Patient whole course monitoring method based on personalized guidance
By collecting multi-dimensional data and using a segmented hidden Markov model to identify disease stages, a personalized transition stage monitoring plan is generated, which solves the problem of disconnected nursing intervention in traditional methods and improves the continuity and safety of nursing care throughout the entire disease course.
Patent Information
- Application Number
- CN202511150405.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-18
- Publication Date
- 2025-12-05
AI Technical Summary
Traditional methods of monitoring patients throughout the entire course of the disease have failed to effectively address the transitional stages of the disease process, leading to a disconnect in nursing interventions, an inability to adapt to individual patient differences, and an impact on the continuity and safety of nursing care throughout the entire course of the disease.
By collecting multi-dimensional data from patients, a segmented hidden Markov model is used to identify the current stage of the disease and predict the next stage, generating a personalized transition stage monitoring plan and dynamically adjusting nursing measures to adapt to the patient's physiological fluctuations and tolerance.
It enables personalized nursing guidance during the transition between disease stages, improves the continuity and safety of the entire disease management process, and avoids discontinuity and over-intervention in nursing care.
Smart Images
Figure CN121075705A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of disease course monitoring, in particular to a patient whole course monitoring method based on personalized guidance. BACKGROUND
[0002] Traditional patient whole course monitoring methods are mostly based on static stage division as the core, that is, according to clinical diagnosis standards, the disease process is mechanically divided into risk period, acute period, stable period and other independent stages, and fixed monitoring indicators and intervention schemes are formulated for each stage, for example, in diabetes management, patients in the stable period are uniformly subjected to blood glucose monitoring once or twice a week, and the medication is adjusted according to fixed thresholds; in tumor treatment, patients in the recovery period are only subjected to image review according to a fixed cycle, and in nursing, such methods also follow a stage solidification mode: patients in the acute period receive standardized bedside nursing processes, and patients in the recovery period adopt a unified functional exercise plan, and the formulation of nursing measures highly depends on general clinical pathways, lacking response to dynamic changes in individual nursing needs of patients, and nursing operations are mostly limited to task completion within a single stage, ignoring the connection of nursing focuses in the stage transition process.
[0003] However, the traditional method does not realize the particularity of the transition stage in the disease process, that is, the critical period of transition from the current stage to the next stage, in which the patient's physiological indicators fluctuate greatly, the disease stability is poor, and the sensitivity to intervention measures is significantly different from that in a single stage, due to the lack of precise definition and targeted monitoring of the transition stage, nursing intervention often leads to disconnection, such as when the acute period transitions to the stable period, the intensive nursing focused on rescuing the disease is not adjusted in time to progressive nursing guiding functional recovery, which may cause discomfort in patients due to an increase in activity too early, or when the recovery period transitions to the long-term follow-up period, the high-intensity in-hospital nursing guidance is still continued, and patients have difficulty adhering to the nursing scheme after returning home, resulting in a rebound in recovery effect, and due to the lack of consideration of the nursing tolerance differences of patients in the transition stage, the unified nursing monitoring and guidance scheme is difficult to adapt to individual needs, ultimately affecting the continuity and safety of whole course nursing. SUMMARY
[0004] Therefore, the present application proposes a patient whole course monitoring method based on personalized guidance, which can perform key monitoring based on the transition stage between the previous and subsequent course stages, thereby implementing a personalized nursing scheme adapted to the patient.
[0005] The technical scheme of the present application is implemented as follows:
[0006] The patient whole course monitoring method based on personalized guidance comprises the following steps:
[0007] Step S1, collecting physiological index data, behavior data, environmental exposure data and medical record data of the patient as multidimensional data in real time;
[0008] Step S2, determining the current course stage of the patient based on the multidimensional data by a preset stage recognition model;
[0009] Step S3, predicting the type of the next course stage and the end time of the current course stage according to the dynamic characteristics and historical data of the current course stage;
[0010] Step S4, determining the transition stage based on the current course stage and the next course stage, and generating a monitoring scheme for the transition stage;
[0011] Step S5, generating an individualized guidance scheme according to the monitoring data of the transition stage.
[0012] Preferably, the specific steps of step S1 include:
[0013] Step S11, collecting real-time vital signs, biochemical indicators, genomic data and neural electrical signals of the patient as physiological index data through wearable devices and implantable sensors;
[0014] Step S12, recording the daily activity amount, diet structure, sleep rhythm, medication compliance and social interaction frequency of the patient as behavior data through mobile terminals or Internet of Things devices;
[0015] Step S13, collecting air quality, noise intensity, illumination duration and temperature and humidity data of the environment where the patient is located as environmental exposure data through environmental sensors;
[0016] Step S14, connecting to a hospital database to extract the patient's diagnosis history, treatment history, operation record and past complications as medical record data;
[0017] Step S15, taking the physiological index data, behavior data, environmental exposure data and medical record data as multidimensional data.
[0018] Preferably, after collecting the physiological index data, behavior data, environmental exposure data and medical record data of the patient, the step S1 performs data cleaning, standardization processing, dimension reduction and time sequence alignment on the collected data.
[0019] Preferably, the specific steps of step S2 include:
[0020] Step S21, extracting the absolute value, dynamic change rate, rhythm pattern and abnormal fluctuation characteristics of the physiological indicators from the multidimensional data;
[0021] Step S22, input the extracted features into the pre-trained stage identification model, and output the current disease course stage of the patient by the stage identification model;
[0022] Step S23, calculate the confidence of the current disease course stage of the patient output by the stage identification model, and if the confidence is lower than a preset threshold, supplement specific data from the multi-dimensional data and rejudge.
[0023] Preferably, the specific steps of step S22 are as follows:
[0024] Step S221, define the disease course stage as a hidden state of a segmented hidden Markov model, and each hidden state corresponds to a state segment;
[0025] Step S222, obtain the multi-dimensional data of the historical patient, and form a historical observation sequence after processing, and train the model parameters including the initial probability of each hidden state, the transition probability within the state segment, the transition probability between the state segments, and the emission probability of the observation sequence by using the Baum-Welch algorithm;
[0026] Step S223, convert the features extracted from the multi-dimensional data into a real-time observation sequence, decode the real-time observation sequence by the segmented hidden Markov model, and output the disease course stage corresponding to the hidden state with the highest probability as the current disease course stage.
[0027] Preferably, the disease course stage includes a risk period, an early symptom period, an acute period, a stable period, a recovery period, and a long-term follow-up period.
[0028] Preferably, the specific steps of step S3 include:
[0029] Step S31, obtain the historical stage data of the historical patient in the same current disease course stage, perform trend analysis based on the historical stage data, and determine the type of the next disease course stage;
[0030] Step S32, determine the average duration of the current disease course stage based on the historical stage data, and confirm the time period in which the current disease course stage is located by the dynamic characteristics of the current disease course stage;
[0031] Step S33, predict the end time of the current disease course stage based on the average duration of the current disease course stage and the time period in which the current disease course stage is located.
[0032] Preferably, the specific steps of step S4 include:
[0033] Step S41, evaluate the importance of the next disease course stage in the whole disease course of the patient, and determine the range threshold based on the importance;
[0034] Step S42, taking the end time of the current disease course stage as an intermediate value, and combining the range threshold value to respectively intercept the end period of the current disease course stage and the start period of the next disease course stage as a transition stage;
[0035] Step S43, determining the key indicators of the transition from the current disease course stage to the next disease course stage, and generating a scheme for continuous monitoring of the key indicators in the transition stage.
[0036] Preferably, in the step S43 of generating the scheme for continuous monitoring of the key indicators in the transition stage, the risk coefficient of the transition stage is first evaluated, and the frequency of continuous monitoring is determined based on the risk coefficient.
[0037] Preferably, the specific steps of the step S5 include:
[0038] Step S51, analyzing the difference between the monitoring data of the transition stage and the target value, and determining the short-term control target of the transition stage and the preparation target of the next disease course stage;
[0039] Step S52, dynamically adjusting the drug dosage, rehabilitation training load and environmental adjustment instruction according to the short-term control target and the preparation target.
[0040] Compared with the prior art, the beneficial effects of the present application are:
[0041] The personalized guidance-based patient whole-disease-course monitoring method of the present application collects multi-dimensional data of the patient, intelligently judges the current disease course stage of the patient based on the multi-dimensional data through the introduction of a stage recognition model of artificial intelligence, predicts the type of the next disease course stage of the patient according to the current disease course stage, determines the transition stage according to the difference between the front and rear disease course stages, formulates a special monitoring scheme for the transition stage, and then the patient can perform targeted monitoring on the transition stage based on the formulated monitoring scheme when reaching the transition stage, and finally generates a personalized guidance scheme according to the monitoring data of the transition stage to guide the patient to adjust the corresponding medication method, lifestyle and rehabilitation method, etc., which can dynamically match the physiological fluctuations and nursing needs of the patient at the stage conversion, avoid disconnection or over-intervention, adapt to individual tolerance differences, and improve the continuity and safety of whole-disease-course management. BRIEF DESCRIPTION OF DRAWINGS
[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only preferred embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.
[0043] Figure 1Flowchart for the personalized guidance-based patient whole-course monitoring method of the present application.
[0044] Figure 2 Flowchart for step S1 of the personalized guidance-based patient whole-course monitoring method of the present application.
[0045] Figure 3 Flowchart for step S2 of the personalized guidance-based patient whole-course monitoring method of the present application.
[0046] Figure 4 Flowchart for step S22 of the personalized guidance-based patient whole-course monitoring method of the present application.
[0047] Figure 5 Flowchart for step S3 of the personalized guidance-based patient whole-course monitoring method of the present application.
[0048] Figure 6 Flowchart for step S4 of the personalized guidance-based patient whole-course monitoring method of the present application.
[0049] Figure 7 Flowchart for step S5 of the personalized guidance-based patient whole-course monitoring method of the present application. DETAILED DESCRIPTION
[0050] In order to better understand the technical content of the present application, a specific embodiment is provided below, and the present application is further described in conjunction with the accompanying drawings.
[0051] Reference Figures 1 to 7 The personalized guidance-based patient whole-course monitoring method provided by the present application comprises the following steps:
[0052] Step S1, real-time collection of physiological index data, behavior data, environmental exposure data and medical record data of a patient as multidimensional data;
[0053] Step S2, determination of the current course stage of the patient based on the multidimensional data through a preset stage recognition model;
[0054] Step S3, prediction of the type of the next course stage and the end time of the current course stage according to the dynamic characteristics and historical data of the current course stage;
[0055] Step S4, determination of a transition stage based on the current course stage and the next course stage, and generation of a monitoring scheme for the transition stage;
[0056] Step S5, generation of a personalized guidance scheme according to the monitoring data of the transition stage.
[0057] The personalized guidance-based patient whole-course monitoring method of the application is used for monitoring the whole course of a patient, including the monitoring of multiple course stages of the patient from the risk period to the recovery period, wherein the monitoring of the transition stage between the course stages is focused, after the real-time multi-dimensional data of the patient is obtained, the multi-dimensional data is identified through a preset stage identification model, so that the current course stage of the patient is judged, and is recorded as the current course stage, and the multi-dimensional data includes physiological index data of the patient, such as blood pressure, electrocardiogram signals and the like, behavior data includes behaviors of the patient in the current course stage, such as medication or diet, environmental exposure data is the environmental state of the patient, the environment will affect the current course stage of the patient, and the medical record data records the disease history and the recent detection state of the patient, and the stage identification model can quickly and accurately identify the current course stage of the patient according to the multi-dimensional data.
[0058] After the current course stage is determined, due to the differences of different diseases, the type of the next course stage also has differences, so it is necessary to combine the dynamic characteristics of the current course stage and historical data to predict the specific type of the next course stage, and the end time of the current course stage is also predicted, generally, after the current course stage ends, the next course stage will be entered immediately, the transition stage includes part of the time period of the current course stage and the next course stage, the transition stage can be divided by combining the current course period and the next course period, and a corresponding monitoring scheme can be generated for the transition stage, and then when the patient enters the transition stage, the monitoring data of the transition stage can be collected, and a personalized guidance scheme can be generated based on the monitoring data, which can dynamically match the physiological fluctuations and nursing needs of the patient in the stage conversion, avoid disconnection or over-intervention, adapt to individual tolerance differences, and improve the continuity and safety of whole-course management.
[0059] Preferably, the specific steps of the step S1 include:
[0060] Step S11, collecting real-time vital signs, biochemical indicators, genomic data and neural electrical signals of the patient as physiological index data through a wearable device and an implantable sensor;
[0061] Step S12, recording daily activity, diet structure, sleep rhythm, medication compliance and social interaction frequency of the patient as behavior data through a mobile terminal or an Internet of Things device;
[0062] Step S13, collecting air quality, noise intensity, illumination time and temperature and humidity data of the environment where the patient is located as environmental exposure data through an environmental sensor;
[0063] Step S14, access the hospital database to extract the patient's diagnosis medical record, treatment history, operation record and past complications as medical record data;
[0064] Step S15, the physiological index data, behavior data, environmental exposure data and medical record data as multi-dimensional data.
[0065] In collecting physiological index data, the physiological index data of the patient can be collected through wearable devices or implantable sensors, wherein real-time vital signs include heart rate, blood pressure, body temperature, etc., biochemical indicators include blood glucose, tumor markers, inflammatory factors, etc., and behavior data needs to rely on the patient's mobile terminal or corresponding Internet of Things devices to collect corresponding daily activity amount, dietary structure data, etc. In addition, environmental data can affect the patient's disease course stage, so environmental exposure data of the patient can be collected through several types of environmental sensors. Finally, the patient's disease history, medical record and other related medical record data can be directly extracted from the hospital database. Because the multi-dimensional data collected has different baselines, and the data may have missing or dimension overloading problems, after collecting the data, the collected data needs to be cleaned, standardized, dimensionally reduced and time-aligned. Data cleaning includes removing outliers, filling missing values, correcting data collection errors, etc. Standardization can convert data from different sources to a unified dimension. Through principal component analysis, core features can be extracted from high-dimensional data to reduce redundant information, and non-synchronous data can be calibrated on the time axis.
[0066] Preferably, the specific steps of step S2 include:
[0067] Step S21, extracting physiological index absolute value, dynamic change rate, rhythm pattern and abnormal fluctuation characteristics from multi-dimensional data;
[0068] Step S22, inputting the extracted features into the pre-trained stage recognition model to output the current disease course stage of the patient by the stage recognition model;
[0069] Step S23, calculating the confidence of the current disease course stage of the patient output by the stage recognition model, and if the confidence is lower than the preset threshold, supplementing specific data from the multi-dimensional data and rejudging;
[0070] After obtaining the multi-dimensional data, the multi-dimensional data is feature extracted, and the extracted features are input into the pre-trained stage recognition model to identify the current disease course stage of the patient. After identifying the current disease course stage, the confidence can be judged. If the confidence is lower than the preset threshold (such as 80%), rejudgment can be performed. In rejudgment, specific data such as targeted biochemical indicator detection data needs to be supplemented from the multi-dimensional data for judgment.
[0071] Preferably, the specific steps of the step S22 are as follows:
[0072] The step S221 defines six disease stages, including a risk period, an early symptom period, an acute period, a stable period, a recovery period and a long-term follow-up period, as hidden states of a segmented hidden Markov model, and each hidden state corresponds to a state segment.
[0073] The step S222 obtains multi-dimensional data of historical patients, processes the multi-dimensional data to form a historical observation sequence, and trains model parameters, including initial probabilities of the hidden states, transition probabilities within the state segments, transition probabilities between the state segments, and emission probabilities of the observation sequence, by using a Baum-Welch algorithm.
[0074] The step S223 converts features extracted from the multi-dimensional data into a real-time observation sequence, decodes the real-time observation sequence by using the segmented hidden Markov model, and outputs a disease stage corresponding to a hidden state with the highest probability as a current disease stage.
[0075] In the identification of the current disease stage, the segmented hidden Markov model is used, six disease stages are defined as hidden states, each hidden state corresponds to a state segment, historical observation sequences are converted from historical multi-dimensional data of patients, and then training is performed, the parameters of the training include initial probabilities of the hidden states, transition probabilities within the state segments, transition probabilities between the state segments, and emission probabilities of the observation sequence, after the training is completed, the multi-dimensional data of the patient is converted into a real-time observation sequence, and is input into the segmented hidden Markov model, the real-time observation sequence is decoded by using the segmented hidden Markov model, and a hidden state with the highest probability is obtained, and the disease stage corresponding to the hidden state is the current disease stage of the patient.
[0076] Preferably, the specific steps of the step S3 include:
[0077] The step S31 obtains historical stage data of historical patients in the same current disease stage, performs trend analysis based on the historical stage data, and determines a type of a next disease stage.
[0078] The step S32 determines an average duration of the current disease stage based on the historical stage data, and confirms a time period in which the current disease stage is located by using dynamic characteristics of the current disease stage.
[0079] The step S33 predicts an ending time of the current disease stage based on the average duration of the current disease stage and the time period in which the current disease stage is located.
[0080] After determining the current disease course stage, the data of previous patients are compared, and trend analysis is performed to determine which stage the previous patients will enter after the current disease course stage ends, so as to determine the next disease course stage of the patient. In addition, the time length of the patients with the same disease in the stage is averaged to obtain the average time length. Based on the time period of the patient in the current disease course stage and the average time length, the end time of the current disease course stage can be determined.
[0081] Preferably, the specific steps of step S4 include:
[0082] Step S41, evaluating the importance of the next disease course stage in the whole disease course of the patient, and determining the range threshold based on the importance;
[0083] Step S42, taking the end time of the current disease course stage as an intermediate value, and combining the range threshold to respectively cut off the end period of the current disease course stage and the start period of the next disease course stage as a transition stage;
[0084] Step S43, determining the key indicators of the transition from the current disease course stage to the next disease course stage, evaluating the risk coefficient of the transition stage, determining the frequency of continuous monitoring based on the risk coefficient, and generating a scheme for continuous monitoring of the key indicators in the transition stage.
[0085] The end time of the current disease course stage is the beginning of the next disease course stage, and the transition stage needs to include the end of the current disease course stage and the beginning of the next disease course stage. Taking the end time of the current disease course stage as an intermediate value, a certain period of time needs to be extended forward and backward as a transition period. Due to the differences in different diseases and combinations of previous and next disease course stages, the specific time length of the transition period also varies. After evaluating the importance of the next disease course stage in the whole disease course of the patient, the range threshold can be determined. Taking the end time of the current disease course stage as an intermediate value, the end of the current disease course stage and the beginning of the next disease course stage can be cut off respectively by combining the range threshold to obtain the transition stage. After obtaining the transition stage, the key indicators of the transition from the current disease course stage to the next disease course stage need to be determined, and then the key indicators need to be continuously monitored in the transition stage. The frequency of monitoring needs to be determined by the risk coefficient of the transition indicators, so the risk coefficient of the transition stage needs to be evaluated to determine the frequency of continuous monitoring.
[0086] Preferably, the specific steps of step S5 include:
[0087] Step S51, analyzing the difference between the monitoring data of the transition stage and the target value, and determining the short-term control target of the transition stage and the preparation target of the next disease course stage;
[0088] Step S52, dynamically adjusting the drug dosage, the rehabilitation training load and the environmental regulation instruction according to the short-term control target and the preparation target.
[0089] When the patient enters the transition stage, the patient can be continuously monitored based on the monitoring scheme, the monitoring data is compared with the target value, the difference is analyzed, the short-term control target of the transition stage and the preparation target of the next course stage are determined, and finally the individualized guidance scheme can be generated to guide the drug use, rehabilitation and environmental adjustment of the patient.
[0090] The above only describes the preferred embodiments of the present application and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A method for patient longitudinal monitoring based on personalized guidance, characterized in that, The method comprises the following steps: Step S1, collecting physiological index data, behavior data, environmental exposure data and medical record data of a patient as multidimensional data in real time; Step S2, determining the current course stage of the patient based on the multidimensional data through a preset stage recognition model; Step S3, predicting the type of the next course stage and the end time of the current course stage according to the dynamic characteristics and historical data of the current course stage; Step S4, determining a transition stage based on the current course stage and the next course stage, and generating a monitoring scheme for the transition stage; Step S5, generating an individualized guidance scheme according to the monitoring data of the transition stage.
2. The personalized coaching based patient longitudinal monitoring method of claim 1, wherein, The specific steps of the step S1 include: Step S11, collecting real-time vital signs, biochemical indicators, genomic data and neural electrical signals of the patient as physiological index data through wearable devices and implantable sensors; Step S12, recording the daily activity amount, diet structure, sleep rhythm, medication compliance and social interaction frequency of the patient as behavior data through mobile terminals or Internet of Things devices; Step S13, collecting air quality, noise intensity, illumination time and temperature and humidity data of the environment where the patient is located as environmental exposure data through environmental sensors; Step S14, connecting to a hospital database to extract the diagnosis medical record, treatment history, operation record and previous complications of the patient as medical record data; Step S15, taking the physiological index data, behavior data, environmental exposure data and medical record data as multidimensional data.
3. The personalized coaching based patient longitudinal monitoring method of claim 1, wherein, After collecting the physiological index data, behavior data, environmental exposure data and medical record data of the patient, the step S1 performs data cleaning, standardization processing, dimension reduction and time sequence alignment on the collected data.
4. The personalized coaching based patient longitudinal monitoring method of claim 1, wherein, The specific steps of the step S2 include: Step S21, extracting the absolute value, dynamic change rate, rhythm pattern and abnormal fluctuation characteristics of the physiological indicators from the multidimensional data; Step S22, inputting the extracted characteristics into a pre-trained stage recognition model, and outputting the current course stage of the patient by the stage recognition model; Step S23, calculating the confidence of the current course stage of the patient output by the stage recognition model, and re-determining after supplementing specific data from the multidimensional data if the confidence is lower than a preset threshold.
5. The personalized coaching based patient longitudinal monitoring method of claim 4, wherein, The specific steps of the step S22 are: Step S221, defining the course stage as a hidden state of a segmented hidden Markov model, and each hidden state corresponds to a state segment; Step S222, obtaining the multidimensional data of historical patients, and forming historical observation sequences after processing, and training model parameters including the initial probability of each hidden state, the transition probability within the state segment, the transition probability between state segments, and the emission probability of the observation sequence by using the Baum-Welch algorithm; Step S223, converting the characteristics extracted from the multidimensional data into real-time observation sequences, decoding the real-time observation sequences by the segmented hidden Markov model, and outputting the course stage corresponding to the hidden state with the highest probability as the current course stage.
6. The personalized coaching-based patient longitudinal monitoring method of claim 5, wherein, The course stage includes a risk period, an early symptom period, an acute period, a stable period, a rehabilitation period and a long-term follow-up period.
7. The personalized coaching based patient longitudinal monitoring method of claim 1, wherein, The specific steps of the step S3 include: Step S31, obtaining historical stage data of historical patients in the same disease course stage as the current disease course stage, performing trend analysis based on the historical stage data, and determining the type of the next disease course stage; Step S32, determining the average duration of the current disease course stage based on the historical stage data, and confirming the time period in which the current disease course stage is located through the dynamic characteristics of the current disease course stage; Step S33, predicting the end time of the current disease course stage based on the average duration of the current disease course stage and the time period in which the current disease course stage is located.
8. The personalized coaching based patient longitudinal monitoring method of claim 1, wherein, The specific steps of step S4 include: Step S41, evaluating the importance of the next disease course stage in the whole disease course of the patient, and determining the range threshold based on the importance; Step S42, taking the end time of the current disease course stage as the intermediate value, and combining the range threshold to cut off the end period of the current disease course stage and the start period of the next disease course stage as the transition stage respectively; Step S43, determining the key indicators of the transition from the current disease course stage to the next disease course stage, and generating a scheme for continuous monitoring of the key indicators in the transition stage.
9. The personalized coaching-based patient longitudinal monitoring method of claim 8, wherein, When generating the scheme for continuous monitoring of the key indicators in the transition stage, the risk coefficient of the transition stage is first evaluated, and the frequency of continuous monitoring is determined based on the level of the risk coefficient.
10. The personalized coaching based patient longitudinal monitoring method of claim 1, wherein, The specific steps of step S5 include: Step S51, analyzing the difference between the monitoring data of the transition stage and the target value, and determining the short-term control target of the transition stage and the preparation target of the next disease course stage; Step S52, dynamically adjusting the drug dose, rehabilitation training load and environmental adjustment instruction according to the short-term control target and the preparation target.