Adenosine dose intelligent recommendation system for cerebral arteriovenous malformation treatment

Abstract

The present application relates to the technical field of adenosine dose recommendation, in particular to an adenosine dose intelligent recommendation system for cerebral arteriovenous malformation treatment, which can realize the following steps through the mutual cooperation between multiple modules: similar reference patients are screened according to the basic information arrays corresponding to the target patient and historical patients; an adenosine individual response index is determined according to the actual saline injection volume, the actual average flushing rate, the actual arrest duration, the expected saline injection volume, the expected flushing rate and the expected arrest duration in the surgical process; a drug efficacy decay compensation coefficient is determined according to the baseline heart rate before administration and the anatomical distance, the adenosine half-life and the heart rate blood flow velocity conversion constant, and the adenosine dose recommendation interval is determined in combination with the error tolerance, the theoretical adenosine dose and the inhibition drug identification. The present application comprehensively analyzes various data related to adenosine dose recommendation, improves the rationality of adenosine dose assisted recommendation, and further improves the auxiliary effect.

Description

Technical Field

[0001] This invention relates to the field of adenosine dosage recommendation technology, and more specifically to an intelligent adenosine dosage recommendation system for the treatment of arteriovenous malformations in the brain. Background Technology

[0002] With the development of technology, methods to assist doctors in making recommendations are becoming increasingly widespread. For example, they can be used to assist doctors in recommending adenosine dosages for the treatment of arteriovenous malformations in the brain. Currently, the common method used to assist in recommending adenosine dosages is to use fixed-dosage experience based on body weight.

[0003] However, when recommending adenosine dosage based on fixed-dose experience with body weight, the following technical problems often arise: In practice, the adenosine dosage required for the treatment of arteriovenous malformations is not only affected by the patient's weight. Therefore, recommending adenosine dosage based solely on the patient's weight may lead to poor rationality of the auxiliary adenosine dosage recommendation due to the limited consideration of factors, resulting in poor auxiliary effects. Summary of the Invention

[0004] To address the technical problem of poor rationality in adenosine dosage recommendations due to the limited consideration of factors, this invention proposes an intelligent adenosine dosage recommendation system for the treatment of cerebral arteriovenous malformations.

[0005] In a first aspect, the present invention provides an intelligent adenosine dosage recommendation system for the treatment of cerebral arteriovenous malformations, the system comprising: The patient screening module is used to filter out similar reference patients corresponding to the target patient from all historical patients based on the basic information arrays corresponding to the target patient and historical patients. The adenosine individual response index determination module is used to determine the adenosine individual response index corresponding to the target patient based on the actual volume of normal saline injected, the actual average flushing rate, the actual pause duration, the expected volume of normal saline injected, the expected flushing rate, and the expected pause duration of all similar reference patients during the operation. The drug efficacy decay compensation coefficient determination module is used to determine the drug efficacy decay compensation coefficient for the target patient based on the target patient's baseline heart rate and anatomical distance before drug administration, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant. The acquisition and determination module is used to acquire the error tolerance and determine the theoretical adenosine dose for the target patient based on the individual adenosine response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration. The adenosine dosage recommendation interval determination module is used to determine the adenosine dosage recommendation interval for the target patient based on the error tolerance, the drug efficacy attenuation compensation coefficient corresponding to the target patient, the theoretical adenosine dose, and the inhibitory drug identifier, so as to realize the adenosine dosage auxiliary recommendation.

[0006] In conjunction with the first aspect above, in one possible implementation, the step of selecting similar reference patients corresponding to the target patient from all historical patients based on the basic information arrays corresponding to the target patient and historical patients includes: Obtain the basic information array corresponding to the target patient and the basic information array corresponding to each historical patient; The cosine similarity between the basic information array corresponding to the target patient and the basic information array corresponding to each historical patient is determined as the target similarity between each historical patient and the target patient. Select historical patients from all historical patients whose target similarity to the target patient is greater than or equal to a preset similarity threshold, and use them as similar reference patients corresponding to the target patient.

[0007] In conjunction with the first aspect above, in one possible implementation, determining the adenosine individual response index corresponding to the target patient based on the actual volume of saline injected, the actual average flushing rate, the actual duration of cardiac arrest, the expected volume of saline injected, the expected flushing rate, and the expected duration of cardiac arrest in all similar reference patients during the procedure includes: For each similar reference patient, obtain the actual volume of saline injected, the actual average flushing rate, the actual duration of cardiac arrest, the expected volume of saline injected, the expected flushing rate, and the expected duration of cardiac arrest during the operation. The reference reliability of each similar reference patient is determined based on the actual volume of saline injected during the operation, the actual average flushing rate, the expected volume of saline injected, and the expected flushing rate of each similar reference patient. The adenosine individual response index for each similar reference patient is determined based on the difference and sum between the expected and actual pause durations during the operation. The adenosine individual response index for the target patient is determined based on the reference reliability and adenosine individual response index of all similar reference patients corresponding to the target patient.

[0008] In conjunction with the first aspect above, in one possible implementation, determining the reference reliability for each similar reference patient based on the actual volume of saline injected during the procedure, the actual average flushing rate, the expected volume of saline injected, and the expected flushing rate includes: Any similar reference patient is designated as the labeled reference patient. The reference reliability of the labeled reference patient is determined based on the difference between the actual and expected saline injection volume during the operation and the difference between the actual and expected flushing rates during the operation.

[0009] In conjunction with the first aspect above, in one possible implementation, determining the drug efficacy attenuation compensation coefficient for the target patient based on the pre-drug baseline heart rate and anatomical distance, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant, includes: Obtain the baseline heart rate and anatomical distance before drug administration for the target patient, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant; The drug delivery delay time for the target patient is determined based on the anatomical distance corresponding to the target patient and the product of the heart rate-blood flow velocity conversion constant and the baseline heart rate before drug administration for the target patient. Based on the drug delivery delay time and adenosine half-life corresponding to the target patient, the drug efficacy attenuation compensation coefficient corresponding to the target patient is determined.

[0010] In conjunction with the first aspect above, in one possible implementation, the method for obtaining the baseline heart rate of the target patient before drug administration includes: Acquire the heart rate data of the target patient during the preset resting time before drug administration; The mean of all heart rate data within the preset pre-drug resting time period is determined as the pre-drug baseline heart rate for the target patient.

[0011] In conjunction with the first aspect above, in one possible implementation, the method for obtaining the baseline mean arterial pressure and pre-drug heart rate variability characteristics corresponding to the target patient includes: By inserting a pressure sensor into the patient's artery, the average pressure of a single test is collected during a preset resting period before drug administration for the target patient, and the average of all single average pressures during the preset resting period before drug administration is determined as the baseline mean arterial pressure before drug administration for the target patient. By attaching ECG electrodes to the patient's chest, the target patient's ECG waveform is acquired during a preset resting period before drug administration. The Pan-Tompkins algorithm is used to identify the R-wave peaks in the ECG waveform, and the duration between two adjacent R-wave peaks is measured and recorded as the RR interval. The standard deviation of all RR intervals in the ECG waveform is determined as the pre-drug heart rate variability characteristic of the target patient.

[0012] In conjunction with the first aspect above, in one possible implementation, obtaining the error tolerance includes: Obtain the actual adenosine dose, inhibitory drug identification, drug efficacy decay compensation coefficient, adenosine individual response index, actual pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration for each historical patient; The theoretical adenosine dose for each historical patient is determined based on the actual adenosine dose, inhibitory drug label, and drug efficacy attenuation compensation coefficient for each historical patient. Using the adenosine individual response index, actual pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics of historical patients as training samples, and the theoretical adenosine dose corresponding to historical patients as training labels, the baseline dose intercept constant, pause duration weight coefficient, individual response weight coefficient, baseline blood pressure weight coefficient, and heart rate variability weight coefficient are output through a multiple linear regression model. The predicted adenosine dose for each historical patient is determined based on the baseline dose intercept constant, pause time weighting coefficient, individual response weighting coefficient, baseline blood pressure weighting coefficient, and heart rate variability weighting coefficient. The error tolerance is determined based on the root mean square error between the predicted and theoretical adenosine doses for all historical patients, the pre-set reliability adjustment coefficient, and the mean of the theoretical adenosine doses for all historical patients.

[0013] In conjunction with the first aspect above, in one possible implementation, determining the theoretical adenosine dose for the target patient based on the target patient's individual adenosine response index, expected pause duration, pre-drug baseline mean arterial pressure, and pre-drug heart rate variability characteristics includes: Obtain the adenosine individual response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration for the target patient; Based on the individual adenosine response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration for the target patient, as well as the baseline dose intercept constant, pause duration weighting coefficient, individual response weighting coefficient, baseline blood pressure weighting coefficient, and heart rate variability weighting coefficient, a multiple linear regression model is used to output the predicted adenosine dose for the target patient, which is denoted as the theoretical adenosine dose for the target patient.

[0014] In conjunction with the first aspect above, in one possible implementation, determining the recommended adenosine dose range for the target patient based on the error tolerance, the drug efficacy attenuation compensation coefficient corresponding to the target patient, the theoretical adenosine dose, and the inhibitory drug identifier includes: The product of the drug efficacy attenuation compensation coefficient, theoretical adenosine dose, and inhibitory drug label for the target patient is determined as the recommended adenosine dose for the target patient. Based on the error tolerance and the recommended adenosine dose for the target patient, determine the recommended adenosine dose range for the target patient.

[0015] Secondly, the present invention provides an intelligent adenosine dosage recommendation method for the treatment of cerebral arteriovenous malformations, implemented by an intelligent adenosine dosage recommendation system for the treatment of cerebral arteriovenous malformations. The method includes: Based on the basic information arrays corresponding to the target patient and historical patients, similar reference patients corresponding to the target patient are selected from all historical patients. Based on the actual volume of saline injected, the actual average flushing rate, the actual duration of cardiac arrest, the expected volume of saline injected, the expected flushing rate, and the expected duration of cardiac arrest in all similar reference patients during the operation, the adenosine individual response index corresponding to the target patient was determined. Based on the target patient's baseline heart rate and anatomical distance before drug administration, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant, determine the drug efficacy decay compensation coefficient for the target patient. Obtain the error tolerance and determine the theoretical adenosine dose for the target patient based on the individual adenosine response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration. Based on the error tolerance, the drug efficacy attenuation compensation coefficient corresponding to the target patient, the theoretical adenosine dose, and the inhibitory drug label, the recommended adenosine dose range for the target patient is determined to achieve auxiliary adenosine dose recommendation.

[0016] The present invention has the following beneficial effects: This invention provides an intelligent adenosine dosage recommendation system for the treatment of cerebral arteriovenous malformations. It comprehensively analyzes various data related to adenosine dosage recommendation, solving the technical problem of poor rationality in adenosine dosage assistance recommendations due to considering only one factor. This improves the rationality of adenosine dosage assistance recommendations and, consequently, enhances the auxiliary effect. Specifically, based on the basic information arrays corresponding to the target patient and historical patients, this invention selects similar reference patients with similar basic information from all historical patients. Based on the actual volume of saline injected, the actual average flushing rate, the actual pause duration, the expected volume of saline injected, the expected flushing rate, and the expected pause duration during the operation of these similar reference patients, the adenosine individual response index corresponding to the target patient is quantified. Based on the target patient's pre-drug baseline heart rate and anatomical distance, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant, the drug efficacy decay compensation coefficient corresponding to the target patient is quantified. Based on the individual response index of adenosine for the target patient, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration, the theoretical adenosine dose for the target patient was quantified. Finally, based on the error tolerance, the drug efficacy decay compensation coefficient for the target patient, the theoretical adenosine dose, and the inhibitory drug identifier, the recommended adenosine dose range for the target patient was quantified to achieve auxiliary adenosine dose recommendation. This addresses, to some extent, the technical problem of poor rationality of auxiliary adenosine dose recommendation due to the single consideration of factors, thereby improving the rationality of auxiliary adenosine dose recommendation and thus improving the auxiliary effect. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the structure of an intelligent adenosine dosage recommendation system for the treatment of arteriovenous malformations of the brain according to the present invention; Figure 2 This is a flowchart of an intelligent adenosine dosage recommendation method for the treatment of cerebral arteriovenous malformations according to the present invention; Figure 3 This is a schematic diagram of the structure of a computer device according to the present invention; The reference numerals in the attached figures include: 101 is the patient screening module, 102 is the adenosine individual response index determination module, 103 is the drug efficacy attenuation compensation coefficient determination module, 104 is the acquisition and determination module, 105 is the adenosine dose recommendation range determination module, 300 is the computer device, 301 is the memory, 302 is the processor, and 303 is the computer program. Detailed Implementation

[0018] refer to Figure 1 This diagram illustrates the structure of an intelligent adenosine dosage recommendation system for the treatment of cerebral arteriovenous malformations (AVMs) according to the present invention. The intelligent adenosine dosage recommendation system for the treatment of cerebral AVMs includes: The patient screening module 101 is used to select similar reference patients corresponding to the target patient from all historical patients based on the basic information arrays corresponding to the target patient and historical patients.

[0019] The target patient can be a patient with cerebral arteriovenous malformation (cAVM) who has not yet undergone the target surgery and is to receive adenosine-assisted dosage recommendation. The target surgery can be a transvenous approach embolization of cAVM for the treatment of cAVM. Unless otherwise specified, the surgery in the embodiments of this invention refers to transvenous approach embolization of cAVM. In reality, during transvenous approach embolization of cAVM, arterial blood flow may flow backward, which may lead to difficulty in coagulation of the embolic agent and easy migration with the blood flow, causing danger. Injecting adenosine can induce transient cardiac arrest to reduce local blood flow pressure and velocity, and, together with the injection of the embolic agent, make it accumulate and coagulate at the lesion site. Therefore, the adenosine dosage recommendation is crucial when treating cAVM. The historical patient can be a patient with cAVM who has already undergone the target surgery. The basic information array can represent the patient's basic information, which can be composed of dimensional data under a preset number of preset dimensions. The preset quantity can be a pre-set number, such as 3. The preset dimensions can be pre-set dimensions related to the patient's basic physical condition. For example, preset dimensions can be, but are not limited to, height, weight, and age. Dimension data can be numerical values ​​under the preset dimensions. For example, the dimension data under the height dimension can be height. Similar reference patients can be historical patients with similar basic information to the target patient, serving as a reference for the target patient and primarily used to assist in quantifying the adenosine individual response index corresponding to the target patient.

[0020] As an example, identifying similar reference patients for a target patient may include the following steps: The first step is to obtain the basic information array corresponding to the target patient and the basic information array corresponding to each historical patient.

[0021] The second step is to determine the target similarity between each historical patient and the target patient by using the cosine similarity between the basic information array corresponding to the target patient and the basic information array corresponding to each historical patient.

[0022] The third step is to select historical patients from all historical patients whose target similarity to the target patient is greater than or equal to a preset similarity threshold, and use them as similar reference patients corresponding to the target patient.

[0023] The preset similarity threshold can be a pre-set minimum level of similarity allowed when patients are considered to have similar conditions, and it can be set manually. For example, the preset similarity threshold can be 0.8.

[0024] It should be noted that the number of historical patients is often sufficient, thus allowing for the selection of at least one similar reference patient. If no similar reference patient exists, the basic information array corresponding to more historical patients can be supplemented to facilitate the selection of a similar reference patient. Optionally, similar reference patients are mainly used to assist in quantifying the adenosine individual response index corresponding to the target patient. Therefore, when no similar reference patient exists, it is not necessary to expand the historical patient database; the adenosine individual response index corresponding to the target patient can be directly set as the population baseline default value. The population baseline default value can be the average response level of the general population to adenosine, representing the standard state in which the patient does not exhibit abnormal drug resistance or sensitivity. It can be set manually, or it can be equal to the mean of the adenosine individual response indices corresponding to all historical patients, such as 1.

[0025] The adenosine individual response index determination module 102 is used to determine the adenosine individual response index corresponding to the target patient based on the actual volume of normal saline injected, the actual average flushing rate, the actual pause duration, the expected volume of normal saline injected, the expected flushing rate, and the expected pause duration of all similar reference patients during the operation.

[0026] As an example, determining the individual adenosine response index for a target patient may include the following steps: The first step involves collecting data from medical staff, including the actual volume of saline injected, the actual average flushing rate, the actual duration of cardiac arrest, the expected volume of saline injected, the expected flushing rate, and the expected duration of cardiac arrest for each similar reference patient during the procedure.

[0027] The actual volume of saline injected can be the total volume of saline actually used for flushing after the adenosine injection during the procedure. The actual average flushing rate can be equal to the actual volume of saline injected divided by the actual flushing time, which can characterize the actual flushing rate to some extent. The actual flushing time can be the total time spent injecting saline after the adenosine injection. The actual cardiac arrest duration can be the time the patient's heart actually maintains a momentary cardiac arrest state after the adenosine injection during the procedure. The expected volume of saline injected can be the theoretically required standard volume of saline solution to inject all the adenosine in the catheter into the heart, which can be entered by the physician. The expected flushing rate can be the theoretically required standard injection speed to ensure that the drug solution is not diluted, which can be entered by the physician. The expected cardiac arrest duration can be the ideal time for the patient's heart to stop beating in order to complete the embolization procedure, which can be entered by the physician.

[0028] The second step is to determine the reference reliability of each similar reference patient based on the actual volume of saline injected, the actual average flushing rate, the expected volume of saline injected, and the expected flushing rate during the operation.

[0029] For example, any similar reference patient can be designated as a labeled reference patient, and the reference reliability of the labeled reference patient can be determined based on the difference between the actual and expected saline injection volume during the operation and the difference between the actual and expected flushing rates during the operation.

[0030] For example, the formula for determining the reference reliability of a labeled reference patient can be: ; Where A represents the reference reliability corresponding to the labeled reference patient. It is an exponential function with the natural constant as its base. It is an absolute value function. It is a marker for the actual volume of normal saline injected into the reference patient during the procedure. It is a marker for the expected volume of saline to be injected into the reference patient during the procedure. It is the actual average flushing rate of the reference patient during the operation. It is a reference reference patient's expected flushing rate during the procedure.

[0031] It should be noted that when A larger value often indicates that the actual volume of saline injected during the procedure deviates more from the expected volume. When A larger value generally indicates that the actual average flushing rate of the labeled reference patient deviates more from the expected flushing rate during the procedure. Therefore, a larger value for A generally indicates that the actual volume of saline injected and the actual average flushing rate of the labeled reference patient during the procedure are closer to the expected values. This suggests that the flushing procedure performed by the labeled reference patient during the procedure is likely to be more standardized and almost identical to standard procedures. The causal relationship of the pause time is often purer, and the reference value and reliability of the data are often higher.

[0032] The third step is to determine the adenosine individual response index for each similar reference patient based on the difference and sum of the expected and actual pause durations during the operation.

[0033] For example, the formula for determining the adenosine individual response index corresponding to a similar reference patient can be: ; in, is the adenosine individual response index corresponding to the i-th similar reference patient. i is the sequence number of the similar reference patient corresponding to the target patient. It is the expected pause duration during the operation of the i-th similar reference patient. It is the actual duration of cardiac arrest during the operation of the i-th similar reference patient.

[0034] It should be noted that when When the value equals 1, it usually indicates that the actual pause duration equals the expected pause duration, which can characterize the baseline state and represent standard resistance. A value greater than 1 often indicates that the actual cardiac arrest duration is shorter than the expected cardiac arrest duration, and often suggests that the i-th similar reference patient has some resistance to adenosine. Therefore, when recommending adenosine dosage for target patients with similar conditions to the i-th similar reference patient, the dosage can be appropriately increased. A value less than 1 often indicates that the actual duration of cardiac arrest is longer than the expected duration of cardiac arrest. This often indicates that the i-th similar reference patient is sensitive to adenosine, and the dosage can be appropriately reduced when recommending adenosine dosage for target patients with similar conditions to the i-th similar reference patient.

[0035] The fourth step is to determine the adenosine individual response index for the target patient based on the reference reliability and adenosine individual response index of all similar reference patients corresponding to the target patient.

[0036] For example, the formula for determining the individual adenosine response index corresponding to the target patient can be: ; in, is the individual adenosine response index corresponding to the target patient. N is the number of similar reference patients corresponding to the target patient. i is the sequence number of the similar reference patients corresponding to the target patient. B is the reference reliability corresponding to the i-th similar reference patient. B is the cumulative value of the reference reliability corresponding to all similar reference patients corresponding to the target patient. It is the adenosine individual response index corresponding to the i-th similar reference patient.

[0037] It should be noted that, Can be used as The weight. When The larger the value, the more standardized the flushing procedure of the i-th similar reference patient is likely to be during surgery, and the more likely it is to be almost identical to the standard procedure. The causal relationship of the pause time is often purer, and the reference value and reliability of the data are often higher. This can characterize the degree of "resistance / sensitivity" of the i-th similar reference patient to adenosine drugs. Therefore, It can characterize the degree of "resistance / sensitivity" of a target patient to adenosine drugs.

[0038] The drug efficacy attenuation compensation coefficient determination module 103 is used to determine the drug efficacy attenuation compensation coefficient for the target patient based on the target patient's baseline heart rate and anatomical distance before drug administration, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant.

[0039] As an example, determining the drug efficacy attenuation compensation factor for a target patient may include the following steps: The first step is to obtain the baseline heart rate and anatomical distance of the target patient before drug administration, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant.

[0040] The pre-drug baseline heart rate characterizes the patient's resting heart rate before adenosine injection. The pre-drug baseline heart rate for the target patient can be obtained by using a medical monitor to acquire the target patient's heart rate data within a preset pre-drug resting time period, and then determining the average of all heart rate data within that preset pre-drug resting time period as the pre-drug baseline heart rate for the target patient. The preset pre-drug resting time period can be a pre-set time period within the resting state before adenosine injection, and its duration can be pre-set, such as 180 seconds. The resting state can be the basic physiological state of a person being awake, quiet, relaxed, without significant physical activity, without emotional excitement, and without being stimulated by strong external sounds or lights. For example, in this embodiment of the invention, the resting state can refer to a patient lying flat on the operating table, having completed preoperative preparation but not yet subjected to invasive procedures, with stable vital signs and cardiovascular compensation mechanisms at baseline levels.

[0041] Anatomical distance refers to the total physical length that the drug solution needs to travel from the injection port of the external catheter, along the venous pathway, and finally to the right atrium (AV node) of the heart; that is, the total length of the catheter and blood vessels through which the drug solution flows. Adenosine half-life refers to the time required for adenosine to decrease in concentration by half after being metabolized by red blood cells and vascular endothelial cells in the blood. The heart rate-flow velocity conversion constant can be used to describe the linear relationship between changes in a patient's heart rate and changes in venous blood flow velocity. In reality, the faster the heart rate, the faster the venous return, and the shorter the time it takes for the drug to reach the heart. The heart rate-flow velocity conversion constant can be obtained by acquiring heart rate data and venous blood flow velocity measured by Doppler ultrasound for each historical patient at the same time period, fitting a straight line with heart rate data as the x-axis and venous blood flow velocity as the y-axis, and recording the slope of the fitted line as the heart rate-flow velocity conversion constant.

[0042] The second step is to determine the drug delivery delay time for the target patient based on the anatomical distance corresponding to the target patient and the product of the heart rate-blood flow velocity conversion constant and the baseline heart rate before drug administration for the target patient.

[0043] For example, the formula for determining the drug delivery delay time for a target patient can be: ; Where D is the drug delivery delay time for the target patient, L is the anatomical distance for the target patient, k is the heart rate-blood flow velocity conversion constant, and H is the baseline heart rate for the target patient before drug administration.

[0044] It should be noted that D can characterize the actual physical time it takes for the drug solution to travel from the catheter opening to the target point (AV node) in the heart via the bloodstream. The larger the value, the longer the drug solution is delayed on the way. Because adenosine has a very short half-life, the longer the delay on the way, the more of the drug will be metabolized before it reaches the heart, which often leads to the need to give a higher "compensation dose" to ensure the efficacy of the drug.

[0045] The third step is to determine the drug efficacy attenuation compensation coefficient for the target patients based on the drug delivery delay time and adenosine half-life.

[0046] For example, the formula for determining the drug efficacy attenuation compensation coefficient for the target patient can be: ; Where C is the drug efficacy attenuation compensation coefficient for the target patient. It is an exponential function with the natural constant as its base. D is the drug delivery delay time for the target patient. d is the adenosine half-life. 0.693 is... An approximation of the elimination rate constant in drug metabolism models is often used to describe the relationship between the elimination rate constant and the half-life. . It is the natural logarithm function.

[0047] It should be noted that when C is larger, it often indicates that the drug is more severely damaged during transit, such as when the tubing is particularly long or the flow rate is particularly slow.

[0048] The acquisition and determination module 104 is used to acquire the error tolerance and determine the theoretical adenosine dose for the target patient based on the individual adenosine response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration.

[0049] Error tolerance characterizes the percentage fluctuation error within the recommended safe dose range. Pre-dose baseline mean arterial pressure characterizes the patient's mean arterial pressure at rest before adenosine injection. The method for obtaining the pre-dose baseline mean arterial pressure for the target patient can be as follows: by inserting a pressure sensor into the patient's artery, the average pressure of a single reading is collected within a preset pre-dose resting time period for the target patient, and the mean of all single readings within the preset pre-dose resting time period is determined as the pre-dose baseline mean arterial pressure for the target patient. Single average pressure = (high pressure + 2 × low pressure) / 3.

[0050] Pre-drug administration heart rate variability can characterize the heart rate variability of a patient at rest before adenosine infusion. The method for obtaining the pre-drug administration heart rate variability for the target patient can be as follows: ECG waveforms of the target patient are acquired during a preset pre-drug administration resting period using ECG electrodes attached to the patient's chest. The R-wave peaks in the ECG waveforms are identified using the Pan-Tompkins algorithm. The duration between two adjacent R-wave peaks is measured and recorded as the RR interval. The standard deviation of all RR intervals in the ECG waveforms is then determined as the pre-drug administration heart rate variability for the target patient.

[0051] As an example, the acquisition and determination module 104 can specifically implement the following steps: The first step is to obtain the actual adenosine dose, inhibitory drug identification, drug efficacy attenuation compensation coefficient, adenosine individual response index, actual pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration for each historical patient.

[0052] The actual adenosine dose can be the dose of adenosine actually injected into the patient during surgery. The inhibitory drug label can be obtained through expert scoring, and this label indicates the presence of other drugs that inhibit adenosine metabolism or have a similar strong conduction blocking effect on the atrioventricular node. If present, the inhibitory drug label can be set to 0.5; if absent, it can be set to 1. Other drugs that inhibit adenosine metabolism or have a similar strong conduction blocking effect on the atrioventricular node can be, but are not limited to: dipyridamole, carbamazepine, digoxin, and verapamil.

[0053] The second step is to determine the theoretical adenosine dose for each historical patient based on the actual adenosine dose, inhibitory drug label, and drug efficacy attenuation compensation coefficient.

[0054] For example, the formula for determining the theoretical adenosine dose for a historical patient can be: ; in, It is the theoretical adenosine dose corresponding to the j-th historical patient. It is the actual adenosine dose corresponding to the j-th historical patient. It is the drug efficacy attenuation compensation coefficient corresponding to the j-th historical patient. It is the inhibitory medication identifier corresponding to the j-th historical patient.

[0055] It should be noted that, It can characterize the actual effective dose of the drug that acts purely on the atrioventricular node of the heart of the j-th historical patient.

[0056] The third step involves using the adenosine individual response index, actual pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics of historical patients as training samples, and the theoretical adenosine dose corresponding to historical patients as training labels. Through a multiple linear regression model, the baseline dose intercept constant, pause duration weight coefficient, individual response weight coefficient, baseline blood pressure weight coefficient, and heart rate variability weight coefficient are output.

[0057] The basal dose intercept constant represents the minimum absolute basal adenosine dose required to stop the heart in an average person when the individual response index, actual pause duration, pre-dose baseline mean arterial pressure, and pre-dose heart rate variability are all ideal or zero. The pause duration weighting coefficient measures the influence of the physician-required pause duration on the adenosine dose. The individual response weighting coefficient measures the influence of the patient's innate resistance or sensitivity to adenosine on the dose. The baseline blood pressure weighting coefficient measures the influence of the patient's baseline intravascular pressure before administration on the dose. The heart rate variability weighting coefficient measures the influence of the patient's sympathetic nervous system excitation level on the dose.

[0058] The fourth step is to determine the predicted adenosine dose for each historical patient based on the baseline dose intercept constant, pause time weighting coefficient, individual response weighting coefficient, baseline blood pressure weighting coefficient, and heart rate variability weighting coefficient.

[0059] For example, the formula for determining the predicted adenosine dose for a historical patient can be: ; in, This represents the predicted adenosine dose for the j-th historical patient. j is the patient's serial number. , , , and These are the baseline dose intercept constant, pause time weighting coefficient, individual response weighting coefficient, baseline blood pressure weighting coefficient, and heart rate variability weighting coefficient. It is the natural logarithm function. It is the actual duration of cardiac arrest during the operation of the j-th historical patient. It is the adenosine individual response index corresponding to the j-th historical patient. It is the baseline mean arterial pressure before drug administration for the j-th historical patient. It represents the pre-drug-dosage heart rate variability characteristic corresponding to the j-th historical patient.

[0060] It should be noted that, It can characterize the effective drug dose acting on the atrioventricular node of the heart in the j-th historical patient, as predicted by a multiple linear regression model. The closer it is to the effective dose, the better. The higher the accuracy of the predictions, the better the multiple linear regression model.

[0061] The fifth step is to determine the error tolerance based on the root mean square error between the predicted and theoretical adenosine doses for all historical patients, the pre-set reliability adjustment coefficient, and the mean of the theoretical adenosine doses for all historical patients.

[0062] The preset reliability adjustment coefficient can be a manually set safety control valve used to manually amplify the statistical error predicted by the algorithm. Its value range can be from 1 to 1.5, for example, it can be 1.2.

[0063] For example, the formula for determining the error tolerance can be: ; Where w is the error tolerance. U is the root mean square error between the predicted adenosine dose and the theoretical adenosine dose for all historical patients. This is the mean of the theoretical adenosine doses for all historical patients. f is the pre-set reliability adjustment coefficient.

[0064] It should be noted that w can characterize the percentage fluctuation error of the safe recommended dose range, and its normal reasonable value range is usually between 0.1 and 0.5.

[0065] The sixth step is to obtain the adenosine individual response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration for the target patient.

[0066] Step 7: Based on the individual adenosine response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration for the target patient, as well as the baseline dose intercept constant, pause duration weighting coefficient, individual response weighting coefficient, baseline blood pressure weighting coefficient, and heart rate variability weighting coefficient, a multiple linear regression model is used to output the predicted adenosine dose for the target patient, which is denoted as the theoretical adenosine dose for the target patient.

[0067] For example, the formula for determining the theoretical adenosine dose for a target patient can be: ; Where Q is the theoretical adenosine dose for the target patient, which can be equal to the predicted adenosine dose for the target patient. , , , and These are the baseline dose intercept constant, pause time weighting coefficient, individual response weighting coefficient, baseline blood pressure weighting coefficient, and heart rate variability weighting coefficient. It is a natural logarithm function. t is the expected pause duration for the target patient during the procedure. This is the individual adenosine response index for the target patient. G is the baseline mean arterial pressure for the target patient before drug administration. It refers to the pre-drug heart rate variability characteristics of the target patient.

[0068] It should be noted that Q can represent the predicted amount of effective drug required to act on the atrioventricular node of the target patient's heart.

[0069] The adenosine dose recommendation interval determination module 105 is used to determine the adenosine dose recommendation interval for the target patient based on the error tolerance, the drug efficacy attenuation compensation coefficient corresponding to the target patient, the theoretical adenosine dose, and the inhibitory drug identifier, so as to realize the adenosine dose auxiliary recommendation.

[0070] As an example, determining the recommended adenosine dosage range for a target patient may include the following steps: The first step is to determine the recommended adenosine dose for the target patient by multiplying the drug efficacy attenuation compensation coefficient, the theoretical adenosine dose, and the inhibitory drug label.

[0071] For example, the formula for determining the recommended adenosine dose for a target patient can be: ; Where F is the recommended adenosine dose for the target patient. C is the efficacy attenuation compensation coefficient for the target patient. Q is the theoretical adenosine dose for the target patient, which can be equal to the predicted adenosine dose for the target patient. It is the identifier for the inhibitory medication corresponding to the target patient.

[0072] It should be noted that F can represent the estimated value of the actual adenosine dose required by the target patient and can be used as a recommended value for the optimal adenosine dose.

[0073] The second step is to determine the recommended adenosine dose range for the target patient based on the error tolerance and the recommended adenosine dose for the target patient.

[0074] For example, the recommended adenosine dosage range for the target patient can be [ , F represents the recommended adenosine dose for the target patient. w represents the tolerance for error.

[0075] It should be noted that w can characterize the percentage fluctuation error within the safe recommended dose range. F can characterize the estimated amount of adenosine actually needed by the target patient and can be used as a better recommended adenosine dose. Therefore, It can be used as the lower limit of the recommended dosage range for adenosine. It can be used as the upper limit of the recommended adenosine dosage range to assist doctors in selecting the final adenosine dosage.

[0076] refer to Figure 2 Based on the same inventive concept as the above-described system embodiments, this invention provides an intelligent adenosine dosage recommendation method for the treatment of cerebral arteriovenous malformations, comprising the following steps: Step S1: Based on the basic information arrays corresponding to the target patient and historical patients, filter out similar reference patients corresponding to the target patient from all historical patients; Step S2: Based on the actual volume of saline injected, the actual average flushing rate, the actual duration of cardiac arrest, the expected volume of saline injected, the expected flushing rate, and the expected duration of cardiac arrest of all similar reference patients during the operation, determine the adenosine individual response index corresponding to the target patient. Step S3: Determine the drug efficacy decay compensation coefficient for the target patient based on the baseline heart rate and anatomical distance before drug administration, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant. Step S4: Obtain the error tolerance and determine the theoretical adenosine dose for the target patient based on the individual adenosine response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration. Step S5: Based on the error tolerance, the drug efficacy attenuation compensation coefficient corresponding to the target patient, the theoretical adenosine dose, and the inhibitory drug identifier, determine the recommended adenosine dose range for the target patient to achieve auxiliary adenosine dose recommendation.

[0077] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. For example, as shown... Figure 3 As shown, the computer device 300 includes: a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302, wherein when the processor 302 executes the computer program 303, the computer device can execute the aforementioned intelligent adenosine dosage recommendation method for the treatment of cerebral arteriovenous malformations.

[0078] In summary, this invention, based on the basic information arrays corresponding to the target patient and historical patients, selects similar reference patients from all historical patients whose basic information is similar to that of the target patient. Based on the actual volume of saline injected, the actual average flushing rate, the actual duration of cardiac arrest, the expected volume of saline injected, the expected flushing rate, and the expected duration of cardiac arrest during the surgical procedure of the similar reference patients, the adenosine individual response index corresponding to the target patient is quantified. Based on the target patient's pre-drug baseline heart rate and anatomical distance, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant, the drug efficacy decay compensation coefficient corresponding to the target patient is quantified. Based on the individual response index of adenosine for the target patient, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration, the theoretical adenosine dose for the target patient was quantified. Finally, based on the error tolerance, the drug efficacy decay compensation coefficient for the target patient, the theoretical adenosine dose, and the inhibitory drug identifier, the recommended adenosine dose range for the target patient was quantified to achieve auxiliary adenosine dose recommendation. This addresses, to some extent, the technical problem of poor rationality of auxiliary adenosine dose recommendation due to the single consideration of factors, thereby improving the rationality of auxiliary adenosine dose recommendation and thus improving the auxiliary effect.

Claims

1. An intelligent adenosine dosage recommendation system for the treatment of cerebral arteriovenous malformations, characterized in that, The system includes: The patient screening module is used to filter out similar reference patients corresponding to the target patient from all historical patients based on the basic information arrays corresponding to the target patient and historical patients. The adenosine individual response index determination module is used to determine the adenosine individual response index corresponding to the target patient based on the actual volume of normal saline injected, the actual average flushing rate, the actual pause duration, the expected volume of normal saline injected, the expected flushing rate, and the expected pause duration of all similar reference patients during the operation. The drug efficacy decay compensation coefficient determination module is used to determine the drug efficacy decay compensation coefficient for the target patient based on the target patient's baseline heart rate and anatomical distance before drug administration, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant. The acquisition and determination module is used to acquire the error tolerance and determine the theoretical adenosine dose for the target patient based on the individual adenosine response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration. The adenosine dosage recommendation interval determination module is used to determine the adenosine dosage recommendation interval for the target patient based on the error tolerance, the drug efficacy attenuation compensation coefficient corresponding to the target patient, the theoretical adenosine dose, and the inhibitory drug identifier, so as to realize the adenosine dosage auxiliary recommendation.

2. The adenosine dosage intelligent recommendation system for the treatment of cerebral arteriovenous malformations according to claim 1, characterized in that, The step of selecting similar reference patients corresponding to the target patient from all historical patients based on the basic information arrays corresponding to the target patient and historical patients includes: Obtain the basic information array corresponding to the target patient and the basic information array corresponding to each historical patient; The cosine similarity between the basic information array corresponding to the target patient and the basic information array corresponding to each historical patient is determined as the target similarity between each historical patient and the target patient. Select historical patients from all historical patients whose target similarity to the target patient is greater than or equal to a preset similarity threshold, and use them as similar reference patients corresponding to the target patient.

3. The adenosine dosage intelligent recommendation system for the treatment of cerebral arteriovenous malformations according to claim 1, characterized in that, The adenosine individual response index for the target patient is determined based on the actual volume of saline injected, the actual average flushing rate, the actual duration of cardiac arrest, the expected volume of saline injected, the expected flushing rate, and the expected duration of cardiac arrest in all similar reference patients during the procedure. For each similar reference patient, obtain the actual volume of saline injected, the actual average flushing rate, the actual duration of cardiac arrest, the expected volume of saline injected, the expected flushing rate, and the expected duration of cardiac arrest during the operation. The reference reliability of each similar reference patient is determined based on the actual volume of saline injected during the operation, the actual average flushing rate, the expected volume of saline injected, and the expected flushing rate of each similar reference patient. The adenosine individual response index for each similar reference patient is determined based on the difference and sum between the expected and actual pause durations during the operation. The adenosine individual response index for the target patient is determined based on the reference reliability and adenosine individual response index of all similar reference patients corresponding to the target patient.

4. The adenosine dosage intelligent recommendation system for the treatment of cerebral arteriovenous malformations according to claim 3, characterized in that, The reference reliability for each similar reference patient is determined based on the actual volume of saline injected, the actual average flushing rate, the expected volume of saline injected, and the expected flushing rate during the procedure. This includes: Any similar reference patient is designated as the labeled reference patient. The reference reliability of the labeled reference patient is determined based on the difference between the actual and expected saline injection volume during the operation and the difference between the actual and expected flushing rates during the operation.

5. The adenosine dosage intelligent recommendation system for the treatment of cerebral arteriovenous malformations according to claim 1, characterized in that, The determination of the drug efficacy attenuation compensation coefficient for the target patient, based on the pre-drug baseline heart rate and anatomical distance, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant, includes: Obtain the baseline heart rate and anatomical distance before drug administration for the target patient, as well as the adenosine half-life and heart rate-blood flow velocity conversion constant; The drug delivery delay time for the target patient is determined based on the anatomical distance corresponding to the target patient and the product of the heart rate-blood flow velocity conversion constant and the baseline heart rate before drug administration for the target patient. Based on the drug delivery delay time and adenosine half-life corresponding to the target patient, the drug efficacy attenuation compensation coefficient corresponding to the target patient is determined.

6. The adenosine dosage intelligent recommendation system for the treatment of cerebral arteriovenous malformations according to claim 5, characterized in that, Methods for obtaining the baseline heart rate before drug administration for the target patient include: Acquire the heart rate data of the target patient during the preset resting time before drug administration; The mean of all heart rate data within the preset pre-drug resting time period is determined as the pre-drug baseline heart rate for the target patient.

7. The adenosine dosage intelligent recommendation system for the treatment of cerebral arteriovenous malformations according to claim 1, characterized in that, Methods for obtaining the baseline mean arterial pressure and pre-drug heart rate variability characteristics for the target patients include: By inserting a pressure sensor into the patient's artery, the average pressure of a single test is collected during a preset resting period before drug administration for the target patient, and the average of all single average pressures during the preset resting period before drug administration is determined as the baseline mean arterial pressure before drug administration for the target patient. By attaching ECG electrodes to the patient's chest, the target patient's ECG waveform is acquired during a preset resting period before drug administration. The Pan-Tompkins algorithm is used to identify the R-wave peaks in the ECG waveform, and the duration between two adjacent R-wave peaks is measured and recorded as the RR interval. The standard deviation of all RR intervals in the ECG waveform is determined as the pre-drug heart rate variability characteristic of the target patient.

8. The adenosine dosage intelligent recommendation system for the treatment of cerebral arteriovenous malformations according to claim 1, characterized in that, The acquisition of error tolerance includes: Obtain the actual adenosine dose, inhibitory drug identification, drug efficacy decay compensation coefficient, adenosine individual response index, actual pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration for each historical patient; The theoretical adenosine dose for each historical patient is determined based on the actual adenosine dose, inhibitory drug label, and drug efficacy attenuation compensation coefficient for each historical patient. Using the adenosine individual response index, actual pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics of historical patients as training samples, and the theoretical adenosine dose corresponding to historical patients as training labels, the baseline dose intercept constant, pause duration weight coefficient, individual response weight coefficient, baseline blood pressure weight coefficient, and heart rate variability weight coefficient are output through a multiple linear regression model. The predicted adenosine dose for each historical patient is determined based on the baseline dose intercept constant, pause time weighting coefficient, individual response weighting coefficient, baseline blood pressure weighting coefficient, and heart rate variability weighting coefficient. The error tolerance is determined based on the root mean square error between the predicted and theoretical adenosine doses for all historical patients, the pre-set reliability adjustment coefficient, and the mean of the theoretical adenosine doses for all historical patients.

9. The adenosine dosage intelligent recommendation system for the treatment of cerebral arteriovenous malformations according to claim 8, characterized in that, The theoretical adenosine dose for the target patient is determined based on the individual adenosine response index, expected pause duration, baseline mean arterial pressure before administration, and pre-administration heart rate variability characteristics, including: Obtain the adenosine individual response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration for the target patient; Based on the individual adenosine response index, expected pause duration, baseline mean arterial pressure before administration, and heart rate variability characteristics before administration for the target patient, as well as the baseline dose intercept constant, pause duration weighting coefficient, individual response weighting coefficient, baseline blood pressure weighting coefficient, and heart rate variability weighting coefficient, a multiple linear regression model is used to output the predicted adenosine dose for the target patient, which is denoted as the theoretical adenosine dose for the target patient.

10. The intelligent adenosine dosage recommendation system for the treatment of cerebral arteriovenous malformations according to claim 1, characterized in that, The process of determining the recommended adenosine dose range for the target patient based on error tolerance, the drug efficacy attenuation compensation coefficient for the target patient, the theoretical adenosine dose, and the inhibitory drug identifier includes: The product of the drug efficacy attenuation compensation coefficient, theoretical adenosine dose, and inhibitory drug label for the target patient is determined as the recommended adenosine dose for the target patient. Based on the error tolerance and the recommended adenosine dose for the target patient, determine the recommended adenosine dose range for the target patient.

Images