CN118919079A - Painless gastroscope diagnosis and treatment preoperative anesthesia evaluation method and system - Google Patents

Abstract

The present invention relates to the field of anesthesia risk assessment technology, and specifically to a method and system for preoperative anesthesia assessment for painless gastroenteroscopic diagnosis and treatment. The present invention analyzes the changes in intraoperative monitoring data of historical patients to assess their intraoperative risks, and then measures the weight factor of each preoperative examination indicator with the help of the distribution characteristics of the number of historical patients under different physical sign parameters under the preoperative examination indicators, so as to accurately assess historical patients with similar preoperative physical sign data, i.e., physical constitution, by combining the weight factor, and then assess the intraoperative risk of current patients with the help of the intraoperative risk of historical patients, thereby improving the accuracy of preoperative anesthesia risk assessment.

Description

A method and system for evaluating anesthesia before painless gastroenteroscopic diagnosis and treatment

Technical Field

The present invention relates to the technical field of anesthesia risk assessment, and in particular to a method and system for assessing anesthesia before painless gastroenteroscopic diagnosis and treatment.

Background Art

Painless gastroenteroscopy is a medical examination method that helps doctors diagnose digestive system diseases. The difference between it and traditional gastroenteroscopy is that it reduces the patient's discomfort and pain during the operation by using sedatives or anesthetics, thereby improving the acceptability of the examination. In order to ensure the safety of the patient's anesthesia during the operation and reduce the potential risk of postoperative complications, it is crucial to conduct a preoperative anesthesia evaluation for patients who are about to undergo painless gastroenteroscopy.

The American Society of Anesthesiologists (ASA) grading system is currently the most commonly used standardized tool for grading anesthesia risk. Anesthesiologists assess each patient's physical condition and surgical risk before anesthesia to develop a more appropriate anesthesia plan for them. However, the ASA grading system also has certain limitations. Specifically, it analyzes the patient's preoperative physical signs based on the anesthesiologist's clinical experience, which is affected by experience and inefficient. In addition, the system cannot analyze the patient's related physical signs during surgery, making it difficult to assess the patient's anesthesia sensitivity, resulting in low accuracy and efficiency in preoperative anesthesia risk assessment.

Summary of the invention

In order to solve the technical problem of low accuracy of preoperative anesthesia assessment in the prior art, the purpose of the present invention is to provide a method and system for preoperative anesthesia assessment for painless gastroenteroscopic diagnosis and treatment. The technical scheme adopted is as follows:

The present invention provides a method for evaluating anesthesia before painless gastroenteroscopic diagnosis and treatment, the method comprising:

Obtaining preoperative physical sign data and intraoperative monitoring data of all patients who have undergone painless gastroenteroscopy in history, and obtaining preoperative physical sign data of the current patient to be examined; the preoperative physical sign data includes physical sign parameters of each preoperative examination indicator, and the intraoperative monitoring data includes all dynamic parameter sequences of each intraoperative monitoring indicator;

All reference patients of the patient to be examined are screened from all historically examined patients according to the preoperative physical sign data; under each intraoperative monitoring index, according to the fluctuation changes of the dynamic parameter sequences of all reference patients, combined with the change differences between the dynamic parameter sequences corresponding to different reference patients, the intraoperative risk index of each reference patient is obtained;

Under each preoperative examination indicator, according to the quantity distribution of the reference patients corresponding to different physical sign parameters, the indicator distribution model under each preoperative examination indicator is obtained; the fusion distribution model is obtained by comprehensively analyzing the quantity distribution of the reference patients corresponding to different physical sign parameters under all preoperative examination indicators and the variation range of the physical sign parameters under each preoperative examination indicator; according to the difference between the fusion distribution model and the indicator distribution model, the weight factor of the corresponding preoperative examination indicator is obtained;

According to the preoperative physical sign data and the weight factor of each preoperative examination indicator, combined with the intraoperative risk indicator of each reference patient, the anesthesia risk of the current patient to be examined is evaluated.

Furthermore, the method for obtaining the reference patient includes:

All patients examined in the past and the patients to be examined currently are regarded as first patients to be classified; and corresponding metric distances are obtained according to the similarities between the preoperative physical sign data corresponding to different first patients to be classified;

Based on the clustering algorithm and the metric distance between the preoperative physical sign data, all patient clusters are obtained; all historically examined patients in the patient cluster to which the current patient to be examined belongs are used as all reference patients for the current patient to be examined.

Furthermore, the method for obtaining the intraoperative risk indicator includes:

Taking any intraoperative monitoring indicator as a target monitoring indicator, under the target monitoring indicator, obtaining a sensitivity weight of the target monitoring indicator according to all adjacent extreme value differences in the dynamic parameter sequences of all the reference patients;

Taking any of the reference patients as the target patient, under the target monitoring index, according to the change difference between the target patient and each of the other reference patients corresponding to the dynamic parameter sequence, obtain the fluctuation deviation between the target patient and each of the other reference patients; taking the mean of all the fluctuation deviations as the index deviation of the target patient relative to all the other reference patients under the target monitoring index;

The sensitivity weight of each intraoperative monitoring indicator is used to weight the deviation of the corresponding indicator and the weighted average result is normalized to obtain the intraoperative risk indicator of the target patient.

Furthermore, the method for obtaining the sensitivity weight includes:

In the intraoperative monitoring data of each of the reference patients, the cumulative result of the change differences of all adjacent extreme values in the dynamic parameter sequence under the target monitoring index is taken as the first fluctuation parameter, and the variance of the dynamic parameter sequence under the target monitoring index is taken as the second fluctuation parameter; the product of the first fluctuation parameter and the second fluctuation parameter is taken as the fluctuation parameter of the target monitoring index in the intraoperative monitoring data corresponding to the reference patient;

The mean values of all the fluctuation parameters of the target monitoring index in the intraoperative monitoring data of all the reference patients are normalized, and the normalized result is used as the sensitivity weight of the target monitoring index.

Furthermore, the method for obtaining the fusion distribution model includes:

Screening all low-risk patients from all the reference patients according to the intraoperative risk index; determining characteristic physical sign parameters of each preoperative examination index according to the distribution of physical sign parameters of all the low-risk patients under each preoperative examination index, and using the characteristic physical sign parameters to standardize all physical sign parameters of all the reference patients respectively to obtain adjusted physical sign parameters;

Under each preoperative examination indicator, the distribution model under each preoperative examination indicator is obtained according to the distribution of the number of reference patients corresponding to different adjusted sign parameters; the distribution models under all preoperative examination indicators and the variation range of the adjusted sign parameters under each preoperative examination indicator are combined to obtain a fusion distribution model.

Furthermore, the method for obtaining the adjustment vital sign parameters includes:

All the reference patients whose intraoperative risk index is less than the preset threshold are regarded as low-risk patients; under each preoperative examination index, the extreme difference among all the physical sign parameters corresponding to all the low-risk patients is regarded as the extreme physical sign parameter, and the physical sign parameter corresponding to the maximum total number of the low-risk patients is regarded as the reference physical sign parameter;

Under each preoperative examination indicator, the extreme sign parameter is used as the denominator, the difference between the sign parameter of each reference patient and the reference sign parameter is used as the numerator, and the fractional ratio is used as the adjusted sign parameter of the corresponding sign parameter.

Furthermore, the method for obtaining the fusion distribution model includes:

Under each preoperative examination indicator, a two-dimensional coordinate system is constructed with the size of the adjusted physical sign parameter as the horizontal axis and the total number of reference patients as the vertical axis; the total number of reference patients under the corresponding physical sign parameter of each adjusted physical sign parameter before adjustment is mapped to the two-dimensional coordinate system and a curve is fitted to obtain a distribution model under each preoperative examination indicator;

Under each preoperative examination indicator, the range of all the adjusted physical sign parameters is normalized as the vertical axis weight, and the vertical axis weight is used to perform weighted adjustment on the total number of reference patients corresponding to each adjusted physical sign parameter in the distribution model to obtain the weighted total number of reference patients corresponding to each adjusted physical sign parameter;

Under each of the adjusted physical sign parameters, the corresponding weighted total quantities under all preoperative examination indicators are accumulated to obtain the adjusted quantity of reference patients under each of the adjusted physical sign parameters; the adjusted quantity of reference patients under each of the adjusted physical sign parameters is mapped to the newly constructed two-dimensional coordinate system and the curve is fitted to obtain a fusion distribution model.

Furthermore, the method for obtaining the weight factor includes:

The DTW distance between the fusion distribution model and the indicator distribution model under each preoperative examination indicator is used as the x in the exponential function exp(-x) with the natural constant e as the base, and the exponential function value is used as the weight factor of the corresponding preoperative examination indicator.

Furthermore, the method for evaluating the anesthesia risk of the patient to be examined currently includes:

All reference patients and the current patient to be examined are taken as the second patients to be classified; based on the weight factor of each preoperative examination indicator, the weighted Euclidean distance between the preoperative physical sign data of different second patients to be classified is obtained; based on the density clustering algorithm and the weighted Euclidean distance, all sample patient clusters are obtained; all reference patients in the sample patient cluster to which the current patient to be examined belongs are taken as all target patients of the current patient to be examined;

A training sample set is constructed based on the preoperative physical sign data of all the target patients and the intraoperative risk indicators, and the anesthesia risk assessment model is trained using the training sample set; the preoperative physical sign data of the current patient to be examined is input into the trained anesthesia risk assessment model, and the estimated intraoperative risk indicator of the current patient to be examined is output.

The present invention also proposes a preoperative anesthesia assessment system for painless gastrointestinal endoscopic diagnosis and treatment, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the preoperative anesthesia assessment method for painless gastrointestinal endoscopic diagnosis and treatment when executing the computer program.

The present invention has the following beneficial effects:

The present invention obtains the preoperative vital sign data and intraoperative monitoring data of all patients with historical examinations of painless gastroenteroscopy, and obtains the preoperative vital sign data of the patient to be examined currently; the preoperative vital sign data include the vital sign parameters of each preoperative examination index, and the intraoperative monitoring data include all dynamic parameter sequences of each intraoperative monitoring index; the intraoperative monitoring data can be used to evaluate the anesthetic effect and safety risk of the patient during sedation and anesthesia, and the preoperative vital sign data reflects the physical fitness of each patient, so as to subsequently evaluate the intraoperative anesthesia risk of the current patient to be examined based on the intraoperative anesthesia risk of historically examined patients with similar physical fitness; based on the preoperative vital sign data, all reference patients with similar physical fitness to the current patient to be examined are preliminarily screened out from all historically examined patients; under each intraoperative monitoring index, based on the fluctuation changes of the dynamic parameter sequences of all reference patients, combined with the change differences between the corresponding dynamic parameter sequences of different reference patients, the physical fitness of each reference patient is obtained. The intraoperative risk index of the patient is obtained; under each preoperative examination index, the index distribution model is obtained according to the number distribution of reference patients corresponding to different physical sign parameters; the fusion distribution model is obtained by comprehensively analyzing the number distribution of reference patients corresponding to different physical sign parameters under all preoperative examination indicators, and the variation range of physical sign parameters under each preoperative examination indicator; according to the difference between the fusion distribution model and the index distribution model, the weight factor of the corresponding preoperative examination indicator is obtained; if the difference between the index distribution model and the fusion distribution model is smaller, it means that the number distribution of reference patients under the preoperative examination indicator has more obvious clustering, that is, it presents an obvious multi-peak distribution feature, then the preoperative examination indicator is more important for analyzing the intraoperative anesthesia risk of the reference patient, and the weight factor is larger; according to the preoperative physical sign data and the weight factor of each preoperative examination indicator, combined with the intraoperative risk index of each reference patient, the anesthesia risk of the current patient to be examined is evaluated. The present invention evaluates the intraoperative risk of historical patients by analyzing the changes in intraoperative monitoring data, and then measures the weight factor of each preoperative examination indicator with the help of the distribution characteristics of the number of historical patients under different physical sign parameters under the preoperative examination indicators. The weight factor is then combined to accurately evaluate historical patients with similar preoperative physical sign data, i.e., physical constitution. The intraoperative risk of the current patient can be evaluated with the help of the intraoperative risk of the historical patient, thereby improving the accuracy of the preoperative anesthesia risk assessment.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

FIG1 is a flow chart of a method for evaluating anesthesia before painless gastroenteroscopic diagnosis and treatment according to an embodiment of the present invention;

FIG. 2 is a flow chart of a method for obtaining intraoperative risk indicators provided by an embodiment of the present invention.

DETAILED DESCRIPTION

In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the following is a detailed description of the method and system for preoperative anesthesia assessment for painless gastroenteroscopic diagnosis and treatment proposed by the present invention, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

The specific scheme of the method and system for preoperative anesthesia assessment for painless gastroenteroscopic diagnosis and treatment provided by the present invention is described in detail below with reference to the accompanying drawings.

Please refer to FIG. 1 , which shows a flow chart of a method for preoperative anesthesia assessment for painless gastroenteroscopic diagnosis and treatment provided by an embodiment of the present invention, which specifically includes:

Step S1, obtain the preoperative physical sign data and intraoperative monitoring data of all patients who have undergone historical examinations of painless gastroenteroscopy, and obtain the preoperative physical sign data of the current patient to be examined; the preoperative physical sign data includes the physical sign parameters of each preoperative examination indicator, and the intraoperative monitoring data includes all dynamic parameter sequences of each intraoperative monitoring indicator.

In one embodiment of the present invention, first, the physical examination records of all patients who have undergone painless gastroscopy before the painless gastroscopy are retrieved from the hospital database, and the preoperative physical sign data of each historical examination patient is constructed based on the physical examination records; at the same time, the preoperative physical sign data of the current patient to be examined who is about to undergo painless gastroscopy is constructed based on the physical examination records; and the physical sign monitoring records of each historical examination patient obtained by using relevant monitoring instruments during the painless gastroscopy are retrieved, and the intraoperative monitoring data is constructed based on the physical sign monitoring records.

It should be noted that the anesthesia assessment method for each patient currently under examination is the same, and only one of them is used for description here. Intraoperative monitoring data can be used to evaluate the anesthetic effect and safety risk of patients during sedation and anesthesia, and preoperative physical sign data reflects the physical fitness of each patient; both are obtained in order to subsequently examine the intraoperative anesthesia risk of patients with similar physical fitness history and evaluate the intraoperative anesthesia risk of the current patient under examination.

Among them, the preoperative physical sign data include the physical sign parameters of each preoperative examination indicator, and the preoperative examination indicators at least include age, height, weight, blood pressure, blood routine, liver and kidney function, electrocardiogram, etc.; the intraoperative monitoring data contains all dynamic parameter sequences of each intraoperative monitoring indicator, and the intraoperative monitoring indicators at least include blood oxygen, blood pressure, heart rate, etc. Since the operation is a continuous process, the monitoring result of each intraoperative monitoring indicator is a dynamic parameter sequence; the dynamic parameter sequence at least includes a blood oxygen sequence, a blood pressure sequence, and a heart rate sequence. For example, the construction process of the blood pressure sequence is: during the painless gastrointestinal endoscopy of patients with historical examinations, the blood pressure parameters are collected at a preset sampling frequency, and the blood pressure parameters are sorted in the order of collection to construct a sequence; the construction of the blood oxygen sequence and heart rate sequence is the same; the preset sampling frequency is 1s each time, and the implementer can also set it by themselves.

It should be noted that routine blood tests and liver and kidney function tests may include a variety of laboratory indicators. For example, routine blood tests mainly include red blood cell count, hemoglobin, white blood cell count and platelets, liver and kidney function tests include total protein, albumin, globulin, and kidney function tests include serum creatinine, serum uric acid and urea nitrogen. Electrocardiograms are in the form of pictures, so numerical annotations are made based on the evaluation results of electrocardiograms by doctors in relevant departments as corresponding physical sign parameters, such as myocardial ischemia marked as 0, normal marked as 1, and arrhythmia marked as 2. Each of the laboratory indicators and electrocardiogram can be used as a preoperative examination indicator separately, and the physical sign parameters corresponding to each preoperative examination indicator can be used as vector elements to construct a multidimensional vector, thereby obtaining preoperative physical sign data. Similarly, the dynamic parameter sequence of each intraoperative monitoring indicator in the intraoperative physical sign monitoring record of historically examined patients can be used as a vector element to construct a multidimensional vector, thereby obtaining intraoperative monitoring data.

It should be noted that before constructing a multidimensional vector, all vital sign parameters and parameters in the dynamic parameter sequence need to be standardized to eliminate the dimension for subsequent analysis. This is already a prior art and will not be elaborated here. In other embodiments, the implementer can also customize other types or numbers of preoperative examination indicators and intraoperative monitoring indicators, which are not limited here.

Step S2, screen out all reference patients for the current patient to be examined from all historically examined patients based on preoperative physical sign data; under each intraoperative monitoring indicator, obtain the intraoperative risk indicator for each reference patient based on the fluctuation of the dynamic parameter sequences of all reference patients and the difference in changes between the corresponding dynamic parameter sequences of different reference patients.

Considering that among many patients undergoing painless gastroenteroscopy, their physical fitness may be different, resulting in different tolerance of each patient to sedative anesthetics; people with similar physical fitness may have more similar tolerance. In order to conduct an anesthetic risk assessment for the current patient to be examined, the embodiment of the present invention first screens out all reference patients for the current patient to be examined from all historical examination patients based on preoperative physical sign data. The reference patients are historical examination patients with similar physical fitness to the current patient to be examined who are initially screened.

Preferably, in one embodiment of the present invention, considering that clustering can cluster similar preoperative physical sign data into a cluster, and thus patients with similar physical fitness can be obtained; therefore, the method for obtaining reference patients includes:

All historically examined patients and the current patient to be examined are taken as the first patients to be classified; the corresponding metric distance is obtained according to the similarity between the corresponding preoperative physical sign data of different first patients to be classified; all patient clusters are obtained based on the clustering algorithm and the metric distance between the preoperative physical sign data; all historically examined patients within the patient cluster to which the current patient to be examined belongs are taken as all reference patients for the current patient to be examined.

As an example, the Euclidean distance between the corresponding preoperative physical sign data of different first patients to be classified is used as the metric distance between different first patients to be classified, and all patient clusters are obtained based on the K-means clustering algorithm and the preset K value; the preset K value is 9, and the implementer can also set it by himself; all historical examination patients belonging to the same patient cluster as the current patient to be examined are used as reference patients. It should be noted that the K-means clustering algorithm is a well-known technology, and the implementer can also use other clustering algorithms such as CURE hierarchical clustering, which will not be repeated here.

Taking into account the historical examination of patients during the anesthesia process, the corresponding parameters of various intraoperative monitoring indicators should change smoothly. If there are more drastic fluctuations, such as a sharp rise or fall in parameters, it means that the patient's anesthetic effect under the current anesthesia scheme is not good, which may lead to safety risks such as waking up due to shallow anesthesia or coma due to deep anesthesia; and considering that all reference patients with similar physical fitness should have similar fluctuations in the corresponding parameters during anesthesia, if the fluctuation of the intraoperative monitoring data of a reference patient is more drastic than that of other patients, it means that the reference patient is more sensitive and his anesthesia risk is greater; therefore, under each intraoperative monitoring indicator, the embodiment of the present invention obtains the intraoperative risk index of each reference patient according to the fluctuation of the dynamic parameter sequence of all reference patients, combined with the difference in changes between the corresponding dynamic parameter sequences of different reference patients; the intraoperative risk index reflects the degree of anesthesia risk of each reference patient during painless gastrointestinal endoscopy.

Preferably, in one embodiment of the present invention, the method for obtaining the intraoperative risk indicator includes:

Please refer to FIG. 2 , which shows a flow chart of a method for obtaining intraoperative risk indicators provided by an embodiment of the present invention, including:

Step S201, taking any intraoperative monitoring indicator as a target monitoring indicator, and obtaining the sensitivity weight of the target monitoring indicator according to all adjacent extreme value differences in the dynamic parameter sequences of all reference patients under the target monitoring indicator.

In a preferred embodiment of the present invention, considering that under each intraoperative monitoring indicator, if the difference between adjacent extreme values in the dynamic parameter sequence is larger and the variance is larger, it means that the intraoperative monitoring indicator fluctuates more violently; at the same time, if the dynamic parameter sequence under the intraoperative monitoring indicator of all reference patients fluctuates violently, it further indicates that the intraoperative monitoring indicator is more important in assessing the intraoperative anesthesia risk; therefore, the method for obtaining the sensitive weight includes:

In the intraoperative monitoring data of each reference patient, the cumulative result of the change differences of all adjacent extreme values in the dynamic parameter sequence under the target monitoring indicator is taken as the first fluctuation parameter, and the variance of the dynamic parameter sequence under the target monitoring indicator is taken as the second fluctuation parameter; the product of the first fluctuation parameter and the second fluctuation parameter is taken as the fluctuation parameter of the target monitoring indicator in the intraoperative monitoring data of the corresponding reference patient; the mean of all fluctuation parameters of the target monitoring indicator in the intraoperative monitoring data of all reference patients is normalized, and the normalized result is taken as the sensitive weight of the target monitoring indicator.

As an example, the calculation formula for sensitivity weight is: ;in, is the serial number of the target monitoring indicator; For the The sensitivity weight of each target monitoring indicator; is the serial number of the reference patient; is the total number of reference patients; is the sequence number of the absolute value of the difference between adjacent extreme values; is the total number of absolute values of differences between adjacent extreme values in the dynamic parameter sequence; For the Under the target monitoring indicators The first in the dynamic parameter sequence of the reference patient The absolute value of the difference between adjacent extreme values; is the first fluctuation parameter; For the Under the target monitoring indicators The variance of all parameters in the dynamic parameter sequence of a reference patient is also the second fluctuation parameter; is the standard normalization function.

It should be noted that, in this example, a linear normalization method is specifically adopted, and other normalization methods may also be adopted; the acquisition of extreme values is already an existing technology and will not be elaborated here; in the calculation formula of the sensitivity weight, the larger the cumulative value of the absolute value of the difference between all adjacent extreme values, the larger the first fluctuation parameter, indicating that the dynamic parameter sequence fluctuates violently and the fluctuation amplitude is large; the larger the variance, the larger the second fluctuation parameter, further indicating that the dynamic parameter sequence fluctuates violently; the product of the two is a more comprehensive assessment of the fluctuation of the dynamic parameter sequence under each intraoperative monitoring indicator. The larger the fluctuation, the more important the intraoperative monitoring indicator is for assessing the anesthesia risk of all patients, and the larger the sensitivity weight.

Step S202, taking any reference patient as the target patient, under the target monitoring index, according to the difference in changes between the corresponding dynamic parameter sequences of the target patient and each other reference patient, obtain the fluctuation deviation between the target patient and each other reference patient; taking the mean of all the fluctuation deviations as the index deviation of the target patient relative to all other reference patients under the target monitoring index.

As an example, the formula for calculating the indicator deviation is: ;in, is the serial number of the target patient; is the serial number of the target monitoring indicator; is the serial number of other reference patients except the target patient; is the total number of reference patients; For the Under the target monitoring indicators The target patient and The difference in changes between the dynamic parameter sequences corresponding to the reference patients is also called the fluctuation deviation; For the The target patient is ranked relative to all other reference patients. The indicator deviation under each target monitoring indicator.

Among them, the method for obtaining the fluctuation deviation is: converting the dynamic parameter sequence under each intraoperative monitoring indicator into a character string through the SAX algorithm, wherein each parameter in the dynamic parameter sequence corresponds to a character; taking the edit distance between the character strings of the two dynamic parameter sequences corresponding to the target patient and each reference patient as the fluctuation deviation; in other examples, the implementer may also use other measurement methods of the difference in changes between sequences to obtain the fluctuation deviation, such as directly taking the DTW distance between the two dynamic parameter sequences as the fluctuation deviation, which will not be elaborated here.

The indicator deviation comprehensively evaluates the special sensitivity of the target patient by comparing the dynamic parameter sequence differences of the target patient with all other reference patients, which facilitates the subsequent evaluation of the target patient's intraoperative risk indicators in combination with the sensitivity weight.

It should be noted that the SAX algorithm, the calculation of the edit distance and the DTW distance are already existing technologies and will not be described in detail here.

Step S203, using the sensitivity weight of each intraoperative monitoring indicator to weighted average the corresponding indicator deviation, normalizing the weighted average result, and obtaining the intraoperative risk indicator of the target patient.

As an example, the weighted average result is specifically mapped to the hyperbolic tangent function. Since the weighted average result is not a negative value, the mapping result is between 0 and 1, and normalization is then achieved. The normalized result is used as the intraoperative risk indicator for the target patient.

It should be noted that weighted averaging is a well-known method and will not be described in detail here; implementers may also use other normalization methods, which will not be described in detail here.

Step S3, under each preoperative examination indicator, according to the number distribution of reference patients corresponding to different physical sign parameters, obtain the indicator distribution model under each preoperative examination indicator; comprehensively analyze the number distribution of reference patients corresponding to different physical sign parameters under all preoperative examination indicators, and the variation range of physical sign parameters under each preoperative examination indicator, to obtain the fusion distribution model; according to the difference between the fusion distribution model and the indicator distribution model, obtain the weight factor of the corresponding preoperative examination indicator.

Although intraoperative monitoring data can reflect the patient's anesthesia risk more intuitively than preoperative physical signs, the preoperative physical signs data indirectly reflect the patient's anesthesia risk before the operation, that is, the preoperative physical signs data affect the intraoperative monitoring data to a certain extent; but different preoperative examination indicators have different effects on the patient's anesthesia risk. For example, under the premise that other preoperative examination indicators are consistent, patients with lighter weight have lower body fat and muscle content, which affects drug metabolism, and their anesthesia risk is higher than that of patients with normal standard weight; therefore, the feasibility of the present invention requires obtaining the weight factor of each preoperative examination indicator to facilitate the subsequent evaluation of intraoperative risk.

Considering that for each preoperative examination indicator, the intraoperative risk indicator of the patient under different physical sign parameters should be presented as follows: the lower or higher the physical sign parameter, the greater the intraoperative risk; and when the physical sign parameter is in a certain intermediate range, the intraoperative risk is relatively smaller; and considering that the number of reference patients under different physical sign parameters under each preoperative examination indicator is different, a physical sign parameter-patient number model can be obtained, that is, the indicator distribution model, and the indicator distribution model should present a single-peak or multi-peak distribution curve; the multi-peak is due to the obvious clustering of the number of reference patients under different physical sign parameters, that is, each peak may approximately correspond to a reference patient group with similar intraoperative risk; and the single peak is due to the weak clustering of the intraoperative risk of the reference patient under different physical sign parameters, that is, the lower the reference value of different physical sign parameters for assessing intraoperative risk; the indicator distribution model indirectly reflects the distribution of reference patients under different intraoperative risk indicators, and the indicator distribution model of different preoperative examination indicators should have distinctive model characteristics.

Considering that when all preoperative examination indicators are integrated, the distribution of reference patients should also have a certain degree of clustering, that is, when all preoperative examination indicators are integrated and the preoperative physical sign data is regarded as a data point, when analyzing the clustering of reference patients under different data points, it should normally present a multi-peak distribution feature; but due to the high dimensionality of different preoperative physical sign data, there may be a certain dimensional disaster when analyzing the distribution of reference patients, that is, the number of reference patients corresponding to each data point may be very small, and it is difficult to analyze the clustering; therefore, the embodiment of the present invention will integrate the distribution of the number of reference patients corresponding to different physical sign parameters under all preoperative examination indicators, and then combine the variation range of the physical sign parameters under each preoperative examination indicator to fuse the different distributions and obtain a fusion distribution model; by adjusting and fusing the distribution of all preoperative examination indicators, a fusion distribution model is obtained to avoid multi-dimensional disasters, and at the same time, combined with the extensiveness of the distribution of physical sign parameters of different preoperative examination indicators, the wider the distribution, the more important the distribution characteristics corresponding to the analysis of fitting multi-peak features, so that the clustering of reference patients can be accurately evaluated.

If the difference between the indicator distribution model of a single preoperative examination indicator and the fusion distribution model of multiple preoperative examination indicators is smaller, it means that the number distribution of reference patients under the preoperative examination indicator has more obvious clustering, that is, it presents an obvious multi-peak distribution feature. In this case, the more important the preoperative examination indicator is for analyzing the intraoperative anesthesia risk of reference patients, the greater the weight factor of the preoperative examination indicator.

In one embodiment of the present invention, the method for obtaining the index distribution model includes: constructing a two-dimensional coordinate system under each preoperative examination index with the size of the physical sign parameter as the horizontal axis and the total number of reference patients as the vertical axis; mapping the total number of reference patients under each physical sign parameter to the two-dimensional coordinate system and fitting the curve to obtain the index distribution model under each preoperative examination index. The construction concept is the same as that of the probability distribution model, which will not be repeated here.

Preferably, in one embodiment of the present invention, considering that there are differences in the value ranges of the physical sign parameters of different preoperative examination indicators, direct fusion may lead to unsatisfactory fusion results; therefore, the preoperative physical sign parameters need to be adjusted for fusion; and considering that the physical sign parameters of reference patients with lower intraoperative risk indicators can be used as adjustment references, the value ranges of the physical sign parameters of different preoperative indicators are mapped and scaled with the help of the standardization concept to be approximately consistent, thereby obtaining an ideal fusion effect;

Based on this, the method for obtaining the fusion distribution model includes:

All low-risk patients are screened from all reference patients according to intraoperative risk indicators; under each preoperative examination indicator, the characteristic physical sign parameters of each preoperative examination indicator are determined according to the distribution of physical sign parameters of all low-risk patients, and all physical sign parameters of all reference patients are standardized respectively using the characteristic physical sign parameters to obtain adjusted physical sign parameters;

Under each preoperative examination indicator, the distribution model under each preoperative examination indicator is obtained according to the distribution of the number of reference patients corresponding to different adjusted physical sign parameters; the distribution models under all preoperative examination indicators and the variation range of the adjusted physical sign parameters under each preoperative examination indicator are combined to obtain the fusion distribution model.

The method for obtaining the adjustment vital sign parameters includes:

All reference patients whose intraoperative risk index is less than the preset threshold are regarded as low-risk patients; under each preoperative examination index, the extreme difference of all physical sign parameters corresponding to all low-risk patients is regarded as the extreme sign parameter, and the physical sign parameter corresponding to the largest total number of low-risk patients is regarded as the reference physical sign parameter; under each preoperative examination index, the extreme sign parameter is regarded as the denominator, the difference between the physical sign parameter of each reference patient and the reference physical sign parameter is regarded as the numerator, and the fractional ratio is regarded as the adjusted physical sign parameter of the corresponding physical sign parameter.

As an example, the preset threshold is 0.65 to screen out all low-risk patients. The implementer can also set the preset threshold by himself. Under each preoperative examination indicator, the physical sign parameters of all low-risk patients are sorted in ascending order, and the total number of low-risk patients corresponding to each physical sign parameter is counted, and the maximum and minimum physical sign parameters are screened out. At the same time, the physical sign parameters corresponding to the largest total number of low-risk patients are used as reference physical sign parameters.

The calculation formula for adjusting the vital signs parameters is: ;in, It is the serial number of the preoperative examination index; is the serial number of the reference patient; For the Preoperative examination indicators Adjusted vital sign parameters for reference patients; For the Preoperative examination indicators The vital signs and parameters of the reference patients; For the Reference physical sign parameters under the preoperative examination indicators; For the The maximum physical sign parameters under the preoperative examination indicators; For the The minimum physical sign parameters under the preoperative examination indicators; It is an extremely poor physical sign parameter.

It should be noted that due to the diversity of reference patients, the possibility that the maximum sign parameter is equal to the minimum sign parameter is extremely low, that is, the possibility that the extreme difference sign parameter is 0 when used as the denominator is extremely small, so the case where the denominator is zero is ignored.

The calculation formula for adjusting the vital sign parameters uses the idea of maximum and minimum value normalization to adjust all the vital sign parameters under each preoperative examination indicator, thereby obtaining the adjusted vital sign parameters; the difference between the calculation formula for adjusting the vital sign parameters and the calculation formula for the maximum and minimum values lies in the difference in the numerators. The embodiment of the present invention uses the difference between the vital sign parameters and the reference vital sign parameters to replace the difference between the original vital sign parameters and the minimum vital sign parameters, and centers the vital sign parameters, so that all the adjusted vital sign parameters are distributed around the reference vital sign parameters corresponding to the low-risk patients, which can better reflect that the reference patients whose adjusted vital sign parameters are in a certain intermediate range have lower intraoperative risks, so as to facilitate the subsequent construction of the distribution model.

The method for obtaining the fusion distribution model includes:

Under each preoperative examination indicator, a two-dimensional coordinate system is constructed with the size of the adjusted physical sign parameter as the horizontal axis and the total number of reference patients as the vertical axis; the total number of reference patients under the corresponding physical sign parameter before adjustment of each adjusted physical sign parameter is mapped to the two-dimensional coordinate system and a curve is fitted to obtain a distribution model under each preoperative examination indicator;

Under each preoperative examination index, the range of all adjusted physical sign parameters was normalized as the vertical axis weight, and the vertical axis weight was used to perform weighted adjustment on the total number of reference patients corresponding to each adjusted physical sign parameter in the distribution model to obtain the weighted total number of reference patients corresponding to each adjusted physical sign parameter;

Under each adjusted sign parameter, the corresponding weighted total number of all preoperative examination indicators is accumulated to obtain the adjusted number of reference patients under each adjusted sign parameter; the adjusted number of reference patients under each adjusted sign parameter is mapped to the newly constructed two-dimensional coordinate system and the curve is fitted to obtain the fusion distribution model.

As an example, first obtain the distribution model. The method for obtaining the distribution model is the same as the construction idea of the probability distribution model, which will not be repeated here. Since the wider the value range of the adjustment sign parameter under each preoperative examination indicator, the more important its distribution characteristics are in fitting the multi-peak characteristics, the sum of the ranges of the adjustment sign parameters under all preoperative examination indicators is used as the denominator, and the range of the adjustment sign parameters under each preoperative examination indicator is used as the numerator, and normalized to obtain the vertical axis weight. The vertical axis weight indicates the importance of the distribution model under the preoperative examination indicator to the subsequent construction of the fusion distribution model. Then, using the corresponding vertical axis weight, adjust the total number of reference patients corresponding to each adjustment sign parameter in the distribution model to obtain the weighted total number.

Since the sign parameters under all preoperative sign indicators have been standardized, the adjusted sign parameters under different preoperative sign indicators can be fused, that is, a two-dimensional coordinate system is reconstructed, with the horizontal axis representing the size of the adjusted sign parameter and the vertical axis representing the number of patients. The corresponding weighted total quantities under all preoperative examination indicators under each adjusted sign parameter are added up to obtain a new adjusted quantity, which is then mapped to the two-dimensional coordinate system as the vertical axis height, and a curve is fitted to obtain a new fused distribution model. That is, the construction process of the fused distribution model can be regarded as the weighted fusion result of the distribution model under each preoperative examination indicator.

It should be noted that the total number of reference patients corresponding to each adjusted sign parameter in the distribution model is equal to the total number of reference patients under the corresponding sign parameter before adjustment; the total number of reference patients corresponding to each adjusted sign parameter in the fusion distribution model is the weighted result of the total number of reference patients corresponding to each adjusted sign parameter in each distribution model.

Preferably, in one embodiment of the present invention, the method for obtaining the weight factor includes:

The DTW distance between the fusion distribution model and the indicator distribution model under each preoperative examination indicator is taken as the x in the exponential function exp(-x) with the natural constant e as the base, and the exponential function value is taken as the weight factor of the corresponding preoperative examination indicator.

It should be noted that DTW distance is a well-known technology and will not be described in detail here. The smaller the DTW distance, the greater the similarity, so it is negatively correlated and mapped. Other negative correlation mapping methods can also be used, such as performing an inverse operation on it, to perform negative correlation normalization. Implementers can also use other distance difference or similarity measurement methods to obtain weight factors.

Step S4, based on the preoperative physical sign data and the weight factor of each preoperative examination indicator, combined with the intraoperative risk indicator of each reference patient, the anesthesia risk of the current patient to be examined is evaluated.

After obtaining the weight factor of each preoperative examination indicator, the weight factor can be further combined to accurately evaluate the similarity between the preoperative physical sign data, thereby evaluating the anesthesia risk of the current patient to be examined with the help of the intraoperative risk indicators of the reference patients with similar preoperative physical sign data.

Preferably, in one embodiment of the present invention, the method for assessing the anesthesia risk of the current patient to be examined comprises:

All reference patients and the current patient to be tested are taken as the second patients to be classified; based on the weight factor of each preoperative examination indicator, the weighted Euclidean distance between the preoperative physical sign data of different second patients to be classified is obtained; based on the density clustering algorithm and the weighted Euclidean distance, all sample patient clusters are obtained; all reference patients in the sample patient cluster to which the current patient to be tested belongs are taken as all target patients of the current patient to be tested;

A training sample set is constructed based on the preoperative physical sign data and intraoperative risk indicators of all target patients, and the anesthesia risk assessment model is trained using the training sample set; the preoperative physical sign data of the current patient to be examined is input into the trained anesthesia risk assessment model, and the estimated intraoperative risk indicators of the current patient to be examined are output.

As an example, the weighted Euclidean distance between the corresponding preoperative physical sign data of different second patients to be classified is used as the metric distance between different second patients to be classified, and all second patients to be classified are clustered based on the DBSCAN clustering algorithm, wherein the evaluation and calculation of algorithm parameters such as the preset radius distance ε and the preset minimum number of neighborhood points minPts are all existing technologies, and the implementer may also use other clustering algorithms, which are not described here; all reference patients in the sample patient cluster to which the current patient to be tested belongs are taken as target patients; the target patient is a reference patient with similar physical sign data to the current patient to be tested obtained by combining different weight factors;

The preoperative physical sign data of all target patients and the corresponding intraoperative risk indicators are further used to construct a training sample set, and the target patients with a physique more similar to the current patient to be examined are more accurately screened out through weight factors, which helps to improve the training effect of the model; the neural network model is a convolutional neural network, and the convolutional neural network model is trained using the training sample set to obtain a trained anesthesia risk assessment model, and then the preoperative physical sign data of the current patient to be examined can be input into the trained anesthesia risk assessment model, and the estimated surgical risk indicators of the current patient to be examined are output, thereby assisting relevant anesthesiologists in analyzing, evaluating and formulating anesthesia plans for the current patient to be examined.

In summary, the present invention first obtains the preoperative physical sign data and intraoperative monitoring data of all historical patients undergoing painless gastroenteroscopy, and obtains the preoperative physical sign data of the current patient to be examined, and then screens out all reference patients of the current patient to be examined; then analyzes the intraoperative monitoring data of all reference patients to obtain the intraoperative risk index of each reference patient; further analyzes the preoperative physical sign data of all reference patients to evaluate the weight factor of each preoperative examination index; and then evaluates the anesthesia risk of the current patient to be examined based on the preoperative physical sign data and the weight factor of each preoperative examination index, combined with the intraoperative risk index of each reference patient. The present invention evaluates the intraoperative risk of historical patients by analyzing the changes in intraoperative monitoring data of historical patients, and then measures the weight factor of each preoperative examination index with the help of the distribution characteristics of the number of historical patients under different physical sign parameters under the preoperative examination index, so as to accurately evaluate historical patients with similar preoperative physical sign data, i.e., physical constitution, in combination with the weight factor, and then can evaluate the intraoperative risk of the current patient with the help of the intraoperative risk of historical patients, thereby improving the accuracy of preoperative anesthesia risk assessment.

The present invention also proposes a preoperative anesthesia assessment system for painless gastrointestinal endoscopic diagnosis and treatment. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned preoperative anesthesia assessment method for painless gastrointestinal endoscopic diagnosis and treatment are implemented.

It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims (10)

1. A method for evaluating anesthesia before painless gastroenteroscopic diagnosis and treatment, characterized in that the method comprises: Obtaining preoperative physical sign data and intraoperative monitoring data of all patients who have undergone painless gastroenteroscopy in history, and obtaining preoperative physical sign data of the current patient to be examined; the preoperative physical sign data includes physical sign parameters of each preoperative examination indicator, and the intraoperative monitoring data includes all dynamic parameter sequences of each intraoperative monitoring indicator; All reference patients of the patient to be examined are screened from all historically examined patients according to the preoperative physical sign data; under each intraoperative monitoring index, according to the fluctuation changes of the dynamic parameter sequences of all reference patients, combined with the change differences between the dynamic parameter sequences corresponding to different reference patients, the intraoperative risk index of each reference patient is obtained; Under each preoperative examination indicator, according to the quantity distribution of the reference patients corresponding to different physical sign parameters, the indicator distribution model under each preoperative examination indicator is obtained; the fusion distribution model is obtained by comprehensively analyzing the quantity distribution of the reference patients corresponding to different physical sign parameters under all preoperative examination indicators and the variation range of the physical sign parameters under each preoperative examination indicator; according to the difference between the fusion distribution model and the indicator distribution model, the weight factor of the corresponding preoperative examination indicator is obtained; According to the preoperative physical sign data and the weight factor of each preoperative examination indicator, combined with the intraoperative risk indicator of each reference patient, the anesthesia risk of the current patient to be examined is evaluated. 2. A method for evaluating anesthesia before painless gastroenteroscopic diagnosis and treatment according to claim 1, characterized in that the method for obtaining the reference patient comprises: All patients examined in the past and the patients to be examined currently are regarded as first patients to be classified; and corresponding metric distances are obtained according to the similarities between the preoperative physical sign data corresponding to different first patients to be classified; Based on the clustering algorithm and the metric distance between the preoperative physical sign data, all patient clusters are obtained; all historically examined patients in the patient cluster to which the current patient to be examined belongs are used as all reference patients for the current patient to be examined. 3. A method for preoperative anesthesia assessment for painless gastroenteroscopic diagnosis and treatment according to claim 1, characterized in that the method for obtaining the intraoperative risk index comprises: Taking any intraoperative monitoring indicator as a target monitoring indicator, under the target monitoring indicator, obtaining a sensitivity weight of the target monitoring indicator according to all adjacent extreme value differences in the dynamic parameter sequences of all the reference patients; Taking any of the reference patients as the target patient, under the target monitoring index, according to the change difference between the target patient and each of the other reference patients corresponding to the dynamic parameter sequence, obtain the fluctuation deviation between the target patient and each of the other reference patients; taking the mean of all the fluctuation deviations as the index deviation of the target patient relative to all the other reference patients under the target monitoring index; The sensitivity weight of each intraoperative monitoring indicator is used to weight the deviation of the corresponding indicator and the weighted average result is normalized to obtain the intraoperative risk indicator of the target patient. 4. A method for evaluating anesthesia before painless gastroenteroscopic diagnosis and treatment according to claim 3, characterized in that the method for obtaining the sensitivity weight comprises: In the intraoperative monitoring data of each of the reference patients, the cumulative result of the change differences of all adjacent extreme values in the dynamic parameter sequence under the target monitoring index is taken as the first fluctuation parameter, and the variance of the dynamic parameter sequence under the target monitoring index is taken as the second fluctuation parameter; the product of the first fluctuation parameter and the second fluctuation parameter is taken as the fluctuation parameter of the target monitoring index in the intraoperative monitoring data corresponding to the reference patient; The mean values of all the fluctuation parameters of the target monitoring index in the intraoperative monitoring data of all the reference patients are normalized, and the normalized result is used as the sensitivity weight of the target monitoring index. 5. The method for preoperative anesthesia assessment for painless gastroenteroscopic diagnosis and treatment according to claim 1, characterized in that the method for obtaining the fusion distribution model comprises: Screening all low-risk patients from all the reference patients according to the intraoperative risk index; determining characteristic physical sign parameters of each preoperative examination index according to the distribution of physical sign parameters of all the low-risk patients under each preoperative examination index, and using the characteristic physical sign parameters to standardize all physical sign parameters of all the reference patients respectively to obtain adjusted physical sign parameters; Under each preoperative examination indicator, the distribution model under each preoperative examination indicator is obtained according to the distribution of the number of reference patients corresponding to different adjusted sign parameters; the distribution models under all preoperative examination indicators and the variation range of the adjusted sign parameters under each preoperative examination indicator are combined to obtain a fusion distribution model. 6. A method for preoperative anesthesia assessment for painless gastroenteroscopic diagnosis and treatment according to claim 5, characterized in that the method for obtaining the adjustment sign parameters comprises: All the reference patients whose intraoperative risk index is less than the preset threshold are regarded as low-risk patients; under each preoperative examination index, the extreme difference among all the physical sign parameters corresponding to all the low-risk patients is regarded as the extreme physical sign parameter, and the physical sign parameter corresponding to the maximum total number of the low-risk patients is regarded as the reference physical sign parameter; Under each preoperative examination indicator, the extreme sign parameter is used as the denominator, the difference between the sign parameter of each reference patient and the reference sign parameter is used as the numerator, and the fractional ratio is used as the adjusted sign parameter of the corresponding sign parameter. 7. A method for evaluating anesthesia before painless gastroenteroscopic diagnosis and treatment according to claim 5, characterized in that the method for obtaining the fusion distribution model comprises: Under each preoperative examination indicator, a two-dimensional coordinate system is constructed with the size of the adjusted physical sign parameter as the horizontal axis and the total number of reference patients as the vertical axis; the total number of reference patients under the corresponding physical sign parameter of each adjusted physical sign parameter before adjustment is mapped to the two-dimensional coordinate system and a curve is fitted to obtain a distribution model under each preoperative examination indicator; Under each preoperative examination indicator, the range of all the adjusted physical sign parameters is normalized as the vertical axis weight, and the vertical axis weight is used to perform weighted adjustment on the total number of reference patients corresponding to each adjusted physical sign parameter in the distribution model to obtain the weighted total number of reference patients corresponding to each adjusted physical sign parameter; Under each of the adjusted physical sign parameters, the corresponding weighted total quantities under all preoperative examination indicators are accumulated to obtain the adjusted quantity of reference patients under each of the adjusted physical sign parameters; the adjusted quantity of reference patients under each of the adjusted physical sign parameters is mapped to the newly constructed two-dimensional coordinate system and the curve is fitted to obtain a fusion distribution model. 8. A method for evaluating anesthesia before painless gastroenteroscopic diagnosis and treatment according to claim 1, characterized in that the method for obtaining the weight factor comprises: The DTW distance between the fusion distribution model and the indicator distribution model under each preoperative examination indicator is used as the x in the exponential function exp(-x) with the natural constant e as the base, and the exponential function value is used as the weight factor of the corresponding preoperative examination indicator. 9. A method for evaluating anesthesia before painless gastroenteroscopic diagnosis and treatment according to claim 1, characterized in that the method for evaluating the anesthesia risk of the patient to be examined comprises: All reference patients and the current patient to be examined are taken as the second patients to be classified; based on the weight factor of each preoperative examination indicator, the weighted Euclidean distance between the preoperative physical sign data of different second patients to be classified is obtained; based on the density clustering algorithm and the weighted Euclidean distance, all sample patient clusters are obtained; all reference patients in the sample patient cluster to which the current patient to be examined belongs are taken as all target patients of the current patient to be examined; A training sample set is constructed based on the preoperative physical sign data of all the target patients and the intraoperative risk indicators, and the anesthesia risk assessment model is trained using the training sample set; the preoperative physical sign data of the current patient to be examined is input into the trained anesthesia risk assessment model, and the estimated intraoperative risk indicator of the current patient to be examined is output. 10. A system for evaluating anesthesia before painless gastroenteroscopy, characterized in that the system comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the system implements the steps of a method for evaluating anesthesia before painless gastroenteroscopy as described in any one of claims 1 to 9.