A remote follow-up scheduling method based on physiological data analysis

Abstract

The present application relates to the technical field of follow-up scheduling, in particular to a remote follow-up scheduling method based on physiological data analysis. The method acquires multi-dimensional physiological data and medication time of a target object in real time; determines an analysis period based on heart rate deviation degree, obtains abnormal factors by combining physiological data fluctuation and abnormal proportion of each dimension, thereby determining abnormal degree of the analysis period and screening suspected abnormal period; determines a reference time based on the medication time and the night rest period, obtains a risk index by calculating the time interval and heart rate difference of the suspected abnormal period relative to the reference time; aggregates the abnormal degree and the risk index to obtain a single-day abnormal risk coefficient, obtains a physiological state evaluation value by combining the change trend of the abnormal risk coefficient of the history multi-days, and further adaptively adjusts the remote follow-up frequency. The present application realizes deep correlation analysis of physiological fluctuation and clinical background, improves the accuracy of postoperative follow-up, and can effectively optimize medical resource allocation.

Description

Technical Field

[0001] This invention relates to the field of follow-up scheduling technology, and specifically to a remote follow-up scheduling method based on physiological data analysis. Background Technology

[0002] With the rapid development of modern medical technology and the popularization of the concept of comfortable medical care, patients have placed higher demands on their perioperative and postoperative rehabilitation experience. The post-anesthesia stage is a critical period for patients to transition from the acute trauma phase to the stable recovery phase. During this stage, patients face pain caused by surgical trauma, inflammatory reactions, and potential physiological fluctuations due to residual anesthetic drugs. Therefore, it is necessary to pay close attention to the effectiveness of postoperative analgesia and the recovery of the overall physiological state. In order to promptly grasp the patient's analgesia status and recovery progress after discharge, postoperative follow-up and remote pain management have become core links. If postoperative pain is not effectively controlled and intervened, it will not only increase the patient's physical pain and anxiety, but also trigger a series of chain reactions: for example, sustained sympathetic nerve excitation leading to increased blood pressure and heart rate, thereby increasing the risk of cardiovascular events; lung infections caused by the inability to breathe deeply or cough due to pain; more seriously, long-term and recurrent acute pain, if not properly managed, can easily turn into incurable chronic pain, seriously affecting the patient's quality of life and consuming a lot of medical resources.

[0003] Currently, postoperative pain management and follow-up mostly adopt standardized prescriptions and fixed follow-up models. For example, doctors prescribe a fixed dose of analgesics when patients are discharged and set a fixed follow-up schedule (such as telephone follow-ups on the 3rd and 7th day after surgery). However, existing models fail to adequately consider individual patient differences. Patients vary significantly in surgical procedures, baseline physical condition, pain tolerance, and analgesic drug metabolism rates. Fixed follow-up frequencies often lead to insufficient analgesia for high-risk patients (follow-up lag) or overtreatment for stable patients (follow-up redundancy). Secondly, current follow-up relies heavily on patients' subjective reports, lacking objective and continuous physiological data. Patient descriptions of pain are often influenced by emotions and cognition, and transient but dangerous physiological fluctuations occurring during follow-up intervals (such as abnormal vital signs caused by nocturnal burst pain) are difficult to capture and assess in a timely manner. Finally, existing remote monitoring systems primarily focus on threshold alarms for single physiological indicators, failing to conduct in-depth spatiotemporal correlation analysis of physiological fluctuations with external intervention events (such as medication time) and natural physiological cycles (such as nighttime sleep). This makes it difficult to distinguish between normal physiological fluctuations and real physiological crises caused by analgesia failure, easily generating numerous invalid alarms and increasing the burden on medical staff.

[0004] Therefore, the existing postoperative follow-up management model lacks dynamic integration of multidimensional physiological time-series data and correlation analysis with the clinical intervention background. It cannot adaptively adjust the follow-up frequency according to the real-time recovery status of individual patients, making it difficult to achieve precise and efficient remote follow-up scheduling while ensuring analgesic effects. Summary of the Invention

[0005] To address the technical problem that existing postoperative follow-up management models lack dynamic fusion of multidimensional physiological time-series data and correlation analysis with clinical intervention background, thus failing to adaptively adjust follow-up frequency based on the real-time recovery status of individual patients, the present invention aims to provide a remote follow-up scheduling method based on physiological data analysis. The specific technical solution adopted is as follows: This invention provides a remote follow-up scheduling method based on physiological data analysis, the method comprising the following steps: Real-time acquisition of multi-dimensional physiological data and medication administration time of the target object; among which, physiological data includes heart rate data; The analysis period is determined based on the deviation of the resting heart rate data from the preset baseline; the abnormal factors corresponding to each dimension of physiological data in each analysis period are obtained based on the fluctuation and abnormality rate of each dimension of physiological data in each analysis period; the degree of abnormality of each analysis period is obtained based on the abnormality of each dimension of physiological data at each moment in each analysis period and the abnormal factors, thereby screening out suspected abnormal periods. Reference times are determined based on medication administration time and nighttime rest period; risk index for each suspected abnormal time period is obtained based on the time interval between each suspected abnormal time period and the nearest reference time period, as well as the difference in heart rate data. Aggregate the degree of abnormality and risk index of each suspected abnormal period within a single day to obtain the abnormal risk coefficient for that day; based on the magnitude and changes of the abnormal risk coefficient over multiple consecutive historical days, obtain the physiological state assessment value at the current moment. The frequency of remote follow-up for target subjects is dynamically adjusted based on physiological status assessment values.

[0006] Furthermore, the method for obtaining the analysis period is as follows: In the resting state, the heart rate data at each moment is subtracted from the preset baseline and then normalized to obtain the degree of heart rate deviation at the corresponding moment. When the heart rate deviation exceeds the preset heart rate deviation threshold, the corresponding time will be marked as the analysis time. The time period consisting of consecutive adjacent analysis moments on the time axis is defined as the analysis period.

[0007] Furthermore, the method for obtaining the abnormal factor is as follows: For any dimension of physiological data and any analysis period, the difference between the physiological data of that dimension and the baseline data at each moment within the analysis period is obtained as anomaly analysis data. When the abnormal analysis data exceeds the preset abnormal analysis threshold, the corresponding time will be marked as the abnormal analysis time of the physiological data in that dimension. The percentage of times an anomaly is analyzed within the total time frame of the analysis period is taken as the anomaly percentage. The fluctuation analysis value is obtained by subtracting the minimum value from the maximum value of the physiological data in this dimension within the analysis period and dividing the result by the total duration of the analysis period. The product of the abnormality rate and the normalized fluctuation analysis value is used as the abnormality factor corresponding to the physiological data of this dimension within the analysis period.

[0008] Furthermore, the method for obtaining the degree of abnormality is as follows: For any moment within any analysis period, the physiological data corresponding to that moment as the abnormal analysis moment are all considered as abnormal physiological data, thereby determining the set of abnormal physiological data at that moment. Based on the overlap of abnormal physiological data sets at adjacent time points, time segments that are continuous on the time axis of the analysis period and whose abnormal physiological data sets are not empty are taken as comprehensive abnormal segments. The longest comprehensive abnormal segment is taken as the main segment, and the ratio of the number of abnormal physiological data dimensions contained in the main segment to the total number of physiological data dimensions is taken as the first analysis value; the duration of the main segment in this analysis period is taken as the second analysis value. Based on the first analysis value, the second analysis value, and the corresponding maximum anomaly factor within the analysis period, the degree of anomaly for that analysis period is obtained; wherein, the first analysis value, the second analysis value, and the maximum anomaly factor are all positively correlated with the degree of anomaly.

[0009] Furthermore, the method for obtaining the suspected abnormal time period is as follows: When the degree of abnormality exceeds the preset abnormality threshold, the corresponding analysis period will be designated as a suspected abnormal period.

[0010] Furthermore, the method for obtaining the reference time is as follows: For any medication administration time, the time corresponding to the preset drug onset time is taken as the first reference time, starting from that medication administration time and moving backward along the time axis; The entire day is divided according to a preset basic inspection cycle, and the time corresponding to each cycle node is obtained as the second reference time. The reference time is determined based on the first reference time and the second reference time.

[0011] Furthermore, the method for obtaining the risk index is as follows: For any suspected abnormal time period, the reference time that is the closest to the start time of the suspected abnormal time period in terms of time sequence is taken as the target reference time; The interval between the target reference time and the start time of the suspected abnormal period is used as the initial duration. When the initial duration is greater than the preset correlation attenuation limit, the preset correlation attenuation limit will be used as the analysis duration; when the initial duration is less than or equal to the preset correlation attenuation limit, the initial duration will be used as the analysis duration. The difference between the mean of all heart rate data within the suspected abnormal period and the heart rate data at the target reference time is used as the basis for analysis. If the suspected abnormal time period falls within the nighttime rest period, the preset nighttime sensitive gain coefficient will be used as the time period gain coefficient; if the suspected abnormal time period does not fall within the nighttime rest period, the preset normal gain coefficient will be used as the time period gain coefficient; wherein, the preset nighttime sensitive gain coefficient is greater than the preset normal gain coefficient. Based on the analysis duration, analysis difference, and the time period gain coefficient corresponding to the suspected abnormal period, the risk index of the suspected abnormal period is obtained; among them, the analysis duration is negatively correlated with the risk index, while the analysis difference and the time period gain coefficient are both positively correlated with the risk index.

[0012] Furthermore, the method for obtaining the abnormal risk coefficient is as follows: The average of the anomalies in all suspected anomaly periods within a single day is obtained as the overall anomaly analysis value. The maximum value of the risk index for each suspected abnormal period within a single day is used as the main risk analysis value; Based on the overall anomaly analysis value and the main risk analysis value, the daily anomaly risk coefficient is obtained; both the overall anomaly analysis value and the main risk analysis value are positively correlated with the anomaly risk coefficient.

[0013] Furthermore, the method for obtaining the physiological state assessment value is as follows: Dates with abnormal risk coefficients exceeding the preset risk concern threshold from a series of consecutive historical days are designated as concern days. The average interval between adjacent concern days is calculated and used as the interval assessment value. Linear fitting is performed on the sequence of abnormal risk coefficients over multiple consecutive historical days in chronological order to obtain the fitting slope that characterizes the trend of abnormal risk coefficient changes. A trend evaluation value is obtained based on the fitting slope; wherein, the fitting slope and the trend evaluation value are negatively correlated. The normalized result of the product of the interval assessment value and the trend assessment value is used as the physiological state assessment value.

[0014] Furthermore, the method for dynamically adjusting the remote follow-up frequency of the target subjects based on physiological state assessment values ​​is as follows: When the physiological status assessment value is greater than or equal to the preset safety assessment threshold, the initial follow-up frequency will be used as the current remote follow-up frequency for the target object. When the physiological status assessment value is less than the preset safety assessment threshold, the target adjustment frequency is obtained based on the initial follow-up frequency, the preset safety assessment threshold, and the physiological status assessment value; among them, the physiological status assessment value and the target adjustment frequency are negatively correlated. When the target adjustment frequency is greater than the preset frequency limit, the preset frequency limit will be used as the current remote follow-up frequency of the target object. When the target adjustment frequency is less than or equal to the preset frequency upper limit, the target adjustment frequency will be used as the current remote follow-up frequency of the target object.

[0015] The present invention has the following beneficial effects: This invention first determines the analysis period based on the deviation of resting heart rate data from a preset baseline, achieving precise positioning of the high-stress physiological window and effectively filtering out daily physiological noise interference. To quantify the depth and duration of deviation in single-dimensional physiological data, it further obtains the corresponding abnormal factors for each dimension of physiological data within each analysis period based on the fluctuations and abnormal percentages of each dimension's physiological data. This accurately reflects the severity and density of deviations from the steady state of each physiological system, facilitating differentiated characterization of the sensitivity of different physiological dimensions to pain stimuli. Furthermore, based on the abnormalities and abnormal factors of each dimension's physiological data at each moment within each analysis period, it obtains the degree of abnormality for each analysis period, accurately reflecting the confidence level of simultaneous abnormalities in multi-dimensional physiological data. This accurately filters out suspected abnormal periods, and the multi-dimensional physiological data collaborative verification mechanism significantly reduces the system's false alarm rate caused by single-sensor errors. Considering the clear time-sensitivity of analgesic intervention effects and the high-risk period at night, a reference time is determined based on the medication administration time and the nighttime rest period, establishing a relationship between physiological performance, expected efficacy, and natural physiological rhythms. The system establishes a spatiotemporal correlation benchmark for pain management. To assess the immediate effectiveness of analgesia protocols, it obtains a risk index for each suspected abnormal time period based on the time interval between each suspected abnormal time period and the nearest reference time, as well as the difference in heart rate data. This accurately reflects the urgency of pain control failure and the intensity of physiological state dysregulation, facilitating the precise capture of clinical crises caused by insufficient analgesia. Considering the need for continuous, 24 / 7 rehabilitation quality assessment, it aggregates the degree of abnormality and risk index of each suspected abnormal time period within a single day to obtain a daily abnormal risk coefficient, achieving a comprehensive quantification of the overall rehabilitation pressure on the target subjects. To identify the long-term evolution trend from acute pain to chronic pain, it obtains the current physiological state assessment value based on the magnitude and changes of abnormal risk coefficients over multiple consecutive days, while also considering the frequency of risk events and dynamic deterioration trends, significantly improving the foresight of risk prediction. Furthermore, based on the physiological state assessment value, it dynamically adjusts the remote follow-up frequency of the target subjects, constructing an adaptive mapping mechanism between follow-up resource allocation and individual real-time risk. This effectively avoids resource redundancy while ensuring postoperative patient safety, maximizing medical benefits. Attached Figure Description

[0016] Figure 1 A schematic flowchart illustrating a remote follow-up scheduling method based on physiological data analysis, provided as an embodiment of the present invention; Figure 2 This is a structural diagram of a remote follow-up scheduling system based on physiological data analysis, provided as an embodiment of the present invention. Detailed Implementation

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0018] The following description, in conjunction with the accompanying drawings, details a specific scheme for a remote follow-up scheduling method based on physiological data analysis provided by the present invention.

[0019] Example 1: This invention proposes a remote follow-up scheduling method based on physiological data analysis. Please refer to [link / reference]. Figure 1 The diagram illustrates a schematic flowchart of a remote follow-up scheduling method based on physiological data analysis according to an embodiment of the present invention. The method includes the following steps: Step S1: Acquire multi-dimensional physiological data and medication administration time of the target object in real time; among which, physiological data includes heart rate data.

[0020] Specifically, this embodiment uses a single target subject (i.e., an individual in the post-anesthesia recovery phase) as an example for analysis; all subsequent references to this target subject will refer to it. Considering the significant individual differences among different target subjects during post-operative recovery, a fixed follow-up frequency is insufficient to meet the needs of precise rehabilitation. To monitor the physiological recovery status of the target subject in real time, this embodiment acquires multi-dimensional temporal characteristics of the target subject through multi-source sensor collaboration. Specifically, it collects heart rate, respiratory rate, blood oxygen saturation, and body movement data in real time during daily life using a smart wearable device (e.g., a smart bracelet) worn by the target subject with motion monitoring capabilities; simultaneously, it acquires blood pressure data from a portable blood pressure monitor connected to a smart terminal. For ease of consistency in subsequent algorithms, this embodiment collectively refers to the aforementioned heart rate, blood pressure, and respiratory rate data as physiological data. It should be noted that, in order to ensure the safety of the target subject's life and identify acute physiological risks in a timely manner, this embodiment sets the corresponding vital sign safety range for each dimension of physiological data based on pre-stored clinical medical standards or doctor's advice; during the data collection process, if the physiological data monitored in real time exceeds the corresponding vital sign safety range, the emergency alarm mechanism is immediately triggered and the routine assessment process is interrupted, thereby prioritizing the safety of the target subject's vital signs before analyzing the pain control effect.

[0021] After acquiring the aforementioned multi-dimensional physiological data, this embodiment further preprocesses the original physiological data. Specifically, it employs normalization algorithms such as decimal scaling to map physiological data of different dimensions and orders of magnitude to a similar scale space, thereby eliminating the impact of differences in the numerical magnitudes of different indicators on subsequent evaluations. Therefore, all subsequent physiological data are preprocessed. It should be noted that, considering the significant differences in sampling frequency between heart rate data collected by a smart bracelet (high-frequency continuous sampling) and blood pressure data acquired by a portable blood pressure monitor (discrete triggered sampling), this embodiment also includes a data alignment step in the preprocessing process to ensure the synchronization and comparability of multi-dimensional physiological data on the same time axis. Specifically, using the heart rate data sequence with the highest sampling frequency as the reference time axis, it employs a previous value preservation or linear interpolation algorithm to fill in and align the low-frequency sampled blood pressure and respiratory rate data, ensuring that physiological data of different dimensions have corresponding numerical representations at each reference time. This time alignment preprocessing provides a definite data alignment basis for subsequent steps of intersecting abnormal physiological data sets based on time, effectively solving the temporal misalignment problem in the process of multi-source heterogeneous data fusion. Simultaneously, the activity state of the target object is identified based on the collected body motion data, determining its resting and motion states. This helps to eliminate heart rate and respiratory fluctuations caused by exercise load in subsequent analysis, improving the accuracy of subsequent assessments. The decimal scaling standardization algorithm is well-known and will not be elaborated further.

[0022] In addition, in order to establish a spatiotemporal correlation between physiological fluctuations and external intervention behaviors, this embodiment records the time of each medication administration (i.e., the actual time record of analgesic drug administration) held by the target subject through a smart terminal, as an important reference benchmark for subsequent assessment of the risk of the efficacy window period.

[0023] Step S2: Determine the analysis period based on the deviation of the resting heart rate data from the preset baseline; obtain the abnormal factors corresponding to each dimension of physiological data in each analysis period based on the fluctuation and abnormality rate of each dimension of physiological data in each analysis period; obtain the degree of abnormality of each analysis period based on the abnormality of each dimension of physiological data at each moment in each analysis period and the abnormal factors, and then screen out suspected abnormal periods.

[0024] Specifically, considering that continuous 24 / 7 monitoring generates massive amounts of routine physiological data, in order to reduce computational load and accurately capture the physiological stress response of the target subject, this embodiment determines the analysis period based on the deviation of resting heart rate data from a preset baseline, thus identifying the time periods requiring focused attention. It is known that real physiological pain can trigger interconnected changes in multi-dimensional physiological data (such as increased blood pressure and shortness of breath). To comprehensively assess the severity of physiological fluctuations and ensure that the physiological state assessment process fully integrates the dynamic evolution characteristics of various dimensions of physiological data, and based on the fluctuations and abnormal proportions of each dimension of physiological data within each analysis period, the corresponding abnormal factors for each dimension of physiological data within each analysis period are obtained. This accurately reflects the amplitude and density of fluctuations in the corresponding dimension of physiological data within the corresponding analysis period; whereby the larger the abnormal factor, the more the corresponding dimension of physiological data deviates from a stable state within the corresponding analysis period. Considering that isolated, single physiological data anomalies may be caused by transient sensor noise, in order to improve the anti-interference ability of pain recognition and eliminate interference from non-pain factors through collaborative verification of multi-dimensional physiological data, the degree of anomaly for each analysis period is obtained based on the anomalies and anomalous factors of each dimension of physiological data at each moment within each analysis period. This accurately reflects the likelihood of the target subject experiencing real pain in each analysis period; the greater the degree of anomaly, the more likely the corresponding analysis period is to be triggered by real physiological pain stimuli. Furthermore, the degree of anomaly is used to screen out suspected abnormal periods, laying the foundation for subsequent risk assessment. It should be noted that the preset baseline is obtained by calculating the average value of historical heart rate data from the target subject's preoperative stable state or pre-discharge resting state, ensuring that the preset baseline conforms to the individualized physiological characteristics of the target subject.

[0025] Preferably, in one feasible method of this embodiment, the method for obtaining the analysis period is as follows: under resting state, the heart rate data at each moment is subtracted from a preset baseline and then normalized to obtain the degree of heart rate deviation at the corresponding moment, accurately reflecting the relative intensity of the heart rate surge at the corresponding moment. In this embodiment, after subtracting the preset baseline from the heart rate data, the difference between the heart rate data and the preset baseline is normalized by dividing by the upper limit of the vital sign safety range corresponding to the heart rate data. The greater the degree of heart rate deviation, the stronger the physiological stress response of the target object at the corresponding moment. Therefore, this embodiment sets the preset heart rate deviation threshold to 0.6 to ensure effective filtering of normal heart rate fluctuations caused by daily mild activities. The implementer can set the preset heart rate deviation threshold according to the age, basic medical history and clinical monitoring sensitivity requirements of the target object, which is not limited here. When the heart rate deviation is greater than the preset heart rate deviation threshold, the corresponding moment is marked as the analysis moment, and abnormal nodes worthy of special attention are extracted. In order to capture the duration span of the physiological stress response, the time period consisting of consecutive adjacent analysis moments on the time axis is determined as the analysis period. It should be noted that the analysis period should include at least 3 analysis moments to avoid misjudgments caused by occasional fluctuations in sensor readings or momentary arrhythmias.

[0026] Preferably, in one feasible embodiment of this invention, the method for obtaining abnormal factors is as follows: for any dimension of physiological data and any analysis period, the difference between the physiological data of that dimension and its baseline data (i.e., the preset baseline corresponding to the physiological data of that dimension) at each moment within the analysis period is obtained as abnormal analysis data; the larger the abnormal analysis data, the more serious the deviation of the physiological data of that dimension at the corresponding moment. When the abnormal analysis data is greater than the preset abnormal analysis threshold, it indicates that the physiological data of that dimension has produced obvious stress manifestations at the corresponding moment, and the corresponding moment is marked as the abnormal analysis moment of the physiological data of that dimension; in this embodiment, the preset abnormal analysis threshold is set to 20% of the baseline data corresponding to the physiological data of that dimension, to ensure that the selected abnormal analysis moments meet the medically significant change criteria. The implementer can set the preset abnormal analysis threshold according to the physiological characteristics of the specific dimension and the accuracy of the sensor, which is not limited here; It is known that real pain stimuli cause continuous changes in physiological state. To assess the temporal density of abnormal states, the proportion of abnormal analysis moments within the total analysis period is used as the abnormality percentage. To quantify the overall intensity of change in this dimension of physiological data within the analysis period and eliminate biases caused by varying period lengths, the maximum value minus the minimum value of this dimension of physiological data within the analysis period is divided by the total duration of the analysis period (in minutes), resulting in the fluctuation analysis value. A higher abnormality percentage and fluctuation analysis value indicate that this dimension of physiological data is not only frequently in an abnormal state within the analysis period but also experiences significant oscillations. To comprehensively balance the breadth (temporal coverage) and depth (fluctuation intensity) of physiological responses and prevent a single extreme value from dominating the evaluation results, the product of the abnormality percentage and the normalized fluctuation analysis value is used as the abnormality factor corresponding to this dimension of physiological data within the analysis period. This embodiment normalizes the fluctuation analysis value using the max-min normalization method and maps it to... The interval is used to eliminate the dimensional differences in fluctuation calculations of different physiological dimensions (such as heart rate and blood pressure). The dataset corresponding to the maximum-minimum normalization method here is the set of historical fluctuation analysis values ​​calculated from the physiological data of the target object in the past period (such as the most recent week). If the number of samples in the set of historical fluctuation analysis values ​​is less than the preset cold start threshold (such as 10, which can be set according to the actual situation), the preset clinical standard fluctuation range corresponding to the physiological data of that dimension is used as the normalization benchmark to avoid the distortion of calculation results due to missing samples. In order to avoid the division error caused by the maximum fluctuation analysis value in the dataset being equal to the minimum fluctuation analysis value (i.e., the historical data is completely stable), this embodiment introduces a very small positive number (such as 0.0001) as a smoothing factor in the denominator of the maximum-minimum normalization calculation. That is, the denominator is the result of subtracting the historical minimum fluctuation analysis value from the historical maximum fluctuation analysis value and adding the smoothing factor. This ensures that the accuracy of the normalization mapping is not significantly affected, and that the underlying algorithm can stably output effective values ​​under any extreme data conditions.

[0027] Preferably, in one feasible method of this embodiment, the method for obtaining the degree of abnormality is as follows: considering that severe pain usually causes synchronous abnormalities in multi-dimensional physiological data, in order to accurately identify physiological stress events, for any moment within any analysis period, the physiological data corresponding to that moment as the abnormal analysis moment are all considered as abnormal physiological data, thereby determining the set of abnormal physiological data at that moment; in order to filter out the scattered noise of physiological data rotation fluctuations, based on the overlap of abnormal physiological data sets at adjacent moments, the time segments that are continuous on the time axis of the analysis period and whose abnormal physiological data sets are not empty are taken as comprehensive abnormal segments, and the continuous time periods of concurrent multi-dimensional abnormal physiological data are extracted. To eliminate sporadic and short-lived minor abnormal fluctuations within the analysis period, the longest comprehensive abnormal segment was selected as the principal segment, anchoring the core period of most concentrated and sustained physiological stress within that analysis period. The more physiological dimensions involved in the fluctuations, the higher the confidence level in the occurrence of actual pain. Therefore, the ratio of the number of abnormal physiological data dimensions included in the principal segment to the total number of physiological data dimensions was used as the first analytical value to quantify the breadth of multi-system involvement. Simultaneously, considering that the stress response induced by pain requires a certain period to maintain before having a substantial impact on rehabilitation, the principal segment was selected within the analysis period... The duration percentage within a segment is used as the second analytical value to quantify the sustained stability of the abnormal state. A larger maximum abnormal factor within the analytical period indicates a drastic shift in at least one dimension of physiological data, representing the core intensity of the pain impact. To construct a comprehensive evaluation index integrating the breadth of physiological response, temporal stability, and fluctuation intensity, the degree of abnormality for that analytical period is obtained based on the first analytical value, the second analytical value, and the maximum abnormal factor. The first analytical value, the second analytical value, and the maximum abnormal factor are all positively correlated with the degree of abnormality. Therefore, the formula for calculating the degree of abnormality is: In the formula, Y represents the degree of abnormality; s represents the first analytical value; and t represents the second analytical value. The maximum anomaly factor is denoted by norm, which is the maximum-minimum normalization function. The corresponding dataset is the historical anomaly analysis sample set accumulated since postoperative monitoring of the target object (i.e., containing data from each historical analysis period). The calculation results are as follows: If the number of samples in the historical dataset is less than the preset cold start threshold (e.g., 10), the preset theoretical extreme value (minimum value is 0, maximum value is 3) is directly used as the maximum-minimum benchmark for normalization to prevent numerical polarization in the initial stage of the system; if the number of samples in the historical dataset is greater than or equal to the preset cold start threshold, the actual maximum and minimum values ​​in the historical dataset are used for normalization. In this case, to avoid the denominator of the normalization calculation being zero due to the maximum and minimum values ​​in the historical dataset being equal, a very small positive number (e.g., 0.0001) is introduced into the denominator as a smoothing factor when calculating the maximum-minimum normalization function to ensure that the degree of anomaly can be stably mapped to the target value. The interval provides a standardized numerical basis for setting a unified classification threshold later.

[0028] The greater the known degree of abnormality, the more likely the corresponding analysis period is to be a real physiological crisis or pain event. Therefore, when the degree of abnormality exceeds a preset abnormality threshold, the corresponding analysis period is considered a suspected abnormal period. In this embodiment, the preset abnormality threshold is set to 0.7 to ensure that only analysis periods with high clinical concern are included in subsequent time-series correlation analyses, avoiding false alarms caused by minor daily discomfort. Implementers can set the preset abnormality threshold according to the strictness of follow-up management or clinical sensitivity requirements, which is not limited here.

[0029] Step S3: Determine the reference time based on the medication time and the nighttime rest period; obtain the risk index for each suspected abnormal period based on the time interval between each suspected abnormal period and the nearest reference time, as well as the difference in heart rate data.

[0030] Specifically, considering that simple physiological fluctuations detached from clinical context cannot accurately distinguish between normal postoperative recovery fluctuations and physiological crises caused by insufficient analgesia, in order to provide a causally related benchmark for the assessment of suspected abnormal periods, reference times are determined based on medication time and nighttime rest time, constructing two key risk benchmarks: the expected onset point of drug action and the nighttime stable period. It is known that if physiological abnormalities occur during the window period when analgesics should be effective, or during deep sleep at night, their potential risks and clinical concern value are far greater than during ordinary periods. To quantify the deviation of suspected abnormalities from the expected intervention or rehabilitation context, a risk index is obtained for each suspected abnormal period based on the time interval between each suspected abnormal period and the nearest reference time, as well as the difference in heart rate data. This accurately reflects the degree of pain control failure and rehabilitation risk level of the target subject during the corresponding suspected abnormal period; the higher the risk index, the higher the clinical risk of the corresponding suspected abnormal period, and the greater the need for follow-up intervention.

[0031] Preferably, in one feasible embodiment of this invention, the method for obtaining the reference time is as follows: considering that analgesic drugs require a certain metabolic time to exert their analgesic effect after entering the body, for any medication administration time, the time corresponding to the preset drug onset time is taken as the starting time and moved forward along the time axis; this is used as the first reference time. In this embodiment, the preset drug onset time is set to 30 minutes to ensure that the reference time is anchored to the theoretical onset point of the drug. The implementer can set the preset drug onset time according to the instructions of the analgesic drug used (such as acetaminophen, ibuprofen, or opioids) and the route of administration, which is not limited here; at the same time, considering that if the target object is within a certain period of time (For example, during the day) if medication is not taken for a long time, relying solely on the time of medication administration as a benchmark will lead to a lack of time reference when abnormalities occur, which may easily mask the true risk of sudden pain. In order to ensure the continuity of the monitoring benchmark throughout the day and eliminate the blind spot of assessment during long periods of no intervention, the whole day is divided into time according to a preset basic inspection cycle, and the time corresponding to each cycle node is obtained as a second reference time. In this embodiment, the preset basic inspection cycle is set to 4 hours (e.g., fixed times such as 00:00, 04:00 and 08:00), which ensures that a continuous and stable basic physiological reference system can be established even without external intervention. The implementer can set the preset basic inspection cycle according to the frequency of routine medical rounds, and there is no limitation here. To comprehensively consider the combined impact of exogenous intervention (medication) and daily physiological factors on risk assessment, the first and second reference times are combined to form a reference time set. It should be noted that, to ensure the integrity of the system's computational logic during the initial startup phase (i.e., the initial discharge period before medication is started or the first cycle node is reached), this embodiment pre-stores the target object's initial system operation time (e.g., discharge handover time) in the reference time set as a global initial reference anchor point. This prevents risk index calculation failure due to an empty reference time set during the initial system operation. To ensure that each reference time has analytical significance, duplicate timestamps within overlapping or preset time deviations (e.g., 5 minutes) in the reference time set are deduplicated. Specifically, if the time difference between two reference times is less than or equal to the preset time deviation, only the earlier timetamp is retained, or the average of the two is calculated as the deduplicated reference time. This ensures that when matching a target reference time for any suspected abnormal period, a unique, closest timetamp can be identified from the deduplicated set, eliminating computational redundancy and improving logical certainty.

[0032] Preferably, in one feasible embodiment of this method, the risk index is obtained as follows: considering that the temporal position of the physiological abnormality determines the severity of its clinical nature, in order to examine the correlation between the abnormal event and the most recent baseline state, for any suspected abnormal period, the reference time that is chronologically earlier than the start time of the suspected abnormal period and closest to it is taken as the target reference time, and a specific reference anchor point for the abnormality is established; the interval (in minutes) between the target reference time and the start time of the suspected abnormal period is taken as the initial duration; considering that the inhibitory effect of the intervention state on physiological stress has a time boundary, the timeout is considered as follows. Subsequent abnormalities should be considered as independent sudden events. If the time difference is allowed to accumulate continuously, it will excessively dilute the true physiological risk score. Therefore, when the initial duration is greater than the preset correlation attenuation limit, the preset correlation attenuation limit is used as the analysis duration; when the initial duration is less than or equal to the preset correlation attenuation limit, the initial duration is used as the analysis duration. In this embodiment, the preset correlation attenuation limit is set to 240 minutes. Through the above time clamping process, the problem of infinite amplification of the denominator caused by long periods without external intervention or baseline recording is avoided. Among them, the shorter the analysis duration, the closer the suspected abnormality occurs to the target reference time, and the higher the rehabilitation risk and intervention urgency corresponding to the suspected abnormality period. Considering that the heart rate data at the reference time represents the expected stable baseline, while the heart rate data during the suspected abnormal period represents the degree of deviation, in order to quantify the dispersion of the heart rate data in the suspected abnormal period relative to the target reference time and capture the dynamic rebound characteristics of the physiological state, the absolute value of the difference between the mean of all heart rate data in the suspected abnormal period and the heart rate data at the target reference time is used as the analytical difference; the larger the analytical difference, the more severe the physiological loss of control of the target subject during the suspected abnormal period. Considering the low basal metabolic rate at night, physiological fluctuations occurring during this period often represent a more serious risk to recovery. Therefore, it is determined whether the suspected abnormal period falls within the nighttime rest period (i.e., 23:00 to 06:00 the next day). If the suspected abnormal period falls within the nighttime rest period, a preset nighttime sensitivity gain coefficient is used as the time period gain coefficient; if the suspected abnormal period does not fall within the nighttime rest period (i.e., falls within the regular daytime period), a preset regular gain coefficient is used as the time period gain coefficient. The nighttime sensitivity gain coefficient is greater than the regular gain coefficient. In this embodiment, the preset nighttime sensitivity gain coefficient is set to 1.5, and the preset regular gain coefficient is set to 1.0, to realistically reflect the high-risk clinical attributes of sudden physiological fluctuations under the background of low basal metabolic rate at night. No further limitations are imposed here. To simultaneously consider the specificity of the scenario, the urgency of time, and the severity of heart rate data deviation in risk assessment, a risk index is obtained for the suspected abnormal period based on analysis duration, analysis difference, and the time-segment gain coefficient corresponding to the suspected abnormal period. Analysis duration is negatively correlated with the risk index, while analysis difference and the time-segment gain coefficient are both positively correlated with the risk index. Therefore, the formula for calculating the risk index is: In the formula, F is the risk index; K is the time period gain coefficient; R is the analysis difference; and T is the analysis duration (in minutes). To pre-set a positive reference number and avoid a denominator of 0, its unit is consistent with the analysis duration; in this embodiment, it is set to 1 minute. This ensures that when the suspected anomaly occurs close to the target reference time (i.e., the analysis duration approaches 0), the algorithm can still stably output valid values, avoiding computational overflow. `norm` is the maximum-minimum normalization function; the corresponding dataset here is the dataset of each suspected anomaly period generated by the target object within the historical period. Set, if the above If the number of samples in the set is insufficient to support a statistical distribution (e.g., less than 5), the normalized risk index is directly set to 0.5 (neutral risk value). Dynamic normalization is then enabled once the sample size exceeds the threshold (5). To avoid the denominator of the normalization calculation being zero due to the maximum and minimum values ​​in the historical dataset being equal, a very small positive number (e.g., 0.0001) is introduced into the denominator of this normalization operation as a smoothing factor to ensure that the risk index can be stably mapped to... The interval provides a unified data input for subsequent cross-day risk aggregation.

[0033] Step S4: Aggregate the degree of abnormality and risk index of each suspected abnormal period within a single day to obtain the abnormal risk coefficient for that day; based on the magnitude and changes of the abnormal risk coefficient over several consecutive historical days, obtain the physiological state assessment value at the current moment.

[0034] Specifically, considering that the target subject may experience multiple physiological abnormalities of varying intensities within a single day, the indicators of a single suspected abnormal period cannot comprehensively represent the overall rehabilitation quality of a single day. To transform discrete suspected abnormal periods into a unified evaluation scale with a continuous time span (calculated on a daily basis), and further aggregate the degree of abnormality and risk index of each suspected abnormal period within a single day, an abnormal risk coefficient for that day is obtained, thereby achieving a comprehensive quantification of the daily rehabilitation pressure. A higher abnormal risk coefficient indicates more frequent physiological pain or more severe acute crises experienced by the target subject on that day, and a less ideal rehabilitation status. Simultaneously, considering that post-anesthesia recovery is a gradual process, an occasional abnormal risk coefficient is likely a temporary post-operative reaction, while a continuously rising abnormal risk coefficient over several days indicates a risk of transformation into chronic pain or serious complications. To capture this long-term potential risk evolution pattern, and based on the magnitude and changes of abnormal risk coefficients over several consecutive days, a physiological status assessment value for the current moment is obtained. This accurately reflects the overall safety margin of the target subject in the recent rehabilitation process and the probability of future pain crises, providing a global judgment basis that considers both the current state and future trends for the final follow-up resource allocation.

[0035] Preferably, in one feasible manner of this embodiment, the method for obtaining the abnormal risk coefficient is as follows: the average abnormality of all suspected abnormal periods within a single day is obtained as the overall abnormality analysis value, representing the average stable level of basic pain control of the target object within a single day; the maximum value of the risk index of each suspected abnormal period within a single day is used as the main risk analysis value to anchor the most serious acute risk event that occurs within a single day; it should be noted that if a suspected abnormal period spans two days, it is divided into two sub-periods with the midnight of each day as the dividing point, and the corresponding physiological data points and risk assessment weights are respectively included in the calculation sequence of the respective natural days to ensure the accuracy of the attribution of cross-day physiological risks in the time dimension; It is known that relying solely on overall anomaly analysis values ​​can easily miss fatal acute attacks, while relying solely on principal risk analysis values ​​fails to reflect long-term low-quality recovery. To balance background stability and extreme hazard in risk assessment, a daily anomaly risk coefficient is obtained based on both overall and principal risk analysis values; both overall and principal risk analysis values ​​are positively correlated with the anomaly risk coefficient. This embodiment uses the mean of the overall and principal risk analysis values ​​as the daily anomaly risk coefficient, ensuring that the final coefficient smooths out normal fluctuations while retaining sensitivity to extreme anomalies.

[0036] Preferably, in one feasible manner of this embodiment, the method for obtaining the physiological state assessment value is as follows: In order to assess the frequency of risk occurrence from the dimension of time density, this embodiment takes the dates with abnormal risk coefficients greater than a preset risk attention threshold from the historical consecutive days as attention days, and calculates the average interval between adjacent attention days as the interval assessment value; the larger the interval assessment value, the lower the frequency of significant physiological risks to the target object, and the more stable and safe the recovery status. It should be noted that if the number of attention days in the historical consecutive days is less than 2, making it impossible to calculate the adjacent interval, then a differentiated assignment is made based on the specific number of attention days: when the number of attention days is 0, the result of multiplying the number of days in the historical consecutive days by a preset safety gain coefficient (such as 1.5) is used as the interval assessment value; when the number of attention days is 1, the result of multiplying the number of days in the historical consecutive days by a preset safety gain coefficient (such as 1.5) is used as the interval assessment value; when the number of attention days is 1, the result of multiplying the number of days in the historical consecutive days by a preset safety gain coefficient (such as 1.5) is used as the interval assessment value. The number of days is used as the interval assessment value. Through the above-mentioned differentiation and assignment logic, it is ensured that the assessment system can accurately identify the state difference between absolute safety and low-frequency risk, avoid the phenomenon of numerical inversion of safety scores in the low-risk range, and further improve the statistical discrimination and clinical reference value of physiological state assessment values. In this embodiment, the preset risk attention threshold is set to 0.6 to ensure that only medium- and high-risk days with substantial impact are included in the frequency statistics, and to exclude the flood of attention days caused by slight daily fluctuations. Implementers can set the preset risk attention threshold according to the postoperative risk classification of the target object (such as high-risk surgery or routine surgery), which is not limited here. In this embodiment, the number of consecutive days in history is set to 7 days (i.e., the most recent week) to ensure that the data window can contain sufficient evolution samples and has strong timeliness. The end date of the consecutive days in history is the day before the current day. To predict future risk evolution from a dynamic perspective, a linear fit is performed on a sequence of historical abnormal risk coefficients over multiple consecutive days, arranged chronologically (e.g., using weighted least squares fitting, where a pre-defined weighting function assigns higher weights to dates closer to the current time, with weights increasing linearly with the number of days to ensure the fitting slope accurately captures the recent risk evolution of the target object and prevents the recent deterioration trend from being masked by stable data in the early monitoring period). This yields a fitting slope characterizing the trend of abnormal risk coefficient changes. A larger fitting slope (positive) indicates a daily deterioration of risk; a smaller fitting slope (negative) indicates a daily mitigation of risk. The weighted least squares method is a well-known technique and will not be elaborated further. To transform the fitting slope reflecting risk deterioration into an evaluation index positively correlated with safety, and to obtain a trend evaluation value based on the fitting slope, specifically: this embodiment limits the fitting slope to... Within the core interval (slopes outside this interval are truncated), an absolute slope value of 0.2 represents a daily average change in the risk coefficient within the monitoring period reaching a preset clinically significant proportion (e.g., 20%), indicating a trend reversal in the target subject's pain control status with significant clinical intervention value; and the slope of this interval is then inversely mapped to... The trend assessment value is defined as follows: when the fitted slope is greater than or equal to 0.2, the trend assessment value is 0; when the fitted slope is less than or equal to -0.2, the trend assessment value is 1. A larger trend assessment value indicates a weaker trend of risk deterioration (or accelerated recovery) and a higher expectation of future safety. It should be noted that if the number of days of concern over several consecutive historical periods is extremely small (e.g., less than 3 days), resulting in a lack of statistical significance in linear fitting, slope fitting is not performed, and the trend assessment value is directly set to a preset neutral value (e.g., 0.5), representing a stable trend. To integrate static risk sparsity with dynamic safety expectations, the product of interval assessment values ​​and trend assessment values ​​is normalized using a maximum-minimum normalization method, which serves as the physiological state assessment value. Here, the dataset corresponding to the maximum-minimum normalization method is the historical set of results calculated daily using the same method since the start of monitoring of the target object (product of interval assessment values ​​and trend assessment values). If the number of monitoring days is in the initial stage (e.g., less than 3 days), the physiological state assessment value is directly taken as a preset initial safety value (e.g., 0.8) to ensure a smooth transition of the system to dynamic monitoring. At the same time, to avoid the maximum and minimum product results in the historical dataset being equal, which would result in a normalization denominator of zero, a very small positive number (e.g., 0.0001) is added to the normalization denominator as a smoothing factor. The larger the final physiological state assessment value, the less high-risk events occur in the target object during the recent recovery process, and the more positive the overall risk trend is, indicating a safer physiological state. It should be noted that this embodiment constructs a physiological state assessment value that is positively correlated with the rehabilitation safety of the target subject through the composite calculation of the above-mentioned daily interval and risk change trend. This provides consistent mathematical logic support for the adaptive scheduling mechanism in subsequent steps where the lower the assessment value, the higher the follow-up frequency gain.

[0037] Step S5: Based on the physiological status assessment values, dynamically adjust the remote follow-up frequency of the target subjects.

[0038] Specifically, it is known that follow-up resources (such as medical staff and communication bandwidth) are limited, and the rehabilitation risk of different postoperative target subjects has a high degree of individual dynamic fluctuation. In order to build a dynamic balance mapping mechanism between the intensity of follow-up monitoring and the real-time individual rehabilitation risk, and to ensure that limited medical resources can prioritize covering high-risk groups whose risk is deteriorating, the remote follow-up frequency of target subjects can be dynamically adjusted based on physiological status assessment values. This allows the output of follow-up instructions to adaptively contract or expand with the fluctuation of physiological status assessment values, effectively achieving precise allocation of medical resources and avoiding over-treatment or lack of monitoring while ensuring rehabilitation safety.

[0039] Preferably, in one feasible method of this embodiment, the method for dynamically adjusting the remote follow-up frequency of the target object is as follows: when the physiological status assessment value is greater than or equal to the preset safety assessment threshold, it indicates that the target object's current rehabilitation status is in a safe and stable state, the frequency of physiological abnormalities is low, and the overall risk shows a stable or decreasing trend. At this time, the initial follow-up frequency is used as the current remote follow-up frequency of the target object to maintain basic rehabilitation monitoring. In this embodiment, the preset safety assessment threshold is set to 0.6 to ensure that the dynamic enhancement of follow-up resources is only for target objects with substantial risks. The implementer can set the preset safety assessment threshold according to the manpower load of the follow-up center or the surgical risk level of the target object, which is not limited here. The initial follow-up frequency is set by the staff based on experience and can be directly called. When the physiological status assessment value is lower than the preset safety assessment threshold, it indicates that the target subject has a high risk of pain control failure or a deteriorating physiological status. In order to detect and intervene in potential physiological crises in a timely manner by increasing the follow-up frequency, a target adjustment frequency is obtained based on the initial follow-up frequency, the preset safety assessment threshold, and the physiological status assessment value. The physiological status assessment value and the target adjustment frequency are negatively correlated. The formula for calculating the target adjustment frequency is: In the formula, Adjust the frequency to the target; This is the initial follow-up frequency; P represents the preset safety assessment threshold; P is the physiological state assessment value. A preset smoothing factor (e.g., 0.001) is used to prevent the logarithmic function calculation from failing when the target object is in an extremely high-risk state, resulting in a physiological state assessment value of 0, thus ensuring the computational stability of the algorithm under full-range operating conditions; ln is the natural logarithm function; To ensure the output follow-up frequency is an executable positive integer, a floor sign is used for rounding up. By introducing a preset safety assessment threshold as the numerator, the logarithmic gain term approaches 0 when the physiological state assessment value approaches the preset safety assessment threshold. This allows the target adjustment frequency to smoothly transition into the initial follow-up frequency, completely eliminating command jumps and system oscillations at the threshold boundary. Simultaneously, the logarithmic gain model ensures that the follow-up frequency increases smoothly with increasing physiological risk, effectively avoiding drastic frequency jumps caused by physiological noise in the low assessment value range. This improves the system's robustness under extreme risk conditions and the stability of clinical execution. Considering the maximum capacity of the follow-up center and the potential interference of frequent follow-ups with the target subject's postoperative rest, this embodiment sets a preset frequency limit of 4 times / day to ensure the clinical feasibility of the scheduling instructions. This ensures that the follow-up intensity remains within a reasonable range of human intervention even under extreme risk conditions. Implementers can set the preset frequency limit according to the follow-up center's scheduling system and the clinical standards of the medical institution; no restrictions are imposed here. When the target adjustment frequency is greater than the preset frequency limit, to prevent the algorithm from outputting redundant instructions that are out of touch with reality, the preset frequency limit is used as the target subject's current remote follow-up frequency to clamp and protect the scheduling results. When the target adjustment frequency is less than or equal to the preset frequency limit, to achieve precise risk hedging, the target adjustment frequency is used as the target subject's current remote follow-up frequency to achieve refined frequency scheduling.

[0040] In summary, this embodiment acquires multi-dimensional physiological data and medication administration times of the target subject in real time; determines the analysis period based on the degree of heart rate deviation, and obtains abnormal factors by combining the fluctuations and abnormal proportions of physiological data in each dimension, thereby determining the degree of abnormality in the analysis period and screening suspected abnormal periods; determines a reference time based on medication administration times and nighttime rest periods, and obtains a risk index by calculating the time interval and heart rate difference between suspected abnormal periods and the reference time; aggregates the degree of abnormality and the risk index for a single day to obtain a daily abnormal risk coefficient, and obtains a physiological state assessment value by combining the trend of abnormal risk coefficient changes over multiple historical days, thereby adaptively adjusting the frequency of remote follow-up. This invention achieves deep correlation analysis between physiological fluctuations and clinical background, improves the accuracy of postoperative follow-up, and can effectively optimize the allocation of medical resources.

[0041] Example 2: This invention also proposes a remote follow-up scheduling system based on physiological data analysis; please refer to [link / reference]. Figure 2 The diagram illustrates a remote follow-up scheduling system based on physiological data analysis provided by an embodiment of the present invention. The system includes: a data acquisition module 10, a suspected abnormal period acquisition module 20, a risk index acquisition module 30, a physiological state assessment value acquisition module 40, and a remote follow-up frequency adjustment module 50.

[0042] The data acquisition module 10 is used to acquire multi-dimensional physiological data and medication time of the target object in real time; among which, the physiological data includes heart rate data.

[0043] The suspected abnormal period acquisition module 20 is used to determine the analysis period based on the degree of deviation between the heart rate data under resting state and the preset baseline; based on the fluctuation and abnormality ratio of each dimension of physiological data in each analysis period, it acquires the abnormal factors corresponding to each dimension of physiological data in each analysis period; based on the abnormality of each dimension of physiological data at each moment in each analysis period and the abnormal factors, it acquires the degree of abnormality of each analysis period, and then filters out suspected abnormal periods.

[0044] The risk index acquisition module 30 is used to determine the reference time based on the medication time and the nighttime rest period; and to obtain the risk index for each suspected abnormal period based on the time interval between each suspected abnormal period and the nearest reference time and the difference in heart rate data.

[0045] The physiological state assessment value acquisition module 40 is used to aggregate the degree of abnormality and risk index of each suspected abnormal period within a single day to obtain the abnormal risk coefficient for the single day; and to obtain the physiological state assessment value at the current moment based on the magnitude and changes of the abnormal risk coefficient over several consecutive historical days.

[0046] The remote follow-up frequency adjustment module 50 is used to dynamically adjust the remote follow-up frequency of the target subject based on the physiological state assessment value.

[0047] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the remote follow-up scheduling system based on physiological data analysis and the remote follow-up scheduling method based on physiological data analysis provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiment, which will not be repeated here.

Claims

1. A remote follow-up scheduling method based on physiological data analysis, characterized in that, The method includes the following steps: Real-time acquisition of multi-dimensional physiological data and medication administration time of the target object; among which, physiological data includes heart rate data; The analysis period is determined based on the deviation of the resting heart rate data from the preset baseline; the abnormal factors corresponding to each dimension of physiological data in each analysis period are obtained based on the fluctuation and abnormality rate of each dimension of physiological data in each analysis period; the degree of abnormality of each analysis period is obtained based on the abnormality of each dimension of physiological data at each moment in each analysis period and the abnormal factors, thereby screening out suspected abnormal periods. Reference times are determined based on medication administration time and nighttime rest period; risk index for each suspected abnormal time period is obtained based on the time interval between each suspected abnormal time period and the nearest reference time period, as well as the difference in heart rate data. Aggregate the degree of abnormality and risk index of each suspected abnormal period within a single day to obtain the abnormal risk coefficient for that day; based on the magnitude and changes of the abnormal risk coefficient over multiple consecutive historical days, obtain the physiological state assessment value at the current moment. The frequency of remote follow-up for target subjects is dynamically adjusted based on physiological status assessment values.

2. The remote follow-up scheduling method based on physiological data analysis as described in claim 1, characterized in that, The method for obtaining the analysis period is as follows: In the resting state, the heart rate data at each moment is subtracted from the preset baseline and then normalized to obtain the degree of heart rate deviation at the corresponding moment. When the heart rate deviation exceeds the preset heart rate deviation threshold, the corresponding time will be marked as the analysis time. The time period consisting of consecutive adjacent analysis moments on the time axis is defined as the analysis period.

3. The remote follow-up scheduling method based on physiological data analysis as described in claim 1, characterized in that, The method for obtaining the abnormal factors is as follows: For any dimension of physiological data and any analysis period, the difference between the physiological data of that dimension and the baseline data at each moment within the analysis period is obtained as anomaly analysis data. When the abnormal analysis data exceeds the preset abnormal analysis threshold, the corresponding time will be marked as the abnormal analysis time of the physiological data in that dimension. The percentage of times an anomaly is analyzed within the total time frame of the analysis period is taken as the anomaly percentage. The fluctuation analysis value is obtained by subtracting the minimum value from the maximum value of the physiological data in this dimension within the analysis period and dividing the result by the total duration of the analysis period. The product of the abnormality rate and the normalized fluctuation analysis value is used as the abnormality factor corresponding to the physiological data of this dimension within the analysis period.

4. The remote follow-up scheduling method based on physiological data analysis as described in claim 3, characterized in that, The method for obtaining the degree of abnormality is as follows: For any moment within any analysis period, the physiological data corresponding to that moment as the abnormal analysis moment are all considered as abnormal physiological data, thereby determining the set of abnormal physiological data at that moment. Based on the overlap of abnormal physiological data sets at adjacent time points, time segments that are continuous on the time axis of the analysis period and whose abnormal physiological data sets are not empty are taken as comprehensive abnormal segments. The longest comprehensive abnormal segment is taken as the main segment, and the ratio of the number of abnormal physiological data dimensions contained in the main segment to the total number of physiological data dimensions is taken as the first analysis value; the duration of the main segment in this analysis period is taken as the second analysis value. Based on the first analysis value, the second analysis value, and the corresponding maximum anomaly factor within the analysis period, the degree of anomaly for that analysis period is obtained; wherein, the first analysis value, the second analysis value, and the maximum anomaly factor are all positively correlated with the degree of anomaly.

5. The remote follow-up scheduling method based on physiological data analysis as described in claim 1, characterized in that, The method for obtaining the suspected abnormal time period is as follows: When the degree of abnormality exceeds the preset abnormality threshold, the corresponding analysis period will be designated as a suspected abnormal period.

6. The remote follow-up scheduling method based on physiological data analysis as described in claim 1, characterized in that, The method for obtaining the reference time is as follows: For any medication administration time, the time corresponding to the preset drug onset time is taken as the first reference time, starting from that medication administration time and moving backward along the time axis; The entire day is divided into time according to the preset basic inspection cycle, and the time corresponding to each cycle node is obtained as the second reference time. The reference time is determined based on the first reference time and the second reference time.

7. The remote follow-up scheduling method based on physiological data analysis as described in claim 1, characterized in that, The method for obtaining the risk index is as follows: For any suspected abnormal time period, the reference time that is the closest to the start time of the suspected abnormal time period in terms of time sequence is taken as the target reference time; The interval between the target reference time and the start time of the suspected abnormal period is used as the initial duration; When the initial duration exceeds the preset correlation decay limit, the preset correlation decay limit will be used as the analysis duration. When the initial duration is less than or equal to the preset correlation attenuation limit, the initial duration will be used as the analysis duration. The difference between the mean of all heart rate data within the suspected abnormal period and the heart rate data at the target reference time is used as the basis for analysis. If the suspected abnormal time period falls within the nighttime rest period, the preset nighttime sensitivity gain coefficient will be used as the time period gain coefficient. If the suspected abnormal period does not fall within the nighttime rest period, the preset normal gain coefficient will be used as the period gain coefficient; among which, the preset nighttime sensitive gain coefficient is greater than the preset normal gain coefficient. Based on the analysis duration, analysis difference, and the time period gain coefficient corresponding to the suspected abnormal period, the risk index of the suspected abnormal period is obtained; among them, the analysis duration is negatively correlated with the risk index, while the analysis difference and the time period gain coefficient are both positively correlated with the risk index.

8. The remote follow-up scheduling method based on physiological data analysis as described in claim 1, characterized in that, The method for obtaining the abnormal risk coefficient is as follows: The average of the anomalies in all suspected abnormal periods within a single day is obtained as the overall anomaly analysis value. The maximum value of the risk index for each suspected abnormal period within a single day is used as the main risk analysis value; Based on the overall anomaly analysis value and the main risk analysis value, the daily anomaly risk coefficient is obtained; both the overall anomaly analysis value and the main risk analysis value are positively correlated with the anomaly risk coefficient.

9. The remote follow-up scheduling method based on physiological data analysis as described in claim 1, characterized in that, The method for obtaining the physiological state assessment values ​​is as follows: Dates with abnormal risk coefficients exceeding the preset risk attention threshold from a series of consecutive historical days are designated as attention days. The average interval between adjacent attention days is calculated and used as the interval assessment value. Linear fitting is performed on the sequence of abnormal risk coefficients over multiple consecutive historical days in chronological order to obtain the fitting slope that characterizes the trend of abnormal risk coefficient changes. A trend evaluation value is obtained based on the fitting slope; wherein, the fitting slope and the trend evaluation value are negatively correlated. The normalized result of the product of the interval assessment value and the trend assessment value is used as the physiological state assessment value.

10. The remote follow-up scheduling method based on physiological data analysis as described in claim 1, characterized in that, The method for dynamically adjusting the remote follow-up frequency of target subjects based on physiological state assessment values ​​is as follows: When the physiological status assessment value is greater than or equal to the preset safety assessment threshold, the initial follow-up frequency will be used as the current remote follow-up frequency for the target object. When the physiological status assessment value is less than the preset safety assessment threshold, the target adjustment frequency is obtained based on the initial follow-up frequency, the preset safety assessment threshold, and the physiological status assessment value; among them, the physiological status assessment value and the target adjustment frequency are negatively correlated. When the target adjustment frequency is greater than the preset frequency limit, the preset frequency limit will be used as the current remote follow-up frequency of the target object. When the target adjustment frequency is less than or equal to the preset frequency upper limit, the target adjustment frequency will be used as the current remote follow-up frequency of the target object.

Images