Warfarin dose real-time decision-making method based on multi-factor dynamic analysis
By employing a multi-factor dynamic analysis method, the accuracy and interpretability issues in warfarin dosage adjustment are addressed, providing real-time, interpretable dosage decisions suitable for emergencies and first-time dosing, and can be run on mobile devices, reducing the risks associated with empirical dosing.
Patent Information
- Application Number
- CN202511493870.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-20
- Publication Date
- 2026-03-03
AI Technical Summary
Existing technologies for warfarin dosage adjustment suffer from low accuracy, poor individual adaptability, poor model interpretability, strong data dependence, high computational resource requirements, and lag, making it difficult to provide real-time, interpretable decision-making in emergency situations and for first-time medication.
Using a multi-factor dynamic analysis method, by collecting and preprocessing basic treatment data, the short-term trend of INR is calculated. Different dose averaging strategies and trend fitting methods are used to provide real-time decision-making for warfarin dosage, including emergency treatment, recovery mode and standard adjustment mode, which are suitable for emergency situations and first-time dosing.
It enables dosage decisions without prior training, has high interpretability and real-time response capabilities, allows for fine-tuning of dosage, is suitable for mobile devices, and reduces the risks of empirical medication.
Smart Images
Figure BDA0005643114530000031 
Figure BDA0005643114530000051 
Figure FDA0005643114520000021
Abstract
Description
Technical Field
[0001] This invention belongs to the field of biomedical technology, and in particular relates to a method for real-time warfarin dosage decision based on multi-factor dynamic analysis. Background Technology
[0002] Warfarin is an oral anticoagulant widely used to prevent thrombotic diseases. Its therapeutic window is narrow, with significant individual variability, and its efficacy is easily affected by various factors such as concomitant medications, diet, and disease state. Therefore, patients need frequent monitoring of their international normalized ratio (INR) and dose adjustments based on the results to balance the risks of thrombosis and bleeding.
[0003] The existing technical solutions mainly include:
[0004] Fixed-formula method: The initial or maintenance dose is calculated using a linear formula based on a limited set of clinical parameters (such as age and weight). This method ignores dynamic changes and complex influencing factors in patients, resulting in low accuracy and poor individual adaptability.
[0005] Traditional machine learning prediction models extract patient features through deep learning networks and perform time-series analysis to predict whether INR testing is needed and to adjust dosage. These methods have the following limitations:
[0006] Poor model interpretability: As a "black box" model, its decision-making basis (why to increase / decrease dosage) is difficult to explain to doctors and patients, affecting clinical trust.
[0007] Strong data dependency: The model relies on a large amount of high-quality, well-labeled historical data for training, and is sensitive to missing data and noise. Its performance will degrade in scenarios with insufficient data.
[0008] Lag: The model's predictions are based on historical patterns and lack the ability to respond and make decisions in real time for first-time medication, drastic dosage changes, or abnormal emergency situations.
[0009] High computational resource requirements: Complex deep learning models require high computational resources, which is not conducive to deployment on mobile devices or resource-constrained environments.
[0010] Therefore, there is an urgent need for a warfarin dosage decision scheme that can respond in real time, has clear rules, is highly interpretable, and does not rely on training with a large amount of historical data. Summary of the Invention
[0011] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for real-time warfarin dosage decision based on multi-factor dynamic analysis. The method of this invention does not require pre-training and can perform calculations and decisions immediately after receiving new data, making it particularly suitable for handling emergency situations and initial calculations for first-time medication.
[0012] To achieve the above-mentioned objectives, the technical solution provided by this invention patent is as follows:
[0013] A method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis, the method specifically includes the following steps:
[0014] S1, Data Acquisition: Collecting patients' basic treatment data and preprocessing the basic treatment data;
[0015] S2 performs persistent storage, serialization, and deserialization of basic treatment data; and records the basic treatment data into an array based on time sorting.
[0016] S3, based on the acquired treatment data, makes real-time decisions on warfarin dosage and provides specific warfarin dosage results.
[0017] Furthermore, the basic treatment data includes the INR target range setting, the INR test results for each test, the date of each INR test, and the warfarin dosage for each test.
[0018] Furthermore, the record array includes treatment date, INR value, target range, and current recommended dose.
[0019] Furthermore, the real-time warfarin dosage decision-making specifically includes:
[0020] The real-time warfarin dosage decision-making process specifically involves:
[0021] Based on the acquired record array, the short-term trend of the patient's INR is calculated. The interval between two adjacent INR measurements is recorded and analyzed. Different dose averaging strategies are adopted according to different intervals. When the interval between two adjacent INR measurements is less than or equal to three days, a short interval strategy is adopted, and the trend is directly calculated by combining recent change characteristics. When the interval between two adjacent INR measurements is between four and fourteen days, a medium interval strategy is adopted, and the moving weighted average method is used to assign differentiated weights to the daily trends within the interval period, and the weighted average is used as the effective trend. When the interval between two adjacent INR measurements is greater than fourteen days, a long interval strategy is adopted, and a trend line fitting method is used. Statistical methods such as linear regression are used to fit the pattern of trend changes over time within the interval period.
[0022] Based on the current INR value and historical INR data, different decision sub-processes are entered, and the recommended warfarin dose for the specific patient in this examination is calculated.
[0023] Furthermore, the short-term trend of the patient's INR includes the direction of INR change, the slope of INR change, the proportion of INR change, the number of consecutive increases in INR, and the number of consecutive high INR counts.
[0024] Furthermore, the specific calculation of the recommended warfarin dosage for the patient in this examination is as follows:
[0025]
[0026] Wherein, Bt is the recommended warfarin dose for the patient in this examination; At is the INR test result for this examination; ΔAt is the change in the INR test result compared to the previous result, including the rate of change and the ratio of change; f(ΔAt) is the medication coefficient, which is obtained by fitting ΔAt and historical trends; g(At) is the fitting coefficient for the upward trend; h(At) is the fitting coefficient for the downward trend; Mt is the interval day calculation strategy; II[a,b](At) is the range indicator function; and d is the number of days between two adjacent INR tests.
[0027] Based on the above technical solution, the warfarin dosage real-time decision-making method based on multi-factor dynamic analysis proposed in this invention has achieved the following technical advantages through practical application:
[0028] 1. The present invention provides a method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis. It does not require prior training and can perform calculations and decisions immediately after receiving new data. It is particularly suitable for handling emergency situations and initial calculations for first-time medication.
[0029] 2. The present invention provides a method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis, which not only considers the static value of INR, but also focuses on its dynamic change characteristics, making dosage adjustment more forward-looking and smooth, and avoiding drastic fluctuations based on single-point data. Detailed Implementation
[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the invention is described below through specific examples. However, it should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.
[0031] Example
[0032] A method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis, the method specifically includes the following steps:
[0033] S1, Data Acquisition: Collecting patients' basic treatment data and preprocessing the basic treatment data;
[0034] S2 performs persistent storage, serialization, and deserialization of basic treatment data; and records the basic treatment data into an array based on time sorting.
[0035] S3, based on the acquired treatment data, makes real-time decisions on warfarin dosage and provides specific warfarin dosage results.
[0036] The basic treatment data includes the INR target range setting, the INR test results for each test, the date of each INR test, and the warfarin dosage for each test.
[0037] The record array includes treatment date, INR value, target range, and current recommended dose.
[0038] The real-time warfarin dosage decision-making process specifically involves:
[0039] Based on the acquired record array, the short-term trend of the patient's INR is calculated. The interval between two adjacent INR measurements is recorded and analyzed. Different dose averaging strategies are adopted according to different intervals. When the interval between two adjacent INR measurements is less than or equal to three days, a short interval strategy is adopted, and the trend is directly calculated by combining recent change characteristics. When the interval between two adjacent INR measurements is between four and fourteen days, a medium interval strategy is adopted, and the moving weighted average method is used to assign differentiated weights to the daily trends within the interval period, and the weighted average is used as the effective trend. When the interval between two adjacent INR measurements is greater than fourteen days, a long interval strategy is adopted, and a trend line fitting method is used. Statistical methods such as linear regression are used to fit the pattern of trend changes over time within the interval period.
[0040] Based on the current INR value and historical INR data, different decision sub-processes are entered, and the recommended warfarin dose for the specific patient in this examination is calculated.
[0041] The decision-making sub-process includes emergency handling mode, recovery mode, and standard adjustment mode.
[0042] Emergency treatment mode: For patients with high-risk factors for severe bleeding, rapidly reduce the risk of bleeding; no fine-tuning required.
[0043] Recovery mode: Provides a refined, stepwise dosage adjustment plan for patients who have discontinued medication, safely guiding them to resume anticoagulation therapy after discontinuation and reducing the risks of empirical medication;
[0044] Standard adjustment mode: Under normal circumstances, the dosage is finely adjusted according to the trend to achieve long-term stability.
[0045] The specific calculation of the recommended warfarin dosage for the patient in that particular examination was as follows:
[0046]
[0047] Wherein, Bt is the recommended warfarin dose for the patient in this examination; At is the INR test result for this examination; ΔAt is the change in the INR test result compared to the previous result, including the rate of change and the ratio of change; f(ΔAt) is the medication coefficient, which is obtained by fitting ΔAt and historical trends; g(At) is the fitting coefficient for the upward trend; h(At) is the fitting coefficient for the downward trend; Mt is the interval day calculation strategy; II[a,b](At) is the range indicator function; and d is the number of days between two adjacent INR tests.
[0048] The method of the present invention has the following characteristics:
[0049] 1. High interpretability and clinical fit: Every decision is based on clear, clinically logical rules; doctors and patients can clearly understand the reasons for dosage adjustments, enhancing the credibility and acceptability of the system; this is different from the "black box" decision-making of deep learning models.
[0050] 2. Powerful real-time response capability: No pre-training is required; calculations and decisions can be made immediately upon receiving new data, making it particularly suitable for handling emergency situations and initial calculations for first-time medication.
[0051] 3. Refined dynamic trend analysis: It not only looks at the static value of INR, but also focuses on its dynamic change characteristics, making dose adjustment more forward-looking and smooth, and avoiding drastic fluctuations based on single-point data.
[0052] 4. Low resource dependence and high robustness: It does not rely on large-scale labeled datasets and GPU computing power, and can run efficiently on mobile devices such as smartphones and tablets. It has a higher tolerance for missing data.
[0053] 5. Personalized recovery logic: The unique "recovery mode" design can guide patients to restart anticoagulation therapy more safely and rationally after stopping medication, reducing the risks of empirical medication.
[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them; although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific implementation of the invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the technical solutions claimed in the present invention.
Claims
1. A method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis, characterized in that, The method specifically includes the following steps: S1, Data Acquisition: Collecting patients' basic treatment data and preprocessing the basic treatment data; S2 performs persistent storage, serialization, and deserialization of basic treatment data; The basic treatment data is recorded in an array and sorted by time. S3, based on the acquired treatment data, makes real-time decisions on warfarin dosage and provides specific warfarin dosage results.
2. The method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis according to claim 1, characterized in that, The basic treatment data includes the INR target range setting, the INR test results for each test, the date of each INR test, and the warfarin dosage for each test.
3. The method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis according to claim 1, characterized in that, The record array includes treatment date, INR value, target range, and current recommended dose.
4. The method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis according to claim 1, characterized in that, The real-time warfarin dosage decision-making process specifically involves: Based on the acquired record array, the short-term trend of the patient's INR is calculated. The interval between two adjacent INR measurements is recorded and analyzed. Different dose averaging strategies are adopted according to different intervals. When the interval between two adjacent INR measurements is less than or equal to three days, a short interval strategy is adopted, and the trend is directly calculated by combining recent change characteristics. When the interval between two adjacent INR measurements is between four and fourteen days, a medium interval strategy is adopted, and the moving weighted average method is used to assign differentiated weights to the daily trends within the interval period, and the weighted average is used as the effective trend. When the interval between two adjacent INR measurements is greater than fourteen days, a long interval strategy is adopted, and a trend line fitting method is used. Statistical methods such as linear regression are used to fit the pattern of trend changes over time within the interval period. Based on the current INR value and historical INR data, different decision sub-processes are entered, and the recommended warfarin dose for the specific patient in this examination is calculated.
5. The method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis according to claim 4, characterized in that, The short-term trend of the patient's INR includes the direction of INR change, the slope of INR change, the proportion of INR change, the number of consecutive increases in INR, and the number of consecutive high INR counts.
6. The method for real-time warfarin dosage decision-making based on multi-factor dynamic analysis according to claim 4, characterized in that, The specific calculation of the recommended warfarin dosage for the patient in that particular examination was as follows: Wherein, Bt is the recommended warfarin dose for the patient in this examination; At is the INR test result for this examination; ΔAt is the change in the INR test result compared to the previous result, including the rate of change and the ratio of change; f(ΔAt) is the medication coefficient, which is obtained by fitting ΔAt and historical trends; g(At) is the fitting coefficient for the upward trend; h(At) is the fitting coefficient for the downward trend; Mt is the interval day calculation strategy; II[a,b](At) is the range indicator function; and d is the number of days between two adjacent INR tests.