Multi-mode physical sign restraint strap remote monitoring system

Abstract

The invention discloses a multi-mode physical sign restraint strap remote monitoring system, and relates to the technical field of medical monitoring, the system is composed of a plurality of functional modules, and the system comprises a data acquisition module for acquiring monitoring data of a patient in real time, the monitoring data comprising physical sign data and medication data of the patient; the time sequence association construction module is used for preprocessing the monitoring data to obtain an effective physical sign data subset, performing dynamic time warping on the effective physical sign data subset based on a DTW algorithm to obtain an optimal time sequence anastomosis path, performing parameter estimation on the optimal time sequence anastomosis path by utilizing a linear regression model, and constructing a time sequence association trend of medicine-physical sign response; the intelligent analysis module is used for marking key nodes based on the time sequence correlation trend of the drug-physical sign response and performing multi-dimensional evaluation on the drug effect to obtain a drug effect evaluation value; and the drug use decision module is used for respectively carrying out transverse comparison and longitudinal analysis based on the drug effect evaluation value.

Description

Technical Field

[0001] This invention relates to the field of medical monitoring technology, specifically a multimodal vital sign restraint belt remote monitoring system. Background Technology

[0002] In the wave of information technology, the deep integration of the Internet, the Internet of Things, big data, and artificial intelligence has laid a solid technological foundation for reshaping the landscape of medical monitoring. Remote monitoring technology has emerged in response to this trend, using various intelligent sensors and portable monitoring devices to accurately capture patients' physiological status information such as heart rate, blood oxygen, and blood pressure. Relying on high-speed network links, the data is transmitted in real time to medical center terminals or mobile devices of medical staff, building a "virtual monitoring ward" that transcends space. This allows medical staff to overcome the limitations of physical distance and conduct remote, continuous, and dynamic monitoring of patients' vital signs. At the same time, the superposition of social development and the aging population is profoundly rewriting the ecosystem of demand for medical and health services. On the one hand, the elderly population continues to expand, and the rigid demand for chronic disease management and long-term care is growing rapidly. On the other hand, the awakening of health awareness and the iteration of lifestyles are giving rise to diversified demands for health prevention, personalized diagnosis and treatment, and continuous outpatient monitoring.

[0003] In traditional medical monitoring scenarios, data collection and management for agitated, suicidal, or mentally unstable patients consistently present significant challenges. These patients often exhibit confusion, agitation, or resistance, making it difficult for them to accurately report changes in their physical sensations and symptoms after medication. This results in fragmented descriptions of their condition, missing key information, and even data distortion, directly impacting healthcare professionals' judgment of drug efficacy and the precise adjustment of treatment plans, thus delaying the recovery process. Simultaneously, vital sign monitoring relies excessively on manual intervention: healthcare professionals need to conduct frequent close-range inspections and operate equipment, significantly increasing their workload and exacerbating manpower shortages. Repeated interventions can disrupt patients' already fragile routines, leading to irritability or resistance, further interfering with the stability of monitoring data. This double negative cycle becomes a prominent pain point in the monitoring of special patients, urgently requiring a monitoring solution that combines concealment, continuity, and intelligent analysis capabilities. This solution should reduce manual intervention while ensuring data authenticity and diagnostic accuracy, breaking through the inherent limitations of traditional models. Summary of the Invention

[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a multimodal vital sign restraint remote monitoring system, which solves the problems mentioned in the background art.

[0005] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: A multimodal vital sign restraint remote monitoring system, comprising: The data acquisition module collects patient monitoring data in real time, including the patient's vital signs and medication data. The time-series correlation construction module preprocesses the monitoring data to obtain a subset of effective vital signs data. Based on the DTW algorithm, it performs dynamic time warping on the subset of effective vital signs data to obtain the optimal time-series alignment path. Using a linear regression model, it estimates the parameters of the optimal time-series alignment path and constructs the time-series correlation trend of drug-vital sign response. The intelligent analysis module, based on the time-series correlation trend of drug-symptom response, marks key nodes and performs multi-dimensional evaluation of drug effect to obtain drug effect evaluation value; The medication decision-making module performs horizontal and vertical comparisons based on drug effect assessment values, and then implements dynamic dose adjustment strategies based on the comparison and analysis results.

[0006] Furthermore, the vital signs data include heart rate, blood oxygen, and body temperature; the medication data includes the type of drug, single dose, administration time, and route of administration.

[0007] Furthermore, the process of obtaining the effective subset of vital sign data is as follows: The min-max standardization method is used to standardize the vital signs data and map them to the [0,1] interval; for medication data, the original feature data is output directly. Collected vital signs data By analyzing the time series data and the medication time data, a time difference series is constructed. Based on the time difference series, vital sign data within a preset time window after medication are selected and correlation analysis is performed to obtain a valid data subset. .

[0008] Furthermore, the process of dynamically time-warping the subset of effective vital sign data based on the DTW algorithm is as follows: Obtain a subset of valid vital signs data Sequences of medication events and sequences of vital signs were constructed separately. A distance matrix is ​​constructed based on the drug use event sequence and the vital sign sequence. The optimal time-series matching path from the top-left corner (1,1) to the bottom-right corner (m,p) of the distance matrix is ​​found using dynamic programming. The k-th element of the path is... Given a path length of K, the optimal time-matching path is obtained. The formula is: in, For the k-th element Euclidean distance.

[0009] Furthermore, the process of estimating the parameters of the optimal temporal matching path is as follows: Optimal timing matching path Using the timestamp t of the time series as the dependent variable, a linear regression equation is constructed. Using the least squares method to evaluate the parameters and Perform estimation; there are m sets of sample data. , For timestamps, Given the DTW distance at the corresponding time, we obtain the minimum sum of squared residuals. ; Based on minimizing the sum of squared residuals Solve for parameters and ; in for The estimated value of the slope; for An estimate of the intercept.

[0010] Furthermore, the process of constructing the time-series correlation trend of drug-symptom response is as follows: based on The positive and negative values ​​of the correlation between medication and physical signs were used to monitor the time evolution trend of the correlation.

[0011] Furthermore, the specific process of annotating key nodes is as follows: Key nodes are automatically labeled using the first derivative in the temporal correlation trend of drug-symptom response.

[0012] Furthermore, the process of conducting a multi-dimensional assessment of drug effects is as follows: Based on the temporal correlation trend of drug-symptom response, a drug effect assessment model is constructed. This is achieved by determining the weights of assessment indicators and the duration of the effect. Peak effect The weighted summation yields the drug effect assessment value E.

[0013] Furthermore, the specific process of the horizontal comparison and vertical analysis is as follows: Horizontal comparison: Compare the E-values ​​of different drugs or different doses of the same drug, construct a distribution matrix of E-values ​​for drug types, and statistically analyze the mean, median, and standard deviation of E-values ​​for different sedatives; Longitudinal analysis: Observe the trend of E-value changes in the same patient after different doses to assess the patient's adaptation to the drug and the impact of changes in physical condition on drug response.

[0014] Furthermore, based on the drug effect assessment value E, a decision-making model is constructed by comparing drug effects horizontally and analyzing patient responses longitudinally, integrating multiple factors, calculating dosage adjustment amounts, generating suggestions, and dynamically providing feedback for optimization, thereby achieving precision medication.

[0015] (III) Beneficial Effects This invention provides a multimodal vital sign restraint remote monitoring system, which has the following beneficial effects: (1) This invention constructs a vital signs-drug effect correlation analysis system by constraining patients' needs for efficacy monitoring and medication regimen optimization after medication administration; it breaks through the limitations of traditional methods that only focus on vital sign monitoring, adds a drug recording module and effect analysis algorithm, deeply integrates patients' medication information with real-time vital sign data, and generates the temporal correlation trend of drug-vital sign response through temporal correlation analysis, providing accurate reference for individualized drug treatment, solving the problems of difficulty in accurately assessing efficacy after constrained patients' medication administration and lack of data support for dosage adjustment, and improving the safety and effectiveness of constrained patients' medication administration in scenarios such as emergency surgery.

[0016] (2) This invention incorporates patient medication information into the monitoring and analysis scope for the first time, and constructs a closed-loop monitoring system of "medication entry - sign monitoring - effect analysis - result feedback". By linking medication data and sign data, it breaks through the limitation of isolated analysis of sign data, and realizes the leap from simply focusing on the patient's physiological state to deeply exploring the process of drug influence on the patient's physiological state. It provides a more comprehensive and targeted reference for clinical medication decision-making and fills the technical gap that restricts the accurate monitoring of drug effects on patients. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the system of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: Please see Figure 1 This embodiment provides a multimodal vital sign restraint remote monitoring system, the detection system comprising: The data acquisition module collects patient monitoring data in real time, including the patient's vital signs and medication data. Multimodal sensors are embedded in the restraint straps, making them indistinguishable from ordinary restraint straps. They provide real-time, covert monitoring of patients' vital signs, preventing patients from noticing or being deliberately interfered with, thus ensuring the authenticity and reliability of the data. The real-time collection of patients' vital sign data is transmitted to the central monitoring platform at the nurse station via Bluetooth Low Energy or wireless network, and synchronized to mobile terminals, allowing medical staff to view patients' vital signs at any time. Vital sign data includes monitoring of key vital signs such as heart rate, blood oxygen, and body temperature. The patient data collected in real time by multimodal sensors is combined into vital sign data, which is represented as... .

[0020] When restricting patients' medication, medical staff enter the patient's medication data through the central monitoring platform interface at the nurse station, including the type of drug, single dose, time of administration, and route of administration (such as oral or intravenous injection). Finally, the medication operations entered by the patient are combined into medication data, which is represented as follows: .

[0021] The time-series correlation construction module preprocesses the monitoring data to obtain a subset of effective vital signs data. Based on the DTW algorithm, it performs dynamic time warping on the subset of effective vital signs data to obtain the optimal time-series alignment path. Using a linear regression model, it estimates the parameters of the optimal time-series alignment path and constructs the time-series correlation trend of drug-vital sign response. The process of preprocessing monitoring data: After acquiring real-time vital signs and medication data, the data is time-aligned and standardized to obtain a subset of effective vital signs data.

[0022] Acquire the collected vital signs data medium time series Medication data The time for taking traditional Chinese medicine is Construct time difference series Based on time difference series The vital signs data within a preset time window after medication were selected, and correlation analysis was performed to obtain a valid subset of data. .

[0023] Vital signs data within a preset time window after medication administration: From the time of drug administration, relevant data reflecting the effect of the drug on vital signs are collected within a specific time interval pre-set according to clinical experience, drug characteristics, etc., and used to accurately analyze the relationship between drug use and vital signs.

[0024] Specifically, using the standard timestamps of the hospital information system as a benchmark, the time series data of vital signs are analyzed. and medication time Alignment was performed; vital sign data were standardized using the min-max standardization method, mapping the vital sign data to the [0,1] interval to facilitate unified analysis between different vital sign indicators; for medication data, the original features were directly used in sequence construction and distance calculation. Dynamic time warping of a subset of valid vital sign data based on the DTW algorithm: S201: Obtain a subset of valid vital sign data Sequences of medication events and sequences of vital signs were constructed separately. Medication data is feature-encoded to construct a medication event sequence. ; in, This represents the j-th drug use characteristic; For example, for the use of sedatives, the dosage can be... Medication time coding (with the time of medication as 0 o'clock, coded by minute) and other features are used as characteristics, such as , ,in, This is the coded value for medication time; Acquire vital sign data, encode features, and construct vital sign sequences. ; in, Indicates the characteristic of the p-th individual; For example: using standardized heart rate sequences For example: If we analyze the correlation between blood oxygen and medication, the formula would be: .

[0025] S202: Distance matrix construction: Build Distance matrix (Taking the correlation between heart rate and medication as an example, when analyzing blood oxygen saturation...) ), calculate drug characteristics Physical characteristics The Euclidean distance is The formula is: Each element in the distance matrix quantifies the degree of difference between medication characteristics and physical signs at the corresponding time point, providing a basis for finding the optimal matching path in the future.

[0026] S203: Calculation of the optimal timing matching path: The optimal time-series matching path from the top-left corner (1,1) to the bottom-right corner (m,p) of the distance matrix is ​​found using dynamic programming; let the k-th element of the path be... Given a path length of K, the optimal time-matching path is obtained. The formula is: in, For the Kth element The Euclidean distance; k is the path length; It should be noted that the smaller the DTW distance, the closer the correlation between the medication sequence and the vital sign sequence. By calculating the DTW distance, the medication sequence and the vital sign sequence can be flexibly aligned on the time axis to uncover the potential patterns of vital sign changes after medication. Even when the two are not synchronized in terms of time series length and rhythm, the correlation can be accurately analyzed.

[0027] Parameter estimation for the optimal time-matching path: Optimal timing matching path Using the timestamp t of the time series as the dependent variable and the timestamp t of the time series as the independent variable, a linear regression equation is constructed, with the following formula: in, The intercept term represents the initial time. The degree of basic correlation between the drug administration sequence and the vital sign sequence; The slope reflects the rate of change of the DTW distance over time (a negative value indicates that the correlation increases over time, while a positive value indicates that it decreases). The error term follows a normal distribution. .

[0028] Parameter estimation: Using the least squares method to evaluate the parameters and Perform estimation; there are m sets of sample data. , For timestamps, For the DTW distance at the corresponding time, calculate the minimum residual sum of squares. The formula is: in, The intercept term represents the basic correlation between the medication sequence and the vital signs sequence at the initial time t=0. The slope reflects the rate of change of the DTW distance over time. Solve for the parameters: in, , , for The estimated value of the slope; for An estimate of the intercept.

[0029] Perform time-series correlation trend analysis: Using the above linear regression model, dynamic correlation monitoring is performed: based on The positive or negative judgment is used to determine the trend of the correlation between medication and physical signs (such as whether the correlation gradually increases after medication). Response feature extraction: combining intercept Quantifying the initial correlation strength helps determine the immediate effect of drug efficacy; Multifactor expansion: Drug dosage, type, etc. can be further incorporated into the model as covariates to construct a multiple linear regression equation and comprehensively analyze the temporal correlation mechanism of drug-symptom response.

[0030] The intelligent analysis module, based on the time-series correlation trend of drug-symptom response, marks key nodes and performs multi-dimensional evaluation of drug effects; Mark key nodes: Key nodes are automatically labeled on the time-series correlation trend of drug-symptom response. The first derivative of the curve is calculated. Find the moment when the heart rate drops the fastest. (when At time 0, if the slope is constant, it represents the entire process; if actual data fluctuations are considered, it can be determined through local extremum analysis. Calculate the area under the curve (AUC) to mark the point where blood oxygen fluctuations stabilize. This means that when the standard deviation of blood oxygen levels is less than a preset threshold for 5 consecutive minutes... (like When the data is standardized, the fluctuation is considered stable, and the corresponding time point is the stable moment. Based on the temporal correlation trend of drug-symptom response, a drug effect assessment model is constructed, and specific drug effect assessment values ​​are obtained through the following steps: Standardized analysis system for multimodal vital signs parameters: Determine the weights of the evaluation indicators: based on the effective time. Duration of the effect Peak effect Key indicators are assigned weights, and the sum of all weights is 1, such as... , , .

[0031] Standardized processing indicators: Standardization of onset time: Setting a standard onset time If the actual time to take effect Less than The score is calculated based on the onset time, using the following formula: Conversely, the formula is: Duration of effect standardization: based on the clinically desired duration of drug effect. Based on the actual duration Greater than The duration score is calculated using the following formula: Conversely, the formula is: Peak effect standardization: Setting standard values ​​for peak effect based on different physical characteristics. The actual peak effect value is The peak effect score formula is: Among them, when Greater than (time), or (when Less than hour).

[0032] Calculate the drug effect assessment value: Using the weighted summation formula, the formula is: The final drug effect assessment value E is obtained, with the value ranging from 0 to 1. A higher value indicates a better drug effect.

[0033] The medication decision-making module performs horizontal and vertical comparisons based on drug effect assessment values, and implements dynamic dose adjustment strategies based on the comparison and analysis results. Based on the obtained drug effect assessment value E, the following analysis was performed: Horizontal comparison: Comparing the E values ​​of different drugs or different doses of the same drug to intuitively judge the quality of drug efficacy.

[0034] Longitudinal analysis: Observe the trend of E-value changes in the same patient after different doses to assess the patient's adaptability to the drug and the impact of changes in physical condition on drug response; Horizontal comparison: Construct a distribution matrix of E-values ​​for drug types, and statistically analyze the mean, median, and standard deviation of E-values ​​for different sedatives (such as drug A and drug B).

[0035] For example, calculate the E value for 50 patients using drug A (5mg dose) and 50 patients using drug B (5mg dose): If drug A... Drug B Furthermore, since the standard deviations of both are less than 0.15, it can be preliminarily determined that drug A is more effective at this dosage.

[0036] The temporal correlation trend of drug-sign response was used to assist in the verification; the heart rate decrease rate curves of the two drugs were compared: if the absolute value of the slope of the curve corresponding to drug A was greater... Batch / minute·mg, The result (times / minute·mg) indicates that drug A has a faster onset of action, which aligns with the conclusions drawn from the comparison with the E value, thus enhancing the credibility of the decision.

[0037] Comparison of different dosages of the same drug: Taking a certain sedative as an example, E-value data for three dosage groups (5mg, 7.5mg, and 10mg) were extracted, and dose-E trend curves were plotted. The curve equation was fitted using polynomial regression, and the formula is as follows: If the fitting result shows and This indicates that the E value first increases and then decreases with increasing dose, and there is an optimal dose range, such as when the E value reaches its peak of 0.82 at 7.5mg.

[0038] The dose-E relationship was validated by combining clinical indicators such as the incidence of restraint conflict. If the E value decreased in the 10mg group while the incidence of restraint conflict increased from 12% in the 7.5mg group to 20%, it can be determined that the 10mg dose carries a risk of excessive sedation, thus helping to determine the "effective and safe" dose range. Longitudinal analysis: Calculate the rate of change of E value The fluctuation range of the E-value quantifies the dynamic characteristics of the patient's response to the drug; the trend of the E-value changing with the number of drug administrations is observed. If patient B is using the drug for the first time... The second dose (3 days apart) The third dose (2 days apart) Furthermore, the fluctuation range decreased from 0.21 to 0.08, indicating that the patient's tolerance to the medication improved and the body gradually established a stable response; events that changed the patient's physical condition (such as removal of the drainage tube on the 3rd day after surgery and the onset of lung infection on the 5th day) were extracted and marked as time nodes. Comparing the changes in E-values ​​before and after the event: If, after patient C's lung infection at time T, the E-value decreases from 0.7 to 0.55, and the amplitude of heart rate fluctuations increases by 30%, it can be determined that the infection leads to a decrease in the body's ability to metabolize and respond to drugs.

[0039] A rule base for the association between "state change - E-value impact" is constructed. The mapping relationship between key state factors (such as infection and increased pain scores) and changes in E-value is mined through decision tree algorithm. The output rule is "lung infection occurs → E-value decreases by 15%-20% and heart rate fluctuation increases by 25%", which helps medical staff predict changes in drug efficacy.

[0040] Decision model construction, multi-factor weighted decision: The decision-making model constructs a dynamic dose adjustment mechanism by integrating cross-sectional and longitudinal analysis results. The model defines three key decision indicators and their weights: The weight of drug efficacy differences in horizontal comparison is 0.4. This indicator focuses on the differences in E-values ​​between different drugs, as well as the changes in the slope of the dose-E-value curve. The individual patient response weight in the longitudinal analysis was 0.3, mainly considering the rate of change of the patient's E value and the influence coefficient of physical status on the drug response; The clinical risk control weight is 0.3, with a focus on assessing potential clinical risks such as constraint conflict risk and respiratory depression risk.

[0041] The formula for calculating the dose adjustment ΔD is as follows: ΔD = (β1 × standardized value of difference in cross-sectional comparison effect + β2 × standardized value of individual response characteristics + β3 × standardized value of risk control) × current dosage Dcurrent Where β1, β2, and β3 are the corresponding weighting coefficients.

[0042] Strategy output and execution process, dose adjustment suggestion generation: When a horizontal comparison shows that the E value of drug A is 20% higher than that of drug B, and a longitudinal analysis shows that the patient adapts well to drug A (the rate of change in E value is positive), and risk control indicators such as the respiratory depression risk score are at a low level, the model will output a suggestion to "increase the dosage of drug A by 10%-15%".

[0043] Develop detailed adjustment rules to address individual differences: For patients with large fluctuations in E value, a "small step adjustment + high frequency monitoring" strategy is adopted, that is, the dose is adjusted by 5% each time, and the E value and vital signs are monitored every 15 minutes after the adjustment. If a patient's condition deteriorates (e.g., due to infection), the "maintenance dose + adjunctive intervention" regimen is recommended as the first choice, including combined physical sedation and optimized infection treatment.

[0044] Closed-loop feedback mechanism: After dosage adjustment, the E-value and related vital signs are continuously tracked, and the input parameters of the decision model are updated every 30 minutes. If the adjusted E-value does not meet expectations (e.g., the E-value only increases by 5% after increasing the dosage), a secondary decision-making process is automatically triggered to reassess the drug type, dosage combination, and patient status, ensuring that the adjustment strategy dynamically adapts to the patient's needs.

[0045] By comparing drugs horizontally to clarify the differences in efficacy between drugs and dosages, and analyzing individual patient response patterns vertically, the medication decision-making module outputs dynamic dosage adjustment strategies based on a multi-factor decision-making model. This module achieves deep integration from data mining to clinical action, providing a quantifiable and traceable decision-making path for precise medication management that constrains patients. It enhances drug efficacy while maximizing patient safety and comfort, making it a key link in realizing the clinical value of the multimodal vital sign restraint belt remote monitoring system.

[0046] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0047] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0048] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A multimodal vital sign restraint remote monitoring system, characterized in that: The system includes: The data acquisition module collects patient monitoring data in real time, including the patient's vital signs and medication data. The time-series correlation construction module preprocesses the monitoring data to obtain a subset of effective vital signs data. Based on the DTW algorithm, it performs dynamic time warping on the subset of effective vital signs data to obtain the optimal time-series alignment path. Using a linear regression model, it estimates the parameters of the optimal time-series alignment path and constructs the time-series correlation trend of drug-vital sign response. The intelligent analysis module, based on the time-series correlation trend of drug-symptom response, marks key nodes and performs multi-dimensional evaluation of drug effect to obtain drug effect evaluation value; The medication decision-making module performs horizontal and vertical comparisons based on drug effect assessment values, and then implements dynamic dose adjustment strategies based on the comparison and analysis results.

2. The multimodal vital sign restraint remote monitoring system according to claim 1, characterized in that: The monitoring data is collected by an embedded multimodal sensor and transmitted in real time to the central monitoring platform at the nurse station via Bluetooth Low Energy or wireless network, and synchronized to the mobile terminal. When restricting patients' medication use, medical staff enter the patients' medication data through the central monitoring platform interface at the nurses' station.

3. The multimodal vital sign restraint remote monitoring system according to claim 2, characterized in that: The process of obtaining a subset of effective vital sign data is as follows: The min-max standardization method is used to standardize the vital signs data and map them to the [0,1] interval; for medication data, the original feature data is output directly. Collected vital signs data By analyzing the time series data and the medication time data, a time difference series is constructed. Based on the time difference series, vital sign data within a preset time window after medication are selected and correlation analysis is performed to obtain a valid data subset. .

4. The multimodal vital sign restraint remote monitoring system according to claim 3, characterized in that: The process of dynamically time-warping the effective vital sign data subset based on the DTW algorithm is as follows: Obtain a subset of valid vital signs data Sequences of medication events and sequences of vital signs were constructed separately. A distance matrix is ​​constructed based on the drug use event sequence and the vital sign sequence. The optimal time-series matching path from the top-left corner (1,1) to the bottom-right corner (m,p) of the distance matrix is ​​found using dynamic programming. The k-th element of the path is... Given a path length of K, the optimal time-matching path is obtained. The formula is: in, For the k-th element Euclidean distance.

5. A multimodal vital sign restraint remote monitoring system according to claim 4, characterized in that: The process of estimating the parameters of the optimal temporal matching path is as follows: Optimal timing matching path Using the timestamp t of the time series as the dependent variable, a linear regression equation is constructed. Using the least squares method to evaluate the parameters and Perform estimation; there are m sets of sample data. , For timestamps, Given the DTW distance at the corresponding time, we obtain the minimum sum of squared residuals. ; Based on minimizing the sum of squared residuals Solve for parameters and ; in, for The estimated value of the slope; for An estimate of the intercept.

6. A multimodal vital sign restraint remote monitoring system according to claim 5, characterized in that: The process of constructing the time-series correlation trend of drug-sign response is as follows: Based on the time evolution trend of the correlation between positive and negative judgments on the degree of correlation between medication and vital signs, dynamic monitoring of the correlation is carried out to obtain the time-series correlation trend.

7. A multimodal vital sign restraint remote monitoring system according to claim 6, characterized in that: The specific process for annotating key nodes is as follows: Key nodes are automatically labeled using the first derivative in the temporal correlation trend of drug-symptom response.

8. A multimodal vital sign restraint remote monitoring system according to claim 1, characterized in that: The process of multidimensionally evaluating drug effects is as follows: Based on the temporal correlation trend of drug-symptom response, a drug effect assessment model is constructed. This is achieved by determining the weights of assessment indicators and the duration of the effect. Peak effect The weighted summation yields the drug effect assessment value E.

9. A multimodal vital sign restraint remote monitoring system according to claim 1, characterized in that: The specific process of the horizontal comparison and vertical analysis is as follows: Horizontal comparison: Compare the E values ​​of different drugs or different doses of the same drug, construct a distribution matrix of E values ​​for drug types, and statistically analyze the mean, median, and standard deviation of E values ​​for different sedatives; Longitudinal analysis: Observe the trend of E-value changes in the same patient after different doses to assess the patient's adaptation to the drug and the impact of changes in physical condition on drug response.

10. A multimodal vital sign restraint remote monitoring system according to claim 9, characterized in that: The specific process of the dynamic dose adjustment strategy based on comparison and analysis results is as follows: Based on the drug effect assessment value E, a decision model is constructed by comparing drug effects horizontally and analyzing patient responses longitudinally, integrating multiple factors, calculating dosage adjustment amounts, generating suggestions, and dynamically providing feedback for optimization.

Images