutes. Alternatively, continuous non-invasive blood pressure monitoring methods, such as pulse wave propagation time (PTT) estimation, can be used to obtain blood pressure sequences with higher temporal resolution.

[0028] Physiological stress data are used to characterize the intensity and temporal characteristics of a patient's physiological response to environmental stimuli, and can be obtained using heart rate data or skin conductance data.

[0029] Heart rate data can be acquired in real time using a wrist-based photoplethysmography (PPG) device, with a sampling rate typically above 1 Hz. Heart rate is one of the most direct indicators of sympathetic nerve activation; when an individual is stimulated, sympathetic nerve excitation leads to a rapid increase in heart rate.

[0030] Skin conductance response (SCR) data can be acquired using GCR sensors, which can be either clip-on or patch-type sensors. SCR reflects changes in sweat gland activity and is a sensitive indicator of stress. When an individual experiences stress or tension, sympathetic nerve activation stimulates increased sweat gland secretion, leading to a significant increase in skin conductivity.

[0031] Through the aforementioned data acquisition module, the system can promptly collect patients' blood pressure and physiological stress data after environmental stimuli occur, providing a complete data foundation for subsequent lag feature extraction and causal correlation analysis.

[0032] return Figure 1 The hysteresis feature extraction module is configured to determine a first delay and a second delay based on the trigger node, blood pressure data, and physiological stress data. The first delay is the delay from environmental stimulus to physiological stress response, and the second delay is the delay from physiological stress response to blood pressure peak.

[0033] This is one of the core innovations of this application. Traditional methods treat elevated blood pressure as a direct response to stimuli, neglecting the time lag characteristics of the physiological systems of the elderly. In reality, the process from external stimulus to the eventual elevation of blood pressure involves two consecutive physiological processes: first, the stimulus is perceived, triggering a stress response in the autonomic nervous system, manifested as an increase in heart rate and skin conductance; subsequently, the stress response further induces a regulatory response in the cardiovascular system, resulting in elevated blood pressure. Each of these processes has specific time delay characteristics, and these characteristics exhibit a significantly different distribution pattern in the elderly compared to younger populations.

[0034] By extracting these two delays and matching them with the statistical distribution of the elderly population, it can be determined whether the current blood pressure fluctuations conform to the typical time pattern induced by emotions, thereby distinguishing between emotional fluctuations and pathological fluctuations.

[0035] The first delay reflects the neural conduction and activation process from environmental stimuli to physiological stress response. This process involves the reception of sensory organs, the conduction of neural signals, cognitive assessment by the brain, and activation of the autonomic nervous system. Due to factors such as decreased neural conduction speed and prolonged cognitive processing time, the first delay of the elderly is usually significantly longer than that of young people, typically ranging from 3 to 30 seconds, while young people usually show obvious stress response within 1 to 5 seconds.

[0036] In the first implementation, the physiological stress data is heart rate data, and the determination of the first delay includes the following steps:

[0037] First, heart rate data is extracted within a first preset time window starting from the trigger node. The length of the first preset time window is set according to the physiological characteristics of the elderly, usually 30-60 seconds. This window length is sufficient to cover the response time from stimulation to peak heart rate for most elderly people.

[0038] Then, within the first preset time window, the first heart rate peak exceeding a preset heart rate threshold is identified in the heart rate data. The preset heart rate threshold can be set to an increase of 10-15 beats / minute relative to the average heart rate in the 30 seconds prior to the trigger point. This threshold setting is based on physiological research and can identify heart rate increases caused by stress responses while excluding normal heart rate fluctuations. When identifying heart rate peaks, a 3-5 second moving average filter can be used to suppress the influence of measurement noise.

[0039] Finally, the time difference between the peak heart rate and the trigger point is calculated to obtain the first delay.

[0040] In the second embodiment, the physiological stress data are skin conductance response data, and the determination of the first delay includes the following steps:

[0041] First, skin conductance response data are extracted within a second preset time window, starting from the trigger node. The second preset time window is usually the same as or slightly longer than the first preset time window, because the onset of the skin conductance response may lag slightly behind the heart rate response.

[0042] Then, within a second preset time window, the first rising edge in the skin conductance response data exceeding a preset conductivity change threshold is identified. A typical pattern of skin conductance response is a latency period after stimulation, followed by a rapid increase in skin conductivity, forming a distinct rising edge. The preset conductivity change threshold is typically set to an increase of 0.05-0.1 microsiemens / second relative to the average skin conductivity in the 30 seconds prior to the trigger point. The identification of the rising edge initiation point can be achieved through first-order derivative analysis: the time derivative of the skin conductance signal is calculated, and the point where the derivative first exceeds the threshold is the rising edge initiation point.

[0043] Finally, the time difference between the rising edge start point and the trigger node is calculated to obtain the first delay.

[0044] It should be noted that heart rate response and skin conductance response each have their advantages and disadvantages as stress indicators: heart rate data is easy to collect and has high temporal resolution, but may be affected by the patient's physical activity; skin conductance response is more sensitive and specific to stress, but requires a dedicated sensor and the signal is easily affected by skin condition. Therefore, in the third embodiment, heart rate data and skin conductance response data are used simultaneously, and the robustness of the first delay extraction is improved through multi-source fusion.

[0045] Specifically, the third implementation includes both heart rate data and skin conductance data. In this third implementation, the determination of the first delay includes the following steps:

[0046] First, the time difference between the time point corresponding to the heart rate peak calculated in the first implementation method and the trigger node is used as the first candidate delay.

[0047] Then, extract the skin conductance response data within the first preset time window; within the first preset time window, identify the first rising edge start point in the skin conductance response data that exceeds the preset conductivity change threshold; calculate the time difference between the rising edge start point and the trigger node to obtain the second candidate delay.

[0048] Finally, based on preset weight coefficients, the first candidate delay and the second candidate delay are weighted and fused to obtain the first delay.

[0049] The expression for weighted fusion is:

[0050] ;

[0051] in, For the final first delay, Delay as the first candidate As the second candidate for delay, and For the preset weighting coefficients, satisfy .

[0052] Weighting coefficient and The settings can be dynamically adjusted based on data quality. In the basic implementation plan, fixed weights can be used. and This reflects the generalizability of heart rate data collection. In more advanced implementations, weights can be dynamically adjusted based on signal quality metrics; for example, when significant motion artifacts appear in the heart rate data, the weights can be reduced. When the signal-to-noise ratio of the skin conductance signal is low, reduce This is a standard technical method and will not be described in detail here.

[0053] By using the above multi-source fusion, we can comprehensively utilize the complementary information of different physiological indicators, improve the accuracy and robustness of the first delay estimation, and reduce the error caused by measurement noise or individual differences in a single indicator.

[0054] Returning to the lag feature extraction module, the second delay reflects the cardiovascular system's regulatory process from physiological stress response to peak blood pressure. This process involves the sympathetic nervous system's regulation of the heart and blood vessels: including a series of physiological changes such as increased cardiac contractility, further increase in heart rate, and peripheral vasoconstriction, ultimately leading to elevated blood pressure. Due to factors such as decreased vascular elasticity and reduced baroreceptor sensitivity, the distribution pattern of the second delay in older adults differs significantly from that in younger adults, exhibiting greater variability.

[0055] The determination of the second delay includes the following steps:

[0056] First, extract blood pressure data within a third preset time window, starting from the trigger node. This third preset time window needs to be long enough to capture the complete blood pressure response process, typically set to 2-3 minutes. Considering the temporal resolution limitations of blood pressure measurement (cuff-type blood pressure monitors typically measure every 1-3 minutes), this window usually contains 2-3 blood pressure measurements.

[0057] Then, within the third preset time window, the first blood pressure peak exceeding a preset blood pressure threshold is identified in the blood pressure data. The preset blood pressure threshold can be set to an increase of 10 mmHg (systolic blood pressure) or 5 mmHg (diastolic blood pressure) relative to the average blood pressure before the trigger point. This threshold setting can capture meaningful blood pressure increases while avoiding misinterpreting normal blood pressure fluctuations as responses.

[0058] Next, the time difference between the peak blood pressure point and the trigger point is calculated to obtain the blood pressure delay.

[0059] Finally, the difference between the blood pressure delay and the first delay is calculated to obtain the second delay:

[0060] ;

[0061] in, For the second delay, The time point corresponding to the peak blood pressure. To trigger the node time, For the first delay, The time point corresponding to the peak of physiological stress (peak heart rate or the starting point of the rising edge of skin electrical activity).

[0062] In this way, the complete response process from stimulus to peak blood pressure is decomposed into two segments: stimulus → stress response and stress response → peak blood pressure. The advantages of this decomposition are: first, each delay has a clearer physiological meaning, making it easier to correspond to known physiological mechanisms; second, the statistical distribution of segmented delays is more stable and predictable, improving the accuracy of probability matching; and third, the combination pattern of the two delays can more finely characterize the physiological response features of an individual.

[0063] Through the aforementioned lag feature extraction module, the system quantifies the continuous physiological response process of "environmental stimulus → physiological stress → increased blood pressure" into two time delay features with clear physiological significance, laying the foundation for subsequent probability matching and causal association scoring.

[0064] return Figure 1 The probability calculation module is configured to match the first delay and the second delay with the corresponding reference delay distribution based on the statistics of the elderly population, respectively, to obtain the first matching degree and the second matching degree.

[0065] This is another core innovation of this application. Traditional methods use fixed thresholds for judgment, such as "heart rate increases by more than X beats per minute" or "blood pressure increases by more than Z mmHg within Y seconds." This hard threshold method cannot adapt to the huge individual differences within the elderly population, nor can it utilize the group characteristics contained in large-scale statistical data. This application proposes a probability distribution matching method that compares the individual's delay characteristics with the statistical prior distribution of the elderly population to quantify "how likely the currently observed delay comes from an emotional triggering process."

[0066] The reference delay distribution was obtained through statistical analysis of historical data from a large-scale elderly population. Specifically, for typical cases of emotion-induced hypertension (confirmed by doctor annotation or counterfactual verification), the first and second delay data were collected, and a distribution was fitted.

[0067] For the first delay in the stimulus-response process, its distribution typically follows a Gamma distribution, which characterizes the right-skewed nature of the delay, meaning that most responses are fast and a few are slow. Its probability density function is:

[0068] ;

[0069] in, This indicates the first delay from the occurrence of environmental stimulus E to the appearance of physiological stress response ES. The probability density, Indicates the first delay. For shape parameters, For rate parameters, It is the Gamma function. For the elderly population, it is typically... The parameter value range is [3,6]. The value is [0.1, 0.2] (unit: 1 / second).

[0070] The second delay between stress and blood pressure typically follows a log-normal distribution, with the following probability density function:

[0071] ;

[0072] in, This indicates the period from the onset of the physiological stress response (ES) to the peak blood pressure. The second delay occurred The probability density, Indicates the second delay. The logarithmic mean is... This is the logarithmic standard deviation. For the elderly population, a typical... The parameter value range is [2.5, 3.5]. The value is [0.5, 0.8], reflecting a large degree of variability.

[0073] These distribution parameters can be learned from historical data through maximum likelihood estimation and stratified according to factors such as age group, gender, and underlying diseases to obtain a more refined population prior.

[0074] First matching degree Indicates the first observed delay The probability density function under the reference distribution is calculated using the following formula:

[0075] ;

[0076] Among them, the denominator The peak density of the reference distribution is used to normalize the probability density to the 0-1 interval. The closer it is to 1, the closer the current first delay is to the typical pattern of emotion-induced triggering; The smaller the value, the less likely the delay pattern is to conform to the statistical characteristics of emotion-induced patterns.

[0077] Similarly, the first matching degree The calculation formula is:

[0078] ;

[0079] It's important to note that the matching degree is not a "probability" in the traditional sense, but rather a normalized similarity measure, and its value does not need to satisfy the constraint that the sum is 1. A high matching degree means that the currently observed delayed features are highly similar to known emotion-induced patterns, supporting the hypothesis that "high blood pressure is induced by emotions."

[0080] In practice, a personalized learning mechanism can be introduced. For patients with multiple medical records, the system can update the individualized reference distribution parameters using Bayesian methods based on their historical delayed data, thereby achieving adaptive adjustment from group prior to individual prior. This personalized learning can better adapt to the unique physiological characteristics of each elderly patient, further improving the accuracy of diagnosis.

[0081] Through the aforementioned probability calculation module, the system quantitatively compares the extracted time delay features with the statistical patterns of the elderly population to obtain two matching degree indicators, which provide core probabilistic evidence for subsequent causal association scoring.

[0082] return Figure 1 The correlation scoring module is configured to calculate a causal correlation score representing the strength of the correlation between blood pressure fluctuations and emotions based on the first and second matching degrees.

[0083] The aforementioned first and second matching scores quantify the similarity between the two delays and the emotion-induced pattern, but it is still necessary to integrate these two independent pieces of evidence into a holistic association strength score. The design of the association scoring module reflects the logic of causal reasoning: the conclusion that "elevation of blood pressure is induced by emotion" can only be supported if and only if both delays conform to the typical pattern of emotion-induced patterns; if either delay significantly deviates from the typical pattern, the credibility of this hypothesis decreases.

[0084] In the first implementation, the formula for calculating the causal association score is:

[0085] ;

[0086] in, The first match degree, This represents the second degree of match.

[0087] The formula is designed following the logic of probability theory that "at least one event occurs." From the opposite perspective: This represents the probability of a first delay mismatch. This represents the probability of a second delay mismatch. Multiplying the two together gives the joint probability that both delays are mismatched. Subtracting this value from 1 gives the probability that at least one delay is matched.

[0088] The advantage of this design lies in its simplicity and reasonable mathematical properties: when and When both are relatively high, It will be close to 1, indicating a strong correlation; when and When any of them is lower, The significant decrease reflects the causal reasoning logic that "a weakness in any link of the chain weakens the overall strength of the evidence"; when and When both are low, A value that is very small indicates a weak or no association.

[0089] However, the first implementation described above does not consider the temporal coordination between the two delays. In fact, even if the two delays each match a reference distribution, if their temporal relationship is unreasonable—for example, the second delay is much shorter than the first delay—it means that blood pressure reaches its peak before the stress response has fully developed, which is inconsistent with physiological mechanisms. Therefore, the second implementation introduces a more refined scoring mechanism.

[0090] Specifically, in the second implementation, the formula for calculating the causal association score is:

[0091] ;

[0092] in, The first match degree, For the second degree of matching, To avoid the default minimum positive number where the logarithmic term is zero, , and The preset weighting coefficients can be set in a typical implementation. =0.4、 =0.4、 =0.2 indicates that the matching degree of the two delays each accounts for 40% of the weight, and the coordination accounts for 20%. These weights can be optimized on the training data through cross-validation. This is a synchronization consistency indicator.

[0093] The formula consists of three parts: the first two terms and The matching degree of the first and second delays was quantified separately. Logarithmic form was used to reflect different information gains; for example, "the matching degree increasing from 0.1 to 0.2" and "from 0.8 to 0.9" have different information gains, consistent with the principles of information theory. (Third term) A coordination constraint between the two delay segments was introduced.

[0094] Synchronous Consistency Indicators satisfy:

[0095] ;

[0096] in, For the second delay, For the first delay, The preset growth coefficient, The typical value range is [0.1, 0.2].

[0097] The above synchronization consistency indicators Using the Sigmoid function form, it has the following characteristics:

[0098] when > That is, when the time from stress response to peak blood pressure is longer than the time from stimulus to stress response, A value close to 1 indicates good timing coordination and conformity to physiological logic; when... ≈ hour, ≈0.5 indicates moderate coordination; when < hour, A value close to 0 indicates a lack of temporal consistency, negatively impacting the score. Growth coefficient. The steepness of the sigmoid function was controlled. The larger the value, the more severe the penalty for temporal inconsistencies.

[0099] It should be noted that, after adopting the logarithmic form, the scoring... The range is no longer limited to [0,1], but can now take negative values ​​or large positive values. For easier subsequent determination, linear transformation or normalization can be used to... Map to the [0,1] interval, or directly compare the original score with the threshold.

[0100] Both implementation methods have their advantages and disadvantages: the first method is computationally simple and easy to interpret, making it suitable for scenarios with high interpretability requirements; the second method introduces more physiological constraints, resulting in stronger discrimination capabilities, and is suitable for scenarios with high accuracy requirements and sufficient training data. In actual deployment, the appropriate solution can be selected based on application needs and data conditions.

[0101] Through the aforementioned correlation scoring module, the system integrates the matching evidence of the two-segment delay into a comprehensive causal correlation score. This score quantifies the strength of support for the hypothesis that "high blood pressure is induced by emotions," providing a core basis for subsequent judgment.

[0102] return Figure 1 The determination module is configured to determine the nature of blood pressure fluctuations based on causal correlation scores.

[0103] The determination rules of the determination module are as follows:

[0104] When the causal association score is greater than the first threshold, the blood pressure is determined to be emotional fluctuation. The first threshold is usually set at 0.7-0.8, and the specific value can be determined based on receiver operating characteristic curve (ROC) analysis of historical data.

[0105] When the causal correlation score is less than the second threshold, the blood pressure is determined to be a non-emotional fluctuation. The second threshold is usually set at 0.3-0.4, and the specific value is determined in the same way as the first threshold.

[0106] When the causal correlation score is between the first threshold and the second threshold, counterfactual verification is performed.

[0107] Specifically, when the causal correlation score S is greater than the first threshold, it indicates that the current blood pressure elevation has very obvious emotional triggering characteristics: both delays highly match the typical pattern of emotional triggering, and the temporal coordination is good. At this time, the system determines that the blood pressure fluctuation is emotional.

[0108] When the causal correlation score S is less than the second threshold, it indicates that the current blood pressure elevation does not conform to the characteristic pattern of emotion-induced fluctuations: there is at least one delay that significantly deviates from the reference distribution, or the temporal coordination is very poor, or no environmental stimulus event is detected at all. In this case, the system determines that the blood pressure fluctuation is non-emotional.

[0109] When the causal association score S falls between the first and second thresholds, it indicates that the current evidence is insufficient to make a high-confidence judgment. This ambiguity may stem from various reasons: the delayed characteristic is on the edge of the typical pattern, the patient has both emotional and potential pathological factors, or there is significant measurement noise. In such cases, the system does not immediately draw a conclusion but instead performs counterfactual verification. The core idea of ​​counterfactual verification is: if the elevated blood pressure is indeed induced by emotions, then the blood pressure should decrease after the emotional stimulus is eliminated or alleviated; conversely, if the blood pressure persists, it supports the pathological hypothesis.

[0110] Specifically, counterfactual verification includes the following steps:

[0111] First, after waiting for a preset time, blood pressure data is retrieved again, and the setback value is calculated. The preset time is usually set to 5-10 minutes, based on the natural recovery time of emotionally induced blood pressure elevation: emotionally induced blood pressure elevation usually begins to decline within 5-15 minutes after the stimulus disappears, while blood pressure elevation caused by non-emotional factors, such as pathological blood pressure elevation, often persists or continues to rise. The setback value is calculated by comparing the peak blood pressure with the re-measured blood pressure value after waiting for the preset time; the difference between the two is the setback value.

[0112] Then, if the drop value is greater than a preset drop threshold, the blood pressure is determined to be an emotional fluctuation; otherwise, it is determined to be a non-emotional fluctuation. The preset drop threshold is usually set at 10 mmHg (systolic blood pressure) or 5 mmHg (diastolic blood pressure). A significant drop in blood pressure supports the emotionally induced hypothesis, while a persistently low blood pressure or a very small drop supports the non-emotionally induced hypothesis.

[0113] In the above-mentioned scheme of this application, if we passively rely on the patient to eliminate the influence of emotions and wait for the blood pressure to drop naturally, there are two problems: First, the emotional recovery time of elderly patients is relatively long, which may lead to an excessively long verification waiting time; Second, the background noise and anxiety atmosphere that are constantly present in the hospital environment may make it difficult for patients to calm down naturally, affecting the effectiveness of counterfactual verification.

[0114] Therefore, this application further proposes that the system also includes a soothing prompt module, configured to generate a prompt instruction for soothing the target patient's emotions when the causal association score is between a first threshold and a second threshold, and send the prompt instruction to an execution terminal for executing the prompt instruction to soothe the target patient.

[0115] The design of the soothing prompt module is based on the concept of proactive intervention: by providing multimodal soothing measures, it accelerates the emotional calming process, thereby obtaining clearer verification results in a shorter time. If the elevated blood pressure is indeed induced by emotions, proactive soothing measures will accelerate the rate at which blood pressure drops; if the elevated blood pressure is dominated by pathological factors, even if soothing measures are provided, blood pressure will not drop significantly. This difference further enhances the accuracy of the judgment.

[0116] The prompts can take various forms: visual cues, auditory cues, environmental control instructions, and text prompts. The execution terminals can be bedside monitoring screens, smart bracelets or mobile phones worn by the patient, smart speakers in the room, and notification terminals for medical staff. Multi-terminal collaboration can provide a more comprehensive reassurance effect.

[0117] In summary, the big data-driven blood pressure emotional fluctuation identification and judgment system for the elderly proposed in this application, by modeling the two-segment time lag characteristics of "environmental stimulus → physiological stress → blood pressure change" and combining the statistical prior distribution of the elderly population for probability matching, has achieved the distinction between emotional and non-emotional blood pressure fluctuations, solving the technical problem of difficulty in distinguishing between emotional and non-emotional blood pressure fluctuations in the prediction of cardiovascular diseases in the elderly.

[0118] It should be understood that the determination result of this application, "non-emotional blood pressure fluctuations," only indicates that the current blood pressure fluctuations are not primarily caused by emotions, suggesting that further attention and examination by medical staff are needed. It does not constitute a diagnosis of any cardiovascular disease in itself.

[0119] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0120] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0121] It should also be noted that in this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions in this application.

[0122] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0123] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A big data-driven system for identifying and judging emotional fluctuations in blood pressure in the elderly, used to distinguish between emotional and non-emotional fluctuations in blood pressure in elderly patients, characterized by: include: The environmental monitoring module is configured to continuously detect environmental stimuli and record the time of occurrence of the environmental stimuli as trigger nodes; The data acquisition module is configured to acquire the target patient’s blood pressure data and physiological stress data after the triggering node, wherein the physiological stress data includes at least one of heart rate data and skin conductance data. The lag feature extraction module is configured to determine a first delay and a second delay based on the trigger node, the blood pressure data, and the physiological stress data. The first delay is the delay from environmental stimulus to physiological stress response, and the second delay is the delay from physiological stress response to blood pressure peak. The probability calculation module is configured to match the first delay and the second delay with the corresponding reference delay distribution based on the statistics of the elderly population, respectively, to obtain the first matching degree and the second matching degree. The correlation scoring module is configured to calculate a causal correlation score representing the strength of the correlation between blood pressure fluctuations and emotions based on the first matching degree and the second matching degree. The determination module is configured as follows: When the causal correlation score is greater than the first threshold, the blood pressure is determined to be an emotional fluctuation. When the causal correlation score is less than the second threshold, the blood pressure is determined to be a non-emotional fluctuation. When the causal correlation score is between the first threshold and the second threshold, the following is executed: After waiting for a preset time, blood pressure data is retrieved again and the drop value of the blood pressure data is calculated. When the drop value is greater than the preset drop threshold, the blood pressure is determined to be an emotional fluctuation; otherwise, it is determined to be a non-emotional fluctuation.

2. The big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly according to claim 1, characterized in that: The environmental stimuli include sound events and interactive events; The sound events include at least one of the following: a queuing event, a medical device alarm event, and an event where the ambient noise intensity is higher than a preset noise threshold. The interactive event is an event in which medical personnel enter a range centered on the target patient and with a preset distance as the radius.

3. The big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly according to claim 2, characterized in that, The environmental monitoring module includes: A sound detection unit is configured to detect the sound event; An image detection unit is configured to acquire a video stream associated with a target patient and detect the interaction events based on video analysis.

4. The big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly according to claim 1, characterized in that, The formula for calculating the causal association score is as follows: in, For the first matching degree, This represents the second matching degree.

5. The big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly according to claim 1, characterized in that, The formula for calculating the causal association score is as follows: in, For the first matching degree, For the second matching degree, To avoid the default minimum positive number where the logarithmic term is zero, , and The preset weighting coefficients, For synchronization consistency metrics, the following must be satisfied: in, For the second delay, For the first delay, This is the preset growth coefficient.

6. The big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly according to claim 1, characterized in that: It also includes a soothing prompt module, configured to generate a prompt instruction for soothing the target patient's emotions when the causal correlation score is between the first threshold and the second threshold, and send the prompt instruction to an execution terminal for executing the prompt instruction to soothe the target patient.

7. The big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly according to claim 1, characterized in that, The physiological stress data is heart rate data, and the determination of the first delay includes: Extract heart rate data within a first preset time window starting from the trigger node; Within the first preset time window, identify the first heart rate peak in the heart rate data that exceeds a preset heart rate threshold; The first delay is obtained by calculating the time difference between the time point corresponding to the heart rate peak and the trigger node.

8. The big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly according to claim 7, characterized in that, The physiological stress data also includes skin conductance data, and the determination of the first delay includes: The time difference between the time point corresponding to the heart rate peak and the trigger node is used as the first candidate delay; Extract skin conductance response data within the first preset time window; Within the first preset time window, identify the starting point of the first rising edge in the skin conductance response data that exceeds the preset conductivity change threshold; Calculate the time difference between the rising edge start point and the trigger node to obtain the second candidate delay; Based on preset weighting coefficients, the first candidate delay and the second candidate delay are weighted and fused to obtain the first delay.

9. The big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly according to claim 1, characterized in that, The physiological stress data are skin conductance data, and the determination of the first delay includes: Extract skin conductance response data within a second preset time window starting from the trigger node; Within the second preset time window, identify the starting point of the first rising edge in the skin conductance response data that exceeds the preset conductivity change threshold; The first delay is obtained by calculating the time difference between the rising edge start point and the trigger node.

10. The big data-driven system for identifying and determining blood pressure and emotional fluctuations in the elderly according to any one of claims 7 to 9, characterized in that, The determination of the second delay includes: Extract blood pressure data within a third preset time window starting from the trigger node; Within the third preset time window, identify the first blood pressure peak in the blood pressure data that exceeds a preset blood pressure threshold; Calculate the time difference between the time point corresponding to the blood pressure peak and the trigger node to obtain the blood pressure delay; The difference between the blood pressure delay and the first delay is calculated to obtain the second delay.

Images