Liver and gall patient risk early warning method and system based on deep learning

Abstract

The present application relates to the technical field of liver and gallbladder patient physical data prediction, and particularly relates to a liver and gallbladder patient risk early warning method and system based on deep learning. The present application obtains a lesion parameter according to the change of each dimension physical data in a long-term time; obtains a dimension influence degree according to the lesion parameter distribution of liver cancer patients and cirrhosis patients in each dimension; obtains a lesion similarity degree according to the similarity of the to-be-analyzed patient and the liver cancer patient in each dimension, and obtains a prediction parameter of the to-be-analyzed patient in combination with the dimension influence degree and the lesion parameter. The risk of the to-be-analyzed patient is predicted according to the prediction parameter. The present application analyzes the physical data of each dimension of the long-term cirrhosis patient, takes into account the complexity of the judgment process and the relevance of the physical data of each dimension, so as to accurately predict the risk of the long-term cirrhosis patient.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hepatobiliary patient physical data prediction, and particularly relates to a hepatobiliary patient risk early warning method and system based on deep learning. BACKGROUND

[0002] In the risk early warning of hepatobiliary diseases, not only image data is widely used, but also multi-source data such as patient clinical indicators and gene data. For example, some research uses patient gene sequencing data and deep learning technology to find gene sites closely related to liver cancer risk, providing new biomarkers for early warning. Deep learning has achieved certain results in liver cancer postoperative recurrence and survival risk prediction. For example, some research has constructed a postoperative recurrence prediction model by analyzing a large amount of clinical and pathological data of liver cancer patients using deep learning algorithm. In summary, the hepatobiliary patient risk early warning method based on deep learning has achieved remarkable results in technical application, research progress and clinical practice, and has shown broad development prospects. With the continuous progress of technology and the deepening of interdisciplinary cooperation, this method is expected to provide stronger support for the early detection, early treatment and personalized management of hepatobiliary diseases.

[0003] In the process of liver cirrhosis patients developing into liver cancer, the physical data of each dimension of the body is too complex, and the change trend of different dimension physical data also has a big difference. Only by judging through individual dimensions, the judging process is too single and the accuracy is poor, which leads to that relevant personnel cannot accurately risk early warning for liver cirrhosis patients. SUMMARY

[0004] To address the problem that the complex and varied physical data of cirrhosis patients progressing to liver cancer, and the significant differences in the trends of these data across different dimensions, relying solely on a single dimension for assessment is overly simplistic and inaccurate, hindering accurate risk warnings for cirrhosis patients, this invention aims to provide a deep learning-based method and system for risk warning of hepatobiliary patients. The specific technical solution is as follows: A deep learning-based method for risk warning of hepatobiliary patients, comprising: acquiring physical data of different long-term cirrhosis patients at each sampling time across different dimensions; selecting one dimension of physical data from a long-term cirrhosis patient as the reference dimension physical data for a reference patient; and analyzing the data of the reference dimension physical data within a preset time period. The study analyzes the following methods: First, it identifies the disease parameters of a reference patient within a predetermined time period, including changes in disease characteristics. Second, it calculates the number of patients with long-term cirrhosis who develop liver cancer. Third, it determines the dimensional influence of the reference dimension on the number of liver cancer patients based on the number of liver cancer patients and the distribution of disease parameters in different long-term cirrhosis patients. Fourth, it obtains the disease parameters of the patient to be analyzed within different dimensions within a predetermined time period. Fifth, it determines the degree of disease similarity between the patient to be analyzed and all liver cancer patients based on the similarity of disease parameters in each dimension. Sixth, it obtains the predicted parameters of the patient to be analyzed based on the dimensional influence of each dimension on the number of liver cancer patients, the disease parameters of the patient to be analyzed in each dimension, and the disease parameters of all long-term cirrhosis patients in each dimension. Finally, it provides risk warnings for the patient to be analyzed based on these predicted parameters.

[0005] Furthermore, the method for obtaining the lesion parameters includes: obtaining the lesion parameters according to a lesion parameter calculation formula, the lesion parameter calculation formula being as follows: In the formula, This indicates the lesion parameters of the reference patient within a preset time period; Indicates the sequence number of the sampling time; This indicates the number of sampling moments within a preset time period; Indicates the first The next sampling time after the previous sampling time; Indicates the first Body data at the sampling time and the first sampling time The slope of the straight line determined by the body data at each sampling time; Indicates the first Body data at the sampling time and the first sampling time The slope of the straight line determined by the body data at each sampling time; This represents the floor function.

[0006] Furthermore, the method for obtaining the degree of influence of the dimension includes: arranging the lesion parameters of the reference dimension for all long-term cirrhosis patients in ascending order within a preset time period to obtain a lesion parameter sequence; dividing the lesion parameter sequence equally to obtain a preset number of lesion parameter intervals; obtaining the probability distribution of the number of long-term cirrhosis patients in each lesion parameter interval and the probability distribution of liver cancer patients in each lesion parameter interval based on the number distribution of long-term cirrhosis patients in each lesion parameter interval; calculating the marginal probability of long-term cirrhosis patients transforming into liver cancer patients; and obtaining the degree of influence of the dimension according to the formula for calculating the degree of influence of the dimension, which is shown below: In the formula, This indicates the number of patients with long-term cirrhosis. Indicates the preset number of disease parameter ranges; This indicates the degree of influence of the reference dimension on the dimension of the number of liver cancer patients. This indicates the number of patients with long-term cirrhosis who develop liver cancer. This represents the number of patients with long-term cirrhosis within each range of pathological parameters. This represents the probability distribution of liver cancer patients in each lesion parameter interval for the reference dimension. This represents the marginal probability that a patient with long-term cirrhosis will develop liver cancer. This represents the probability distribution of the number of people in each lesion parameter range for the reference dimension.

[0007] Furthermore, the method for obtaining the degree of lesion similarity includes: calculating the mean value of lesion parameters in each dimension for all liver cancer patients, as the overall lesion parameters of liver cancer patients in each dimension; taking the lesion parameters in all dimensions of the patient to be analyzed as a first column vector; taking the overall lesion parameters in all dimensions as a second column vector; forming a first matrix by combining the first column vector and the second column vector; and calculating the cosine similarity between the lesion parameters of the patient to be analyzed and the overall lesion parameters of all liver cancer patients, as the degree of lesion similarity between the patient to be analyzed and all liver cancer patients.

[0008] Furthermore, the method for obtaining the prediction parameters includes: obtaining the prediction parameters according to the prediction parameter calculation formula, the prediction parameter calculation formula being as follows: In the formula, Indicates the predictive parameters for the patient to be analyzed; This indicates the number of dimensions in the patient's physical data to be analyzed; Indicates the first The degree of influence of each dimension on the conversion of the number of liver cancer patients; This represents the sum of the influence of all dimensions on the transformation of the number of liver cancer patients; Indicates the patient to be analyzed. Pathological parameters in multiple dimensions; This indicates that all liver cancer patients were at the first stage of treatment. Overall lesion parameters in multiple dimensions; This indicates the degree of lesion similarity between the patient being analyzed and all liver cancer patients; This represents the arctangent function.

[0009] Furthermore, risk warnings are issued for patients to be analyzed based on the predicted parameters, including: when the predicted parameters of the patients to be analyzed are greater than a preset first threshold, the patients to be analyzed are identified as high-risk cirrhosis patients, and risk warnings are issued for the patients to be analyzed.

[0010] Furthermore, the method for obtaining the population distribution probability and the liver cancer patient distribution probability includes: calculating the ratio between the number of long-term cirrhosis patients in each lesion parameter interval and the total number of long-term cirrhosis patients as a reference dimension for the population distribution probability in each lesion parameter interval; and calculating the ratio between the number of liver cancer patients in each lesion parameter interval and the total number of long-term cirrhosis patients as a reference dimension for the liver cancer patient distribution probability in each lesion parameter interval.

[0011] A deep learning-based risk warning system for hepatobiliary patients 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, it implements the steps of the method described above.

[0012] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the deep learning-based risk warning method for hepatobiliary patients described above.

[0013] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the deep learning-based risk warning method for hepatobiliary patients as described above.

[0014] This invention has the following beneficial effects: It acquires physical data from different long-term cirrhosis patients at each sampling time across various dimensions. Since the pathological condition of cirrhosis patients can be reflected through physical data in various dimensions; because in the early stages of cirrhosis, physical data in a certain dimension may only slightly increase over a period of time, but will significantly increase as the cirrhosis worsens, and fluctuations in this dimension within a certain time range are normal physiological fluctuations and have no relation to the degree of cirrhosis, it is necessary to monitor physical data over a longer period and reflect the severity of long-term cirrhosis by referencing the patient's pathological parameters in a reference dimension within a preset time period; because changes in physical data in a certain dimension only indicate a potential progression towards liver cancer, but cannot determine the extent of the impact of that dimension on liver cancer, this invention statistically analyzes how many long-term cirrhosis patients develop liver cancer under various pathological parameters and examines the degree of dimensional influence between changes in physical data in that dimension and the development of liver cancer; by comparing the similar characteristics between the cirrhosis patients to be analyzed and different long-term cirrhosis patients in the big data, predictive parameters for the patients to be analyzed are obtained; and risk warnings for the patients to be analyzed are provided using these predictive parameters. This invention analyzes the body data of patients with long-term cirrhosis in each dimension, taking into account both the complexity of the judgment process and the correlation between the body data in each dimension, thereby enabling accurate risk warnings for patients with long-term cirrhosis. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 The flowchart illustrates a deep learning-based risk warning method for hepatobiliary patients, as provided in one embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a deep learning-based risk warning method and system for hepatobiliary patients proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, 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 pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of a deep learning-based risk warning method and system for liver and gallbladder patients provided by this invention.

[0020] Please see Figure 1 This illustrates a deep learning-based risk warning method for hepatobiliary patients provided by an embodiment of the present invention. The method includes: step S1: acquiring body data of different long-term cirrhosis patients at each sampling time in different dimensions.

[0021] This invention is primarily applied to risk warning scenarios during the transformation of cirrhosis patients into liver cancer. Since the pathological condition of cirrhosis patients can be reflected through various dimensions of physical data, this invention acquires physical data from different long-term cirrhosis patients at each sampling time across different dimensions. It should be noted that in this invention, the dimensions of the physical data are set as alanine aminotransferase (ALT), aspartate aminotransferase (AST), bilirubin, albumin, globulin, alpha-fetoprotein (AFP), and abnormal prothrombin (PIVKA-II). Furthermore, the dimensions of the physical data can be set independently and are not limited here.

[0022] In one embodiment of the present invention, the sampling period is set to one month. It should be noted that in other embodiments of the present invention, the sampling period can be set independently and is not limited here. Furthermore, the number of patients with long-term cirrhosis is set to 10, which is not limited here.

[0023] Step S2: Select one dimension of body data from a patient with long-term cirrhosis as the reference dimension body data for the reference patient; based on the data change characteristics of the reference dimension body data within a preset time period, obtain the lesion parameters of the reference dimension for the reference patient within the preset time period; count the number of patients with long-term cirrhosis who have developed liver cancer; based on the number of liver cancer patients and the distribution of lesion parameters of different patients with long-term cirrhosis in the reference dimension, obtain the degree of influence of the reference dimension on the number of liver cancer patients; obtain the lesion parameters of the patient to be analyzed in different dimensions within the preset time period; based on the similarity characteristics of the lesion parameters of the patient to be analyzed and all liver cancer patients in each dimension, obtain the degree of lesion similarity between the patient to be analyzed and all liver cancer patients; based on the degree of influence of each dimension on the number of liver cancer patients, the lesion parameters of each dimension of the patient to be analyzed, and the lesion parameters of each dimension of all patients with long-term cirrhosis, obtain the predicted parameters for the patient to be analyzed.

[0024] In reality, in the early stages of cirrhosis, a certain dimension of physical data may only show a slight increase initially, but as the cirrhosis worsens, this dimension of physical data will increase significantly. Furthermore, fluctuations in this dimension of physical data within a certain timeframe are normal physiological fluctuations and are unrelated to the degree of cirrhosis. Therefore, it is necessary to monitor physical data over a longer period and reflect the severity of long-term cirrhosis by analyzing changes in a certain dimension of physical data within that period. Thus, in this embodiment of the invention, the pathological parameters of the reference dimension for a reference patient within a preset time period are obtained based on the data change characteristics of the reference dimension of physical data within that preset time period.

[0025] Preferably, in one embodiment of the present invention, the method for obtaining lesion parameters includes: obtaining lesion parameters according to a lesion parameter calculation formula, wherein the lesion parameter calculation formula is as follows: In the formula, This indicates the lesion parameters of the reference patient within a preset time period; Indicates the sequence number of the sampling time; This indicates the number of sampling moments within a preset time period; Indicates the first The next sampling time after the previous sampling time; Indicates the first Body data at the sampling time and the first sampling time The slope of the straight line determined by the body data at each sampling time; Indicates the first Body data at the sampling time and the first sampling time The slope of the straight line determined by the body data at each sampling time; This represents the floor function.

[0026] In the formula for calculating pathological parameters, to prevent sudden changes in physical data in the reference dimension from leading to misjudgments of changes in patients with long-term cirrhosis, the [missing information] is used. Body data at the sampling time and the first sampling time The slope of the straight line determined by the body data at the sampling time, and the slope of the straight line determined by the body data at the sampling time. Body data at the sampling time and the first sampling time The mean difference between the slopes of the straight lines determined by the body data at each sampling time point. To conduct analysis, The larger the value, the more likely it is to be the first Taking a specific sampling time as the starting point, the higher the degree of change in the body data of the reference dimension at other sampling times, the higher the pathological parameter of the reference dimension; at this time, the sampling time used as the starting point... The process involves iterating through the data and analyzing the degree of change in body data at each starting point and subsequent sampling times, until the data reaches the starting point. To avoid excessive repetition in subsequent sampling, which could weaken the anti-interference capability of lesion parameter calculation and thus affect the accuracy of lesion parameters.

[0027] During the progression of cirrhosis to liver cancer, various dimensions of physical data typically change, but not all dimensions do. Therefore, a change in a particular dimension only indicates a potential progression towards liver cancer, but it doesn't definitively confirm it. Thus, it's necessary to statistically analyze the number of long-term cirrhosis patients who develop liver cancer across various disease parameters to assess the influence of changes in that dimension of physical data on the outcome of liver cancer development. In this embodiment of the invention, the number of long-term cirrhosis patients who transform into liver cancer is statistically analyzed, and the influence of subsequent dimensions is studied. Furthermore, by comparing the similarities between the cirrhosis patient to be analyzed and different long-term cirrhosis patients in the big data dataset, predictive parameters for the patient to be analyzed are obtained.

[0028] Preferably, in one embodiment of the present invention, the method for obtaining the degree of influence of the dimension includes: arranging the lesion parameters of the reference dimension of all long-term cirrhosis patients in ascending order within a preset time period to obtain a lesion parameter sequence; and dividing the lesion parameter sequence equally to obtain a preset number of lesion parameter intervals. In one embodiment of the present invention, the preset number is set to 10. It should be noted that the preset number can be set by itself and is not limited here.

[0029] The ratio of the number of long-term cirrhosis patients in each lesion parameter interval to the total number of long-term cirrhosis patients is calculated as a reference dimension to determine the population distribution probability in each lesion parameter interval. The ratio of the number of liver cancer patients in each lesion parameter interval to the total number of long-term cirrhosis patients is calculated as a reference dimension to determine the liver cancer patient distribution probability in each lesion parameter interval. The marginal probability of long-term cirrhosis patients transforming into liver cancer patients is calculated. It should be noted that the marginal probability is a technique well-known to those skilled in the art and will not be elaborated here.

[0030] The degree of dimensional influence is obtained according to the formula for calculating the degree of dimensional influence, as shown below: In the formula, This indicates the number of patients with long-term cirrhosis. Indicates the preset number of disease parameter ranges; This indicates the degree of influence of the reference dimension on the dimension of the number of liver cancer patients. This indicates the number of patients with long-term cirrhosis who develop liver cancer. This represents the number of patients with long-term cirrhosis within each range of pathological parameters. This represents the probability distribution of liver cancer patients in each lesion parameter interval for the reference dimension. This represents the marginal probability that a patient with long-term cirrhosis will develop liver cancer. This represents the probability distribution of the number of people in each lesion parameter range for the reference dimension.

[0031] It should be noted that the above formula is a mutual information formula between the body data of the reference dimension and the event of a long-term cirrhosis patient turning into a liver cancer patient. It is a technical means well known to those skilled in the art, and will not be elaborated here. Only the formula parameters will be explained.

[0032] For patients requiring risk warning, the specific situation of the patients can be obtained by analyzing the similar characteristics of changes in various dimensions of physical data between the patients and different long-term cirrhosis patients.

[0033] Preferably, in one embodiment of the present invention, the method for obtaining the lesion similarity between the patient to be analyzed and all liver cancer patients includes: calculating the mean value of lesion parameters of all liver cancer patients in each dimension as the overall lesion parameters of liver cancer patients in each dimension; taking the lesion parameters of all dimensions of the patient to be analyzed as a first column vector; taking the overall lesion parameters of all dimensions as a second column vector; and forming a first matrix by combining the first column vector and the second column vector.

[0034] The cosine similarity between the lesion parameters of the patient to be analyzed and the overall lesion parameters of all liver cancer patients is calculated, and this cosine similarity is used as the degree of lesion similarity between the patient to be analyzed and all liver cancer patients. It should be noted that the cosine similarity calculation method is a technique well-known to those skilled in the art and will not be elaborated upon here.

[0035] Based on the degree of influence of each dimension on the conversion of the number of liver cancer patients, the pathological parameters of each dimension of the patients to be analyzed, and the pathological parameters of each dimension of all long-term cirrhosis patients, the predictive parameters of the patients to be analyzed are obtained.

[0036] Preferably, in one embodiment of the present invention, the method for obtaining prediction parameters includes: obtaining prediction parameters according to a prediction parameter calculation formula, wherein the prediction parameter calculation formula is as follows: In the formula, Indicates the predictive parameters for the patient to be analyzed; This indicates the number of dimensions in the patient's physical data to be analyzed; Indicates the first The degree of influence of each dimension on the conversion of the number of liver cancer patients; This represents the sum of the influence of all dimensions on the transformation of the number of liver cancer patients; Indicates the patient to be analyzed. Pathological parameters in multiple dimensions; This indicates that all liver cancer patients were at the first stage of treatment. Overall lesion parameters in multiple dimensions; This indicates the degree of lesion similarity between the patient being analyzed and all liver cancer patients; This represents the arctangent function.

[0037] In the formula for calculating the prediction parameters, the first... The larger the ratio between the influence of the first dimension on the number of liver cancer patients and the sum of the influences of all dimensions on the number of liver cancer patients, the stronger the influence of the first dimension. The better the prediction effect of the body data in each dimension on liver cancer lesions, the more effective the analysis of the patient's [number]th dimension is in predicting liver cancer lesions. The pathological parameters of each dimension are compared with those of all liver cancer patients in the first quarter. Differences between overall pathological parameters in each dimension The greater the weight, and When the value is less than 0, it indicates that the patient to be analyzed is in the first stage. The pathological parameters in one dimension were lower than those in all liver cancer patients in the first dimension. The overall lesion parameters in one dimension are small at this time; when... A value greater than 0 indicates that the patient being analyzed is in the [number]th [stage]. The lesion parameters in one dimension were higher than those in all liver cancer patients in the first dimension. If the overall pathological parameters in each dimension are considered, then the patient's cirrhosis may have progressed to liver cancer, indicating a more severe condition and a larger predictive parameter. In this embodiment of the invention, through... The function restricts the range to Between; that is The larger the value, the larger the predictive parameters for the patient being analyzed; the same analysis is performed on each dimension to analyze the predictive parameters of all dimensions of the patient's physical data; and the degree of lesion similarity between the patient being analyzed and all liver cancer patients is also considered. The larger the value, the higher the similarity of the physical data between the patient being analyzed and a liver cancer patient, and the more likely the patient is to be in a liver cancer state; therefore, the larger the predictive parameter, the better.

[0038] Step S3: Provide risk warnings for the patients to be analyzed based on the prediction parameters.

[0039] Preferably, in one embodiment of the present invention, risk warning for the patient to be analyzed based on the predicted parameters includes: when the predicted parameters of the patient to be analyzed are greater than a preset first threshold, the patient to be analyzed is considered a high-risk patient with cirrhosis, and the likelihood of the patient to be analyzed having liver cancer is considered higher, and a risk warning is issued for the patient to be analyzed. In one embodiment of the present invention, the preset first threshold is set to 0.6. It should be noted that in other embodiments of the present invention, the preset first threshold can be set by itself, and is not limited here.

[0040] This completes the risk warning for the patients to be analyzed.

[0041] In summary, the following steps are taken: First, physical data from different long-term cirrhosis patients at each sampling time across various dimensions are obtained. Second, physical data from one dimension of a randomly selected long-term cirrhosis patient is used as the reference dimension physical data for a reference patient. Third, based on the data change characteristics of the reference dimension physical data within a preset time period, the pathological parameters of the reference dimension for the reference patient within that preset time period are obtained. Fourth, the number of all long-term cirrhosis patients who develop liver cancer is counted. Fifth, based on the number of liver cancer patients and the distribution of pathological parameters in the reference dimension among different long-term cirrhosis patients, the degree of influence of the reference dimension on the number of liver cancer patients is obtained. Sixth, pathological parameters of the patient to be analyzed in different dimensions are obtained within a preset time period. Seventh, based on the similarity characteristics of the pathological parameters in each dimension between the patient to be analyzed and all liver cancer patients, the degree of pathological similarity between the patient to be analyzed and all liver cancer patients is obtained. Eighth, predictive parameters for the patient to be analyzed are obtained based on the predictive parameters.

[0042] A second objective of one embodiment of the present invention is to provide a deep learning-based risk warning system for hepatobiliary patients. The system includes a memory, a processor, and a computer program, wherein the memory is used to store the corresponding computer program, the processor is used to run the corresponding computer program, and the computer program, when running in the processor, can implement the methods described in steps S1-S3.

[0043] A third objective of this invention is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the methods described in steps S1-S3.

[0044] The fourth objective of this invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in steps S1-S3.

[0045] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0046] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A deep learning-based risk warning method for hepatobiliary patients, characterized in that, The method includes: acquiring body data of different long-term cirrhosis patients at each sampling time in different dimensions; randomly selecting the body data of one long-term cirrhosis patient in one dimension as the reference dimension body data of a reference patient; obtaining the lesion parameters of the reference dimension of the reference patient in the preset time period based on the data change characteristics of the reference dimension body data within a preset time period; counting the number of patients who have transformed into liver cancer patients among all long-term cirrhosis patients; obtaining the degree of influence of the reference dimension on the transformation of the number of liver cancer patients based on the number of liver cancer patients and the distribution of lesion parameters of different long-term cirrhosis patients in the reference dimension; acquiring the lesion parameters of the patient to be analyzed in different dimensions within the preset time period; obtaining the degree of lesion similarity between the patient to be analyzed and all liver cancer patients based on the similarity characteristics of the lesion parameters of the patient to be analyzed and all liver cancer patients in each dimension; obtaining the predicted parameters of the patient to be analyzed based on the degree of influence of each dimension on the transformation of the number of liver cancer patients, the lesion parameters of the patient to be analyzed in each dimension, and the lesion parameters of all long-term cirrhosis patients in each dimension; and providing risk warning for the patient to be analyzed based on the predicted parameters. The method for obtaining the lesion parameters includes: obtaining the lesion parameters according to the lesion parameter calculation formula, which is shown below: In the formula, This indicates the lesion parameters of the reference patient within a preset time period; Indicates the sequence number of the sampling time; This indicates the number of sampling moments within a preset time period; Indicates the first The next sampling time after the previous sampling time; Indicates the first Body data at the sampling time and the first sampling time The slope of the straight line determined by the body data at each sampling time; Indicates the first Body data at the sampling time and the first sampling time The slope of the straight line determined by the body data at each sampling time; The method for obtaining the degree of influence of the dimension includes: arranging the lesion parameters of the reference dimension for all long-term cirrhosis patients in ascending order within a preset time period to obtain a sequence of lesion parameters; dividing the sequence of lesion parameters equally to obtain a preset number of lesion parameter intervals; obtaining the probability distribution of the number of long-term cirrhosis patients in each lesion parameter interval and the probability distribution of liver cancer patients in each lesion parameter interval based on the number distribution of long-term cirrhosis patients in each lesion parameter interval; calculating the marginal probability of long-term cirrhosis patients transforming into liver cancer patients; and obtaining the degree of influence of the dimension according to the formula for calculating the degree of influence of the dimension, which is shown below: In the formula, This indicates the number of patients with long-term cirrhosis. Indicates the preset number of disease parameter ranges; This indicates the degree of influence of the reference dimension on the dimension of the number of liver cancer patients. This indicates the number of patients with long-term cirrhosis who develop liver cancer. This represents the number of patients with long-term cirrhosis within each range of pathological parameters. This represents the probability distribution of liver cancer patients in each lesion parameter interval for the reference dimension. This represents the marginal probability that a patient with long-term cirrhosis will develop liver cancer. This represents the probability distribution of the number of people in each lesion parameter interval for the reference dimension; the method for obtaining the prediction parameter includes: obtaining the prediction parameter according to the prediction parameter calculation formula, which is shown below: In the formula, Indicates the predictive parameters for the patient to be analyzed; This indicates the number of dimensions in the patient's physical data to be analyzed; Indicates the first The degree of influence of each dimension on the conversion of the number of liver cancer patients; This represents the sum of the influence of all dimensions on the transformation of the number of liver cancer patients; Indicates the patient to be analyzed. Pathological parameters in multiple dimensions; This indicates that all liver cancer patients were at the first stage of treatment. Overall lesion parameters in multiple dimensions; This indicates the degree of lesion similarity between the patient being analyzed and all liver cancer patients; This represents the arctangent function.

2. The method for early warning of hepatobiliary patients based on deep learning according to claim 1, characterized in that, The method for obtaining the lesion similarity includes: calculating the mean of lesion parameters in each dimension for all liver cancer patients, as the overall lesion parameters of liver cancer patients in each dimension; taking the lesion parameters in all dimensions of the patient to be analyzed as a first column vector; taking the overall lesion parameters in all dimensions as a second column vector; forming a first matrix by combining the first column vector and the second column vector; and calculating the cosine similarity between the lesion parameters of the patient to be analyzed and the overall lesion parameters of all liver cancer patients, as the lesion similarity between the patient to be analyzed and all liver cancer patients.

3. The method for early warning of hepatobiliary patients based on deep learning according to claim 1, characterized in that, The risk warning for the patient to be analyzed is based on the predicted parameters, including: when the predicted parameters of the patient to be analyzed are greater than a preset first threshold, the patient to be analyzed is regarded as a high-risk cirrhosis patient and a risk warning is issued for the patient to be analyzed.

4. The method for early warning of hepatobiliary patients based on deep learning according to claim 1, characterized in that, The method for obtaining the population distribution probability and the liver cancer patient distribution probability includes: calculating the ratio between the number of long-term cirrhosis patients in each lesion parameter interval and the total number of long-term cirrhosis patients as a reference dimension for the population distribution probability in each lesion parameter interval; and calculating the ratio between the number of liver cancer patients in each lesion parameter interval and the total number of long-term cirrhosis patients as a reference dimension for the liver cancer patient distribution probability in each lesion parameter interval.

5. A deep learning-based risk warning system for hepatobiliary patients, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 4.

6. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the deep learning-based risk warning method for hepatobiliary patients as described in any one of claims 1 to 4.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the deep learning-based risk warning method for hepatobiliary patients as described in any one of claims 1 to 4.

Images