A method for predicting the volume status and diuretic resistance status of a heart failure patient

Abstract

The application discloses a kind of prediction method for fluid volume status and diuretic resistance state of heart failure patient, it is related to medical instrument and diagnostic technical field. Including: collecting the original data of heart failure patient;Original data is carried out data preprocessing;The original data after preprocessing is input into the fluid volume assessment model of heart failure patient, according to the data index of heart failure patient, two categories are distinguished: fluid volume is insufficient and fluid volume is not insufficient;Heart failure patient of fluid volume is insufficient, directly output diagnosis result and corresponding visit suggestion;Fluid volume is not insufficient for heart failure patient, carries out diuretic resistance detection;The data index of heart failure patient is input into the diuretic resistance identification model of heart failure patient, according to the data index of heart failure patient, two categories are distinguished: diuretic resistance and diuretic non-resistance;Finally, output diagnosis result and corresponding visit suggestion.The application can predict the fluid volume status and diuretic resistance state of patient in real time and accurately.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical devices and diagnostic techniques, and more particularly, to a method for predicting the fluid volume status and diuretic resistance status of a heart failure patient. BACKGROUND

[0002] In the treatment of heart failure (HF) patients, accurate assessment of fluid volume status and identification of diuretic resistance are key steps. Fluid volume abnormalities that contribute to worsening of HF include volume overload, inadequate intake, or overuse of diuretics leading to volume depletion, and abnormal distribution of fluid volume leading to insufficient effective circulating volume. Accurate determination of these three states is the basis for precise volume management. There are many indicators for the determination of volume in the clinic, such as skin elasticity, tongue fur dryness and wetness, extremity dryness and wetness, blood pressure, heart rate, central venous pressure, capillary wedge pressure, inferior vena cava width and collapse rate, lung ultrasound AB line, chest X-ray congestion sign, hematocrit, urine specific gravity, serum lactic acid, straight leg raising test, etc. However, there are very few assessments of fluid volume status in heart failure patients. These indicators are both related and relatively independent, and a single indicator cannot accurately reveal the volume status of the patient. There is currently a lack of objective evaluation of systematic standards for the selection and comprehensive application of these indicators to derive the accurate volume status of the patient.

[0003] Volume overload, or water and sodium retention, is the most common volume abnormality in heart failure patients and is an important cause of disease progression and symptoms. To control blood volume load, more than 90% of heart failure patients currently receive loop diuretic therapy. However, the effect of loop diuretics is affected by many factors such as dose, route of administration, time of administration, concomitant diseases and related treatments, and 20%-30% of volume overload heart failure patients have weakened or disappeared diuretic effect before the goal of reducing volume is achieved, and diuretic resistance occurs. The physiological mechanism of diuretic resistance involves multiple systems and aspects, such as changes in drug metabolism kinetics, drug interactions, liver and kidney function abnormalities, hyponatremia, and renal unit remodeling. At present, the response to diuretic resistance is also based on the patient's clinical characteristics to speculate the possible mechanism and take different methods. Such as avoiding the use of drugs that affect the diuretic effect, increasing the dose of diuretics, changing the mode of action of diuretics, combined use of diuretics with different target sites, application of drugs that increase plasma osmotic pressure, use of drugs that improve renal perfusion and increase diuretic sensitivity, and even invasive ultrafiltration and dehydration therapy. Because patients have different clinical characteristics and different mechanisms of diuretic resistance, the above strategies for patients with diuretic resistance may not be effective, or even ineffective. The current clinical approach is essentially a trial-and-error method, and if some methods are not effective, other methods will be used, and if they are not effective, they will be replaced, which often delays patient treatment. The reason for such a clinical approach is that the individual mechanism of diuretic resistance is not well understood, and the patient's full characteristics and the type of diuretic resistance cannot be accurately determined. There is currently no precise classification of heart failure diuretic resistance in the existing technology.

[0004] At present, there is very little research on fluid volume management in heart failure patients and early identification scoring systems for diuretic resistance. There is no clear single indicator to accurately indicate the volume status of patients, and there is a lack of systematic and objective evaluation criteria. Therefore, the reasonable construction of the fluid volume status evaluation model and the early identification scoring system for diuretic resistance can fill this gap. The present application aims to solve the technical problems of identifying diuretic resistance and fluid volume status in heart failure patients. In the context of precision medicine and artificial intelligence technology, the present application proposes an intelligent patient state discrimination system for quickly identifying the fluid volume status and diuretic resistance status of heart failure patients. The existing technology has the following main problems in the assessment of fluid volume status and the identification of diuretic resistance in heart failure patients:

[0005] (1) Strong subjective dependence: clinical assessment and inquiry methods depend on the experience of doctors and the subjective description of patients, which is prone to misjudgment;

[0006] (2) Single data: the data provided by laboratory tests and imaging techniques is relatively single, and cannot comprehensively and dynamically reflect the fluid volume status of patients;

[0007] (3) Technical threshold is high: many advanced monitoring methods (such as echocardiography, implantable device) have high technical requirements for operators, and the equipment cost is expensive, which is difficult to popularize in primary medical institutions;

[0008] (4) Lack of real-time performance: most of the existing methods cannot realize real-time monitoring, and cannot timely reflect the dynamic changes of the patient's fluid volume state;

[0009] (5) Compliance problem: patients have low compliance to monitoring equipment in daily life, which affects the continuity and accuracy of data.

[0010] Therefore, a method for predicting the fluid volume state and diuretic resistance state of heart failure patients is proposed to solve the problems existing in the prior art, which is a problem that needs to be solved by those skilled in the art. SUMMARY

[0011] Therefore, the present application provides a method for predicting the fluid volume state and diuretic resistance state of heart failure patients, which develops an intelligent diagnosis system by analyzing the clinical data of patients, for predicting the fluid volume state of heart failure patients, and identifying the diuretic tolerance of patients with insufficient fluid volume.

[0012] In order to achieve the above purpose, the present application provides the following technical scheme:

[0013] A method for predicting the fluid volume state and diuretic resistance state of heart failure patients, comprising the following steps:

[0014] Collecting the original data of heart failure patients;

[0015] Data preprocessing of the original data;

[0016] The preprocessed original data is input into the fluid volume evaluation model of heart failure patients, and two categories are distinguished according to the data indicators of heart failure patients: insufficient fluid volume and non-insufficient fluid volume;

[0017] The heart failure patients with insufficient fluid volume directly output the diagnosis results and the corresponding treatment suggestions;

[0018] The heart failure patients with non-insufficient fluid volume are detected for diuretic resistance;

[0019] The data indicators of heart failure patients are input into the diuretic resistance identification model of heart failure patients, and two categories are distinguished according to the data indicators of heart failure patients: diuretic resistance and non-diuretic resistance;

[0020] Finally, the diagnosis results and the corresponding treatment suggestions are output for heart failure patients with diuretic resistance and non-diuretic resistance.

[0021] The method, optionally, the original data of the heart failure patient includes: medical history data, physical sign data, blood test data and heart function data.

[0022] The method, optionally, the medical history data includes: heart failure course, hospitalization record, drug use;

[0023] The physical sign data includes: weight, blood pressure, heart rate, urine volume;

[0024] The blood test data includes: blood test index, and the blood test index includes: electrolyte level, kidney function index and heart marker;

[0025] The heart function data includes: ejection fraction EF and cardiac output CO.

[0026] The method, optionally, the preprocessing includes: missing value filling and standardization.

[0027] The method, optionally, the missing value filling includes: mean filling and interpolation.

[0028] The standardization includes: standardizing each data index, so that the mean of the data is 0 and the standard deviation is 1.

[0029] The method, optionally, the specific content of inputting the original data after preprocessing into the heart failure patient fluid volume evaluation model is:

[0030] Using the intelligent diagnosis system, inputting each data index of the heart failure patient on the user interface, after inputting, clicking the fluid volume prediction button, the input data index is quickly analyzed, and the machine learning model is used to predict the fluid volume state of the heart failure patient, and the prediction result is displayed on the interface.

[0031] The method, optionally, the specific content of inputting the data index of the heart failure patient into the heart failure patient diuretic resistance identification model is:

[0032] If the heart failure patient is in the state of non-insufficient fluid volume, using the intelligent diagnosis system, after inputting the data index on the user interface, clicking the diuretic tolerance prediction button, the input data index is quickly analyzed, and the LASSO regression model is used for analysis, the tolerance of the heart failure patient to the diuretic is provided, and the prediction result is displayed on the interface.

[0033] The method, optionally, the consultation suggestion includes: further test suggestion, adjustment of the existing treatment scheme suggestion, and provision of alternative treatment scheme suggestion.

[0034] Compared with the prior art, the application provides a prediction method for the fluid volume state and diuretic resistance state of a heart failure patient, which has the beneficial effects that:

[0035] (1) Improving the diagnostic accuracy:

[0036] The application constructs a fluid volume evaluation model and a diuretic resistance identification model for heart failure patients. By analyzing the clinical data and laboratory indicators of the patients, the fluid volume state and diuretic resistance state of the patients can be accurately predicted, significantly improving the diagnostic accuracy of heart failure patients. The clinical data of the patients, including medical history, physical signs, blood tests and other information, can be quickly and efficiently processed, realizing rapid and accurate prediction of the fluid volume state and diuretic resistance state of the patients, greatly shortening the diagnosis time and improving the efficiency of clinical decision-making.

[0037] (2) Reducing medical costs:

[0038] By accurately predicting the fluid volume state and diuretic resistance of heart failure patients, the application can avoid unnecessary waste of medical resources, such as unnecessary hospitalization and frequent re-examination, thereby reducing medical costs.

[0039] (3) Accelerating clinical decision-making:

[0040] Based on artificial intelligence machine learning technology, the patient's data can be analyzed and predicted quickly and in real time, and presented on the doctor's interface in real time. This greatly accelerates the process of clinical decision-making, enabling doctors to make accurate diagnosis and treatment decisions more quickly, improving medical efficiency.

[0041] (4) Improving patient treatment effect:

[0042] By accurately predicting the fluid volume state and diuretic resistance of heart failure patients and adjusting the treatment plan accordingly, the patient's condition can be effectively controlled, the occurrence of complications can be reduced, and the patient's treatment effect and quality of life can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only embodiments of the application, and those skilled in the art can obtain other drawings according to the provided drawings without creative labor.

[0044] Figure 1 A flowchart of a prediction method for the fluid volume state and diuretic resistance state of a heart failure patient provided by the application is provided.

[0045] Figure 2 A whole flow chart of a prediction method for fluid volume status and diuretic resistance status of heart failure patients is provided in the present application;

[0046] Figure 3 A principle diagram of a fluid volume evaluation model for heart failure patients is provided in the present application;

[0047] Figure 4 A principle diagram of a diuretic resistance identification model for heart failure patients is provided in the present application. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be apparently and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0049] Referring to Figure 1 and Figure 2 , the present application discloses a prediction method for fluid volume status and diuretic resistance status of heart failure patients, comprising the following steps:

[0050] Collecting original data of heart failure patients;

[0051] Data preprocessing is performed on the original data;

[0052] The preprocessed original data is input into a fluid volume evaluation model for heart failure patients, and two categories are distinguished according to data indexes of heart failure patients: fluid volume insufficiency and fluid volume non-insufficiency;

[0053] For heart failure patients with fluid volume insufficiency, a diagnosis result and corresponding treatment suggestions are directly output;

[0054] For heart failure patients with fluid volume non-insufficiency, diuretic resistance detection is performed;

[0055] The data indexes of heart failure patients are input into a diuretic resistance identification model for heart failure patients, and two categories are distinguished according to the data indexes of heart failure patients: diuretic resistance and diuretic non-resistance;

[0056] Finally, diagnosis results and corresponding treatment suggestions are output for heart failure patients with diuretic resistance and diuretic non-resistance.

[0057] Specifically, referring to Figure 3 and Figure 4For the fluid volume assessment model of heart failure patients, the mRMR feature selection method is used to screen features, and then a machine learning model is used to determine the fluid volume state of the patient. Second, for the diuretic resistance identification model of heart failure patients, the LASSO regression model is used for analysis to distinguish the diuretic tolerance state of the patient.

[0058] Further, the original data of heart failure patients include: medical history data, physical sign data, blood test data and heart function data.

[0059] Further, the medical history data includes: heart failure course, hospitalization record, drug use;

[0060] Physical sign data includes: weight, blood pressure, heart rate, urine volume;

[0061] Blood test data includes: blood test indicators, including: electrolyte levels, kidney function indicators, cardiac markers;

[0062] Heart function data includes: ejection fraction EF, cardiac output CO.

[0063] Specifically, electrolyte levels such as sodium, potassium, and chloride, kidney function indicators such as serum creatinine and urea nitrogen, and cardiac markers such as brain natriuretic peptide, BNP.

[0064] Further, the preprocessing includes: filling missing value processing and standardization processing.

[0065] Further, the filling of missing values includes: using mean filling, interpolation method to fill in;

[0066] Standardization processing includes: standardizing each data indicator, making the mean of the data 0 and the standard deviation 1.

[0067] Further, the specific content of inputting the preprocessed original data into the fluid volume assessment model of heart failure patients is:

[0068] Using the intelligent diagnosis system, input the various data indicators of heart failure patients on the user interface, after input is completed, click the fluid volume prediction button, the input data indicators will be quickly analyzed, and the machine learning model will be used to predict the fluid volume state of heart failure patients, the prediction result is displayed on the interface.

[0069] Specifically, to provide reference for doctors, help doctors make more intelligent diagnosis and treatment decisions.

[0070] Further, the specific content of inputting the data indicators of heart failure patients into the diuretic resistance identification model of heart failure patients is:

[0071] If the heart failure patient is in a state of non-insufficient fluid volume, the intelligent diagnostic system can be used. After inputting the data indicators on the user interface and clicking the diuretic tolerance prediction button, the input data indicators will be quickly analyzed and analyzed using the LASSO regression model. The tolerance of the heart failure patient to diuretics is provided. The prediction results are displayed on the interface.

[0072] Specifically, the tolerance of diuretics provides doctors with deeper insights, helping them adjust treatment plans and ensure more accurate and effective treatment for heart failure patients. The entire process not only simplifies complex data analysis steps, but also significantly improves the speed and accuracy of diagnosis, providing strong support for clinical decision-making.

[0073] Further, the consultation recommendations include further laboratory recommendations, recommendations for adjusting existing treatment plans, and recommendations for providing alternative treatment plans.

[0074] Specifically, after the above process is completed, the detailed prediction results of the specific fluid volume state and diuretic tolerance state of the heart failure patient will be obtained. Based on these prediction results, the system will also generate corresponding diagnostic recommendations to provide valuable reference information for clinicians.

[0075] Finally, a comprehensive report will be seen on the interface, detailing the fluid volume state, diuretic tolerance state, and diagnostic recommendations generated by the system for heart failure patients. This report provides doctors with comprehensive references and helps develop more accurate and effective treatment plans, thereby improving the treatment effect and quality of life of heart failure patients.

[0076] In a specific embodiment:

[0077] 1. Data collection:

[0078] First, collect the original data of heart failure patients, including the following vital signs, physical examination and laboratory indicators, including:

[0079] Urine sodium (mmol / 24h), urine sodium concentration (mmol / L), body weight / kg, urine Na / K, daily intake, FENa, urine potassium (mmol / 24h), TnT (admission), serum creatinine, NTproBNP (admission), urine protein (mg / 24h), RV (mm), thrombolytic dimer, albumin (admission), platelets, lung rales, primary diagnosis, urine creatinine (umol / 24h), diabetes, cystatin C, hematocrit, potassium, total bile acids, chloride, cholinesterase, B-type natriuretic peptide precursor, urinary ketone bodies, infusion volume + oral intake, urine volume (ml), lower extremity edema, high-sensitivity troponin T, uric acid, urea, lung wet rales, lactate dehydrogenase, hypertension, hydroxybutyrate dehydrogenase, cerebrovascular disease, mitochondrial isozyme, aspartate aminotransferase, creatine kinase, oliguria, high-sensitivity C-reactive protein, pulmonary artery systolic pressure, free triiodothyronine, atrial fibrillation / flutter, supine, alanine aminotransferase, dyspnea, gender, absolute eosinophil count, lung dry rales, a total of 52 features.

[0080] 2. Data preprocessing:

[0081] The data preprocessing is performed, and the steps include:

[0082] Missing value filling: for the missing values existing in the data set, mean filling, interpolation method or other suitable methods are used for filling.

[0083] Data standardization: standardization processing is performed on each data index, so that the mean of the data is 0 and the standard deviation is 1, thereby eliminating the dimension difference between different indexes.

[0084] 3. Body fluid volume evaluation model:

[0085] After the data preprocessing is completed, the processed data is input into the heart failure patient body fluid volume evaluation model:

[0086] Feature input: the mRMR (Minimum Redundancy Maximum Relevance) feature selection method is used to screen out 30 features most relevant to the body fluid volume state, including cholinesterase, B-type natriuretic peptide precursor, urinary ketone bodies, infusion volume + oral intake, urine volume (ml), lower extremity edema, high-sensitivity troponin T, uric acid, urea, lung wet rales, lactate dehydrogenase, hypertension, hydroxybutyrate dehydrogenase, cerebrovascular disease, mitochondrial isozyme, aspartate aminotransferase, creatine kinase, oliguria, high-sensitivity C-reactive protein, pulmonary artery systolic pressure, free triiodothyronine, platelets, atrial fibrillation / flutter, supine, alanine aminotransferase, dyspnea, gender, thrombolytic dimer, absolute eosinophil count, lung dry rales.

[0087] Patient classification: input the above 30 feature indicators into the machine learning model to judge the patient's fluid volume status. Output two categories: fluid volume deficiency and non-fluid volume deficiency.

[0088] 4. Fluid volume deficiency processing:

[0089] For patients determined to be fluid volume deficient, the system directly outputs the diagnosis result and the corresponding treatment suggestion, including: fluid replacement treatment plan, dietary adjustment suggestion, regular review plan.

[0090] 5. Fluid volume non-deficiency processing:

[0091] For patients determined to be fluid volume non-deficient, the system further performs diuretic resistance detection:

[0092] Feature input and result output: LASSO (Least Absolute Shrinkage and Selection Operator) regression is used for feature selection and model training, and 24 feature indicators are selected, including urine sodium (mmol / 24h), urine sodium concentration (mmol / L), body weight / kg, urine Na / K, daily intake, FENa, urine potassium (mmol / 24h), TnT (admission), serum creatinine, NTproBNP (admission), urine protein (mg / 24h), RV (mm), thrombolytic dimer, albumin (admission), platelets, lung rales, main diagnosis, urine creatinine (umol / 24h), diabetes, cystatin C, hematocrit, potassium, total bile acid, chloride. Input the above index values, and output two categories of patients according to the LASSO regression model result: diuretic resistance and non-diuretic resistance.

[0093] 6. Diuretic resistance processing:

[0094] For patients determined to be diuretic resistant, the system outputs the diagnosis result and the corresponding treatment suggestion, including: adjusting the type or dose of diuretics, adding other drug treatment, regular monitoring and evaluation of treatment effect.

[0095] 7. Diuretic non-resistance processing:

[0096] For patients determined to be diuretic non-resistant, the system outputs the diagnosis result and the corresponding treatment suggestion, including: continuing the current diuretic treatment, regularly monitoring and evaluating the patient's fluid status, adjusting other related treatment measures.

[0097] Each embodiment in the specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts between each embodiment can be referred to each other.

[0098] The foregoing description of the disclosed embodiments enables a person skilled in the art to make or use the application. Modifications of these embodiments will occur to persons of skill in the art, and that the appended claims are intended to cover all such modifications that do not depart from the true spirit and scope of the application. Therefore, the application is not limited to the embodiments shown but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting fluid volume status and diuretic resistance status in patients with heart failure, characterized in that, Includes the following steps: Collect raw data from heart failure patients; Perform data preprocessing on the raw data; The preprocessed raw data was input into the fluid volume assessment model for heart failure patients, and two categories were identified based on the data indicators of heart failure patients: fluid volume insufficiency and non-fluid volume insufficiency. For heart failure patients with insufficient body fluid volume, the prediction results and corresponding medical advice are directly provided. Heart failure patients who are not fluid volume deficient should undergo diuretic resistance testing. The data indicators of heart failure patients are input into the heart failure patient diuretic resistance identification model, and two categories are distinguished based on the data indicators of heart failure patients: diuretic resistance and diuretic non-resistance; Finally, for heart failure patients who are resistant to diuretics and those who are not, predictive results and corresponding medical advice are provided.

2. The method for predicting fluid volume status and diuretic resistance status in patients with heart failure according to claim 1, characterized in that, The raw data for heart failure patients includes: medical history data, physical signs data, blood test data, and cardiac function data.

3. The method for predicting fluid volume status and diuretic resistance status in patients with heart failure according to claim 2, characterized in that, Medical history data includes: course of heart failure, hospitalization records, and medication use; Vital data include: weight, blood pressure, heart rate, and urine output; Blood test data includes: blood test indicators, including electrolyte levels, kidney function indicators, and cardiac markers; Cardiac function data include: ejection fraction (EF) and cardiac output (CO).

4. The method for predicting fluid volume status and diuretic resistance status in patients with heart failure according to claim 1, characterized in that, Preprocessing includes: Missing value filling and standardization.

5. The method for predicting fluid volume status and diuretic resistance status in patients with heart failure according to claim 4, characterized in that, Missing value imputation includes: imputation using the mean or interpolation method; Standardization processing includes: standardizing each data indicator so that the mean of the data is 0 and the standard deviation is 1.

6. The method for predicting fluid volume status and diuretic resistance status in patients with heart failure according to claim 1, characterized in that, The specific steps for inputting the preprocessed raw data into the fluid volume assessment model for heart failure patients are as follows: Using the intelligent diagnostic system, the user interface allows users to input various data indicators of heart failure patients. After inputting the data, clicking the fluid volume prediction button will quickly analyze the input data indicators and use a machine learning model to predict the fluid volume status of heart failure patients. The prediction results are displayed on the interface.

7. The method for predicting fluid volume status and diuretic resistance status in patients with heart failure according to claim 1, characterized in that, The specific content of inputting the data indicators of heart failure patients into the diuretic resistance identification model for heart failure patients is as follows: If the heart failure patient is not in a state of fluid volume deficiency, the intelligent diagnostic system can be used to input data indicators on the user interface, and then click the diuretic tolerance prediction button. The system will quickly analyze the input data indicators and use the LASSO regression model to provide information on the heart failure patient's tolerance to diuretics. The prediction results will be displayed on the interface.

8. The method for predicting fluid volume status and diuretic resistance status in patients with heart failure according to claim 1, characterized in that, Recommendations for medical attention include: suggestions for further laboratory tests, suggestions for adjusting the existing treatment plan, and suggestions for alternative treatment options.

Images