Detection method and system for auxiliary diagnosis of hepatic encephalopathy

Abstract

The invention relates to the technical field of computer-aided diagnosis, and discloses a detection method and system for aided diagnosis of hepatic encephalopathy, and the system comprises a data collection module which is configured to collect comprehensive body data of a patient, and builds a disease influence data set according to the comprehensive body data; the data selection module is configured to output an influence value of each comprehensive body data in the illness influence data set on the hepatic encephalopathy based on the data importance model, and select the illness influence data set based on the influence value to obtain a core influence data set and a supplementary influence data set; the index calculation module is configured to obtain a core influence data set, divide regions according to a preset core partition value, and count the core data quantity of the comprehensive body data in each region; the HE index calculation module is configured to calculate an HE index according to the quantity of the core data and the quantity of the supplementary data; and a diagnosis early warning module. According to the invention, early, accurate and objective diagnosis of hepatic encephalopathy is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of computer - aided judgment, and more particularly, to a detection method and system for assisting in the diagnosis of hepatic encephalopathy. Background Art

[0002] Hepatic Encephalopathy (HE) is a brain dysfunction caused by severe liver function disorders or portal hypertension, commonly seen in patients with liver cirrhosis. The pathological mechanism of hepatic encephalopathy is complex and is currently considered to be related to factors such as ammonia toxicity, inflammatory response, and neurotransmitter imbalance. This disease can lead to various impairments in cognition, movement, behavior, etc., and can be life - threatening in severe cases.

[0003] Currently, the diagnosis of hepatic encephalopathy mainly relies on clinical manifestations, neuropsychological tests, and the detection of blood indicators (such as blood ammonia concentration). However, neuropsychological tests are greatly affected by the patient's subjective state and the doctor's experience, and the results may not be objective enough. In addition, when some patients are in a mild condition, the clinical symptoms are not obvious, and it is difficult for traditional methods to detect the lesions in a timely manner. The blood ammonia concentration is an important reference index for the diagnosis of hepatic encephalopathy, but it is not completely consistent with the severity of hepatic encephalopathy, and single - detection is prone to misjudgment.

[0004] In summary, there is an urgent need for an auxiliary diagnosis method and system that combines multi - modal data to achieve early, accurate, and objective diagnosis of hepatic encephalopathy. Summary of the Invention

[0005] In view of this, the present invention proposes a detection method and system for assisting in the diagnosis of hepatic encephalopathy, aiming to solve the problem of the lack of early, accurate, and objective diagnosis of hepatic encephalopathy in the current technology.

[0006] On the one hand, a detection system for assisting in the diagnosis of hepatic encephalopathy proposed by the present invention includes:

[0007] A data acquisition module, configured to collect comprehensive physical data of a patient and establish a disease - influencing data set based on the comprehensive physical data. The comprehensive physical data includes: blood ammonia concentration, lactate concentration, electroencephalogram data, and behavior data;

[0008] A data selection module, configured to construct a data importance model according to the pathological characteristics of hepatic encephalopathy, output the influence value of each piece of comprehensive physical data in the disease - influencing data set on hepatic encephalopathy based on the data importance model, and select the disease - influencing data set based on the influence value to obtain a core influence data set and a supplementary influence data set;

[0009] An index calculation module, configured to obtain the core impact dataset, extract the same type of comprehensive physical data in the core impact dataset and sort it, divide regions according to a preset core partition value, and count the core data quantity of the comprehensive physical data in each region;

[0010] Obtain the supplementary impact dataset and calculate the statistical supplementary data quantity;

[0011] An HE index calculation module, configured to calculate the HE index according to the core data quantity and the supplementary data quantity;

[0012] A diagnosis and early warning module, configured to preset multiple index thresholds respectively, compare the HE index with the index thresholds, and divide the early warning levels.

[0013] Further, when the data selection module constructs a data importance model according to the pathological characteristics of hepatic encephalopathy, it includes:

[0014] Use wavelet transform or empirical mode decomposition to extract blood ammonia volatility and lactate peak value, capture the abnormal proportion of EEG signals based on convolutional neural network EEG signal processing, and extract stride volatility and standing stability indicators through an action capture device and a gait analysis model;

[0015] Adopt a spatio-temporal correlation model of patient data constructed based on a multi-scale graph neural network, perform feature aggregation on the time series changes of each type of comprehensive physical data, capture the relationship between short-term fluctuations and long-term trends, and establish a dynamic relationship graph between patient features;

[0016] According to the statistical characteristics of the patient's historical clinical data, assign initial weights; use a reinforcement learning agent, take the prediction accuracy as the reward signal, initially adjust the weights of each comprehensive physical data, introduce the patient's specific factors to perform secondary adjustment on the weights, and use the weights after secondary adjustment as the influence value.

[0017] Further, the specific factors include: patient age, previous attack times of hepatic encephalopathy, and the course of related diseases; the course of related diseases is the sum of the disease durations of diabetes, hyperlactic acidemia, cognitive impairment, and dementia.

[0018] Further, the influence value is obtained by multiplying the initially adjusted value by a secondary adjustment coefficient, and the secondary adjustment coefficient satisfies the following relationship:

[0019]

[0020] Where K is the secondary adjustment coefficient; α is the secondary adjustment global influence factor, indicating the influence degree of secondary adjustment on the influence value; N is the total number of patient specific factors; C i is the contribution coefficient of the i-th specific factor; Fi is the actual value of the i-th specific factor; F i,min is the minimum value of the i-th specific factor in the historical data; F i,min is the maximum value of the i-th specific factor in the historical data.

[0021] Further, when selecting the disease impact data set based on the impact value to obtain a core impact data set and a supplementary impact data set, it includes:

[0022] Preset an impact threshold, and compare the impact value with the impact threshold; when the impact value is greater than or equal to the impact threshold, the corresponding comprehensive body data is used as the core impact data set, and when the impact value is less than the impact threshold, the corresponding comprehensive body data is used as the supplementary impact data set.

[0023] Further, when the index calculation module divides regions according to a preset core partition value and counts the core data quantity of the comprehensive body data in each region, it includes:

[0024] The core partition value includes a first core partition value and a second core partition value, and the first core partition value is less than the second core partition value;

[0025] The index calculation module divides the same kind of comprehensive body data less than the first preset core partition value into the first partition;

[0026] Divides the same kind of comprehensive body data greater than the first preset core partition value and less than or equal to the second preset core partition value into the second partition;

[0027] Divides the same kind of comprehensive body data greater than the second preset core partition value into the third partition;

[0028] Respectively count the quantities of the comprehensive body data in the first partition, the second partition and the third partition, and record them as the core data quantity;

[0029] The calculation method of the supplementary data quantity is the same as that of the core data quantity.

[0030] Further, when calculating the HE index according to the core data quantity and the supplementary data quantity, it includes:

[0031] The HE index is calculated through the following relationship:

[0032]

[0033] where n1 is the core data quantity; N core,i is the quantity of the i-th type of core data; w iis the weight of the i-th type of core data; n2 is the number of supplementary data; N supp,j is the number of the j-th type of supplementary data; γ is the weight of the dynamic adjustment factor, which controls the role of the specificity factor in the HE index; D is the dynamic adjustment factor, which is calculated based on the patient's specificity factor and corrects the HE index.

[0034] Furthermore, the dynamic adjustment factor satisfies the following relationship:

[0035]

[0036] where m is the total number of specificity factors, c k is the contribution coefficient of the k-th specificity factor, F k is the actual value of the k-th specificity factor, F k,max 、F k,min are respectively the minimum and maximum values of the k-th specificity factor in the historical data.

[0037] Furthermore, the diagnosis and early warning module pre-sets a first index threshold and a second index threshold, and the first index threshold is less than the second index threshold;

[0038] Compare the HE index with the index threshold. When the HE index is less than the first index threshold, it is judged that the disease probability is low;

[0039] When the HE index is less than the second index threshold and greater than or equal to the first index threshold, it is judged that the disease probability is medium;

[0040] When the HE index is less than the third index threshold and greater than or equal to the second index threshold, it is judged that the disease probability is high.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0042] By comprehensively collecting multi-modal data such as blood ammonia concentration, lactic acid concentration, electroencephalogram data, and behavior data, the limitations of single-index detection are avoided, and potential lesions of hepatic encephalopathy can be detected earlier; the data selection module constructs a data importance model based on multi-modal data and pathological features, accurately screens the core influencing data set, and improves the sensitivity of diagnosis.

[0043] Wavelet transform, convolutional neural network, and multi-scale graph neural network are used to extract and aggregate features of time series data, which can capture the dynamic association between short-term fluctuations and long-term trends, reduce noise interference and manual judgment errors; the patient's specificity factors (such as age, previous attack times, and the course of related diseases) are used for secondary weight adjustment to ensure the personalization of the diagnosis result and improve the accuracy of the HE index.

[0044] The diagnosis and early warning module divides the HE index into multiple early warning levels (low, medium, high) according to the preset index thresholds, intuitively reflecting the probability of disease, and providing a reference for hierarchical diagnosis for clinicians; the multi-level early warning mechanism facilitates the adoption of different intervention measures according to the severity of the condition, optimizes the allocation of resources, and reduces the possibility of misdiagnosis and missed diagnosis.

[0045] The dynamic adjustment factor corrects the HE index through the contribution coefficient and dynamic weight of the patient-specific factor, enabling the diagnostic model to be dynamically adjusted according to individual patient differences, meeting the needs of different patient groups, and enhancing the reliability of diagnosis; the reinforcement learning agent optimizes the weights with the prediction accuracy as the reward signal, gradually improving the model performance to adapt to complex scenarios.

[0046] The data partitioning and statistical method divides different regions of the core data set through the core partition value, effectively solving the influence of abnormal data and enhancing the robustness of the system; the introduction of a supplementary data set as a supplement to the core data set enables the system to adapt to various data integrity and quality levels, ensuring a high diagnostic effect in different data environments.

[0047] From data collection, feature extraction to HE index calculation and early warning classification, each module realizes process automation, reducing the need for manual intervention; the system combines multi-modal data and intelligent algorithms to provide comprehensive decision-making support for doctors and reduce the dependence on subjective judgment.

[0048] The system architecture can be extended to the detection of other liver disease-related brain function disorders or metabolic diseases (such as diabetic encephalopathy, etc.), with good promotion value; both the data collection module and the model algorithm support flexible expansion, and more index data can be accessed according to clinical needs.

[0049] On the other hand, the detection method for assisting in the diagnosis of hepatic encephalopathy proposed by the present invention is applied to the above-mentioned detection system for assisting in the diagnosis of hepatic encephalopathy, including:

[0050] S1: Collect the comprehensive physical data of the patient, and establish a disease impact data set according to the comprehensive physical data. The comprehensive physical data includes: blood ammonia concentration, lactate concentration, electroencephalogram data, and behavioral data;

[0051] S2: According to the pathological characteristics of hepatic encephalopathy, construct a data importance model, output the influence value of each comprehensive physical data in the disease impact data set on hepatic encephalopathy based on the data importance model, and select the disease impact data set based on the influence value to obtain a core impact data set and a supplementary impact data set;

[0052] S3: Obtain the core impact data set, extract and sort the same type of comprehensive physical data in the core impact data set, divide the region according to the preset core partition value, and count the core data quantity of the comprehensive physical data in each region;

[0053] Obtain the supplementary influence data set and calculate the statistical supplementary data quantity;

[0054] S4: Calculate the HE index according to the core data quantity and the supplementary data quantity;

[0055] S5: Preset multiple index thresholds respectively, compare the HE index with the index thresholds, and divide the warning levels.

[0056] It can be understood that the above detection method and system for assisting in the diagnosis of hepatic encephalopathy have the same beneficial effects, which will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered as a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0058] Figure 1 is the functional framework diagram of the detection system for assisting in the diagnosis of hepatic encephalopathy provided by the embodiment of the present invention;

[0059] Figure 2 is the flowchart of the detection method for assisting in the diagnosis of hepatic encephalopathy provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the drawings and in combination with the embodiments.

[0061] Refer to Figure 1 As shown, the embodiment of the present invention provides a detection system for assisting in the diagnosis of hepatic encephalopathy, including:

[0062] A data acquisition module, configured to acquire the comprehensive physical data of a patient and establish a disease influence data set according to the comprehensive physical data, where the comprehensive physical data includes: blood ammonia concentration, lactic acid concentration, electroencephalogram data, and behavior data;

[0063] A data selection module, configured to construct a data importance model according to the pathological characteristics of hepatic encephalopathy, output the influence value of each comprehensive physical data in the diseased influence dataset on hepatic encephalopathy based on the data importance model, and select the diseased influence dataset based on the influence value to obtain a core influence dataset and a supplementary influence dataset;

[0064] An index calculation module, configured to obtain the core influence dataset, extract the same comprehensive physical data in the core influence dataset and sort them, divide regions according to a preset core partition value, and count the core data quantity of the comprehensive physical data in each region;

[0065] Obtain the supplementary influence dataset and calculate the statistical supplementary data quantity;

[0066] It should be noted that the calculation methods of the supplementary data quantity and the core data quantity are the same.

[0067] A HE index calculation module, configured to calculate the HE index according to the core data quantity and the supplementary data quantity;

[0068] A diagnosis and warning module, configured to preset multiple index thresholds respectively, compare the HE index with the index thresholds, and divide the warning levels.

[0069] It should be noted that the data acquisition module integrates blood ammonia concentration, lactic acid concentration, electroencephalogram data and behavior data, covering multiple pathophysiological characteristics of hepatic encephalopathy, breaking through the limitations of traditional single data sources, and making the diagnosis more comprehensive. The data selection module conducts quantitative analysis on the influence values of different comprehensive physical data by constructing a data importance model, and screens the core influence data and the supplementary influence data. In this way, it can not only focus on key indicators but also take into account other potential factors, ensuring the scientificity and rationality of data processing. The index calculation module and the diagnostic factor calculation module conduct refined analysis on the contributions of the core data and the supplementary data through regional statistics and data classification calculations; the calculation of the HE index synthesizes multiple data, reflecting the multi-level characteristics of the disease and facilitating the dynamic monitoring of the patient's condition. The diagnosis and warning module divides multiple index thresholds and gives graded warnings according to the HE index results, which can intuitively reflect the severity of the condition, help clinicians quickly formulate personalized intervention strategies, and improve the diagnosis efficiency. The functions of each module are clear and the process is clear, and it can realize the full-process automatic processing from data acquisition to HE index calculation and warning grading, reducing subjective judgment and human error.

[0070] Through comprehensive data and multi-level analysis, the system can capture potential changes in the early stage of the disease, help doctors intervene in time, and avoid the deterioration of the condition. The modular design combined with multi-modal data not only improves the accuracy of diagnosis but also provides flexible diagnostic support for different patient groups. Through the division of warning levels, graded intervention strategies can be formulated according to the severity of the condition, optimizing the allocation of medical resources and improving the treatment efficiency.

[0071] In some embodiments of the present application, when constructing a data importance model according to the pathological characteristics of hepatic encephalopathy, the data selection module includes:

[0072] Using wavelet transform or empirical mode decomposition to extract blood ammonia volatility and lactate peak value, capturing the abnormal proportion of EEG signals based on convolutional neural network-based EEG signal processing, and extracting step length volatility and standing stability indicators through an action capture device and a gait analysis model;

[0073] Adopting a spatio-temporal correlation model of patient data constructed based on a multi-scale graph neural network, aggregating features of the time series changes of each comprehensive body data, capturing the relationship between short-term fluctuations and long-term trends, and establishing a dynamic relationship graph between patient features;

[0074] According to the statistical characteristics of the patient's historical clinical data, initial weights are assigned; using a reinforcement learning agent, with the prediction accuracy as the reward signal, the weights of each comprehensive body data are initially adjusted, and a patient-specific factor is introduced to perform a secondary adjustment on the weights, and the weights after the secondary adjustment are used as influence values.

[0075] It should be noted that multiple signal processing methods work together, enabling the system to extract features highly relevant to hepatic encephalopathy from multi-source data, providing reliable support for subsequent diagnosis. The design of combining a multi-scale graph neural network with reinforcement learning enables the system to dynamically adjust the diagnostic model according to time, effectively capturing the complex dynamic changes during the development of the patient's condition, with strong adaptability. The deep fusion of multi-modal data, dynamic optimization of weights, and the introduction of specific factors enable the system to comprehensively and accurately reflect the actual situation of the patient, reducing the risks of missed diagnosis and misdiagnosis. The specific factor in the weight adjustment process takes into account the individual differences of the patient, making the diagnostic result more targeted and providing a scientific basis for the formulation of personalized treatment strategies. The data importance model provides the possibility for early detection of minor lesions by capturing the key features of multi-modal data, helping doctors to perform early intervention when the disease has not significantly developed.

[0076] The above content significantly enhances the system's ability to extract features, dynamic analysis ability, and personalized diagnosis ability for hepatic encephalopathy through technical means such as multi-source data fusion, reinforcement learning optimization, and personalized weight allocation, ensuring that the diagnostic result is more accurate, comprehensive, and reliable. This design not only effectively improves the diagnostic efficiency of hepatic encephalopathy but also provides solid technical support for clinical decision-making, promoting the practical application of the auxiliary diagnosis system in the medical field.

[0077] In some embodiments of the present application, the specific factor includes: patient age, the number of previous episodes of hepatic encephalopathy, and the course of related diseases; the course of related diseases is the sum of the disease durations of diabetes, hyperlacticacidemia, cognitive impairment, and dementia.

[0078] It is understandable that age, the number of previous episodes, and the course of related diseases jointly affect the occurrence and development of hepatic encephalopathy. The comprehensive analysis of specific factors can help doctors better evaluate the severity of the patient's condition, thus providing more targeted suggestions for subsequent treatment. The dynamic adjustment of specific factors enables the system to predict the trend of disease changes based on the individual characteristics of patients and classify and warn of disease risks, helping doctors intervene in a timely manner and reducing the possibility of disease deterioration. By quantifying the number of previous episodes and the course of related diseases of patients, the system can support the long-term management of the patient's medical history, contribute to monitoring disease changes, and provide a basis for subsequent treatment. The role of specific factors in weight adjustment ensures the system's dynamic response ability to the characteristics of each patient, avoids diagnostic biases that may be brought about by fixed weights, and guarantees the rationality and scientific nature of diagnostic results.

[0079] In some embodiments of the present application, the influence value is obtained by multiplying the initially adjusted value by a secondary adjustment coefficient, and the secondary adjustment coefficient satisfies the following relationship:

[0080]

[0081] where K is the secondary adjustment coefficient; α is the secondary adjustment global influence factor, indicating the degree of influence of the secondary adjustment on the influence value, 0.5 ≤ α ≤ 1; N is the total number of patient-specific factors; C i is the contribution coefficient of the i-th specific factor, C i represents the degree of influence of this factor on the weight of comprehensive physical data, which can be set according to clinical experience, 0.8 ≤ Ci ≤ 1; F i is the actual value of the i-th specific factor; F i,min is the minimum value of the i-th specific factor in historical data; F i,min is the maximum value of the i-th specific factor in historical data.

[0082] It should be noted that the secondary adjustment coefficient provides a flexible adjustment mechanism for the diagnostic process by introducing the contribution coefficients of the global influence factor and the patient-specific factor. By setting reasonable coefficient ranges (such as the global influence factor ranging from 0.5 to 1 and the contribution coefficient ranging from 0.8 to 1), the weights of different factors can be adjusted according to actual needs, enabling the diagnostic model to have better adaptability to different situations. By multiplying the influence value after the initial adjustment by the secondary adjustment coefficient, the consideration of the patient's specific situation is further refined. The combination of the number of specific factors, their actual values, and their minimum and maximum values in the historical data provides dynamic correction based on the individual characteristics of the patient. This can better capture the physiological and pathological characteristics of the patient, thereby improving the accuracy of diagnosis. The secondary adjustment coefficient can further correct the influence value according to the specific situation of the patient, thereby optimizing the allocation of data weights and ensuring that the diagnostic system can accurately capture the individual differences of the patient. Especially when faced with complex cases, it can more accurately reflect the severity of the patient's condition. By adjusting the influence value according to the patient's specific factors, the errors that may occur when the model faces different patients can be reduced. For example, for specific patient groups (such as the elderly, those with other chronic diseases, etc.), through the careful setting of the secondary adjustment coefficient, the credibility and reliability of the diagnostic results for this group can be improved.

[0083] The secondary adjustment coefficient can adaptively adjust the weights of different patient-specific factors. Especially when the patient has special physiological conditions or a past medical history, it can better adapt to the individualized treatment needs by adjusting the influence value. During the diagnostic process, the degree of influence of specific factors (such as age, number of previous attacks, etc.) on the diagnostic result is quantified by the contribution coefficient. Through secondary adjustment, the clinical importance of these factors can be accurately reflected, providing support for the accuracy and reliability of the result. The introduction of the secondary adjustment coefficient ensures a reasonable weight of the specific factors in the influence value, enabling not only the general influence of the comprehensive physical data to be considered in the calculation of the influence value, but also a detailed adjustment for individual differences, thereby improving the accuracy of diagnosis. For patients with multiple chronic diseases or a medical history, the secondary adjustment coefficient can supplement and correct their specific health conditions, enhancing the system's ability to handle complex cases and ensuring that a reasonable and accurate diagnosis can still be made even in the case of multiple factors overlapping.

[0084] In some embodiments of the present application, when selecting the disease influence data set based on the influence value to obtain the core influence data set and the supplementary influence data set, it includes:

[0085] Preset an influence threshold and compare the influence value with the influence threshold; when the influence value is greater than or equal to the influence threshold, use the corresponding comprehensive physical data as the core influence data set, and when the influence value is less than the influence threshold, use the corresponding comprehensive physical data as the supplementary influence data set.

[0086] It should be noted that by setting the influence threshold, the influence values can be effectively screened, making the selection of data more in line with the actual diagnostic needs. Data with higher influence values are considered core influence data, which is more directly related to the judgment of hepatic encephalopathy; while data with lower influence values are classified as supplementary data, thereby improving the relevance and effectiveness of the data. The separation of the core influence data set and the supplementary influence data set can ensure that the model focuses on the most critical parameters and data, thus reducing the interference of noise and redundant information on the diagnostic results. The core data can better reflect the patient's condition and improve the accuracy and sensitivity of the diagnosis. By setting the influence threshold, the data can be quickly screened, and the data that has a greater impact on the diagnostic results is given priority. This not only improves the diagnostic efficiency but also avoids excessive unnecessary data analysis, thus saving computing resources and time. The setting of the influence threshold is flexible and can be adjusted according to the actual situation and needs. Clinicians can flexibly set the influence threshold according to the characteristics of different patients or the severity of the condition, making the data selection process highly adaptable and adjustable, and further improving the personalized diagnostic ability of the system.

[0087] By comparing and screening the influence values, the system can divide the data into two categories: the core data set and the supplementary data set. The core influence data set contains the key information that can best reflect the patient's current condition, which has a greater impact on the diagnostic results and has a higher weight; while the supplementary data set is used to provide additional supporting information, which helps to further improve the diagnostic results.

[0088] The core influence data set usually includes more important and diagnostically significant parameters, ensuring that the system can give priority to processing these data during the analysis. The supplementary data set provides additional data support for the model, which can be used to strengthen the comprehensive ability of the model but does not directly affect the main judgment result, thus helping the system to make the most accurate judgment under limited resources. By setting the influence threshold, the system can assign reasonable weights to each piece of data, enabling the influence of diagnostic factors to be reasonably reflected in the selection process of the core data set and the supplementary data set. The different importance of imaging data and physiological data can be refined through this method, thereby improving the accuracy of the overall system. By dividing the data into the core influence data set and the supplementary influence data set, the system can focus on the core data during the analysis, reduce the influence of redundant data, and avoid wasting computing resources. At the same time, this data processing method makes the subsequent calculation and model optimization process more efficient and helps to reduce unnecessary complexity.

[0089] In some embodiments of the present application, when the index calculation module divides regions according to the preset core partition value and counts the number of core data of the comprehensive body data in each region, it includes:

[0090] The core partition values include a first core partition value and a second core partition value, and the first core partition value is less than the second core partition value;

[0091] The index calculation module divides the same type of comprehensive physical data less than the first preset core partition value into the first partition;

[0092] The same type of comprehensive physical data greater than the first preset core partition value and less than or equal to the second preset core partition value is divided into the second partition;

[0093] The same type of comprehensive physical data greater than the second preset core partition value is divided into the third partition;

[0094] Count the quantities of the comprehensive physical data in the first partition, the second partition, and the third partition respectively, and record them as the core data quantities;

[0095] The calculation method of the supplementary data quantity is the same as that of the core data quantity.

[0096] It should be noted that by setting multiple core partition values, the data can be carefully divided into three different intervals, thereby improving the partitioning accuracy of the comprehensive physical data. Each partition represents different degrees of pathological characteristics or data manifestations, which can better reflect the changes in the patient's condition and help the system analyze the data in each partition targeted. This partitioning method can not only help the system identify core data of different degrees (such as indicators from mild to severe), but also provide richer and hierarchical diagnostic information by conducting statistics within each partition. This multi-dimensional performance enables the diagnostic system to comprehensively evaluate the patient's condition from different perspectives. During the process of counting the quantity of core data, the partitioning of different partitions provides more reference indicators, thereby more accurately reflecting the patient's pathological state. For example, the smaller quantity of core data in the lower partition may indicate a milder condition, while the data in the higher partition shows more severe symptoms, which can improve the accuracy of diagnosis. Processing the data with multiple partitions helps to reduce the impact of extreme values on the results and avoid excessive deviation of a single data value from the diagnostic result. Through the setting of core partition values, the system can reasonably partition different data according to their contribution degrees to the diagnostic result, thereby improving the stability and reliability of data processing.

[0097] The data is divided according to the core partition value, enabling various types of data to be assigned to different regions based on their characteristics and influence. In this way, when the system conducts statistics and analysis, it can provide more detailed and hierarchical analysis for the data in different partitions, ensuring that the contribution of each data item is effectively reflected. Through the setting of partitions, the index calculation module can perform corresponding processing according to the distribution of data in different partitions. For example, if a certain type of data is mostly concentrated in the first or second partition, the system will make targeted adjustments based on the distribution of this data to optimize the diagnostic results. Since the data value ranges represented by each partition are different, the diagnostic system can identify the specific pathological state of the patient according to the data quantity and data changes in different partitions. For example, a large amount of data in the low partition may indicate that the disease is in a relatively mild stage, while a large amount of data in the high partition may indicate a more serious condition. This distinction helps to perform more accurate hierarchical diagnosis. Partitioning and statistical analysis of the data can avoid calculating overly complex models and simplify the calculation process. The data quantity and characteristics of each partition are obvious, which can reduce the complexity of calculation and analysis to a certain extent and improve the diagnostic speed at the same time.

[0098] In some embodiments of the present application, when calculating the HE index based on the core data quantity and the supplementary data quantity, it includes:

[0099] The HE index is calculated through the following relationship:

[0100]

[0101] where n1 is the core data quantity; N core,i is the quantity of the i-th type of core data; w i is the weight of the i-th type of core data; n2 is the supplementary data quantity; N supp,j is the quantity of the j-th type of supplementary data; γ is the weight of the dynamic adjustment factor, controlling the role of the specificity factor in the HE index; D is the dynamic adjustment factor, calculated based on the patient's specificity factor, for correcting the HE index.

[0102] It should be noted that by comprehensively calculating factors such as the number of core data, the number of supplementary data, and data weights, the HE index can provide a comprehensive assessment of hepatic encephalopathy, ensuring that it does not rely solely on a single data source (such as core data or supplementary data), thereby obtaining a more accurate and comprehensive diagnostic result. Using a dynamic adjustment factor to correct the HE index can finely adjust the diagnostic result according to the patient's specific factors (such as age, disease course, etc.). This dynamic correction can enhance the personalization of the diagnostic result, enabling the system to better adapt to the clinical characteristics of different patients and improving the accuracy of diagnosis. The weights of each type of core data and supplementary data can be set or dynamically adjusted according to their impact on the condition, enabling the system to flexibly increase or decrease the influence of certain data types according to the specific situation of the patient. This flexibility enables the diagnostic system to provide more practical results in different clinical situations. Since the HE index synthesizes information from multiple dimensions, it can identify the risk of hepatic encephalopathy in patients earlier, especially being able to handle those mild symptoms that are difficult to detect by traditional methods. The dynamic correction of the HE index also helps to timely reflect the changes in the condition.

[0103] By weighted calculating factors such as the number of core data, the number of supplementary data, and the influence of specific factors through the HE index, it is possible to comprehensively evaluate the patient's condition and help doctors better understand the patient's current situation. This comprehensive evaluation helps to avoid the misleading caused by a single indicator and provides a more objective diagnostic basis. The introduction of a dynamic adjustment factor enables the HE index calculation to take into account the individual differences of patients. Especially when the patient's characteristics change, the system can quickly adjust the diagnostic result. For example, factors such as the patient's disease course and age can affect the final value of the HE index, thereby helping doctors develop a more personalized treatment plan. By dynamically adjusting and weighting different data types, the HE index can detect subtle changes in the condition earlier, issue early warnings in a timely manner, and help medical staff intervene before the condition worsens. Especially the early identification of hepatic encephalopathy helps to avoid the further deterioration of the disease and improve the prognosis of the patient. The calculation method of the HE index combines multiple core data and supplementary data, making the final diagnostic result more reliable. The dynamic correction factor can be adjusted in a timely manner according to the patient's historical data and pathological changes, thereby enhancing the credibility of the diagnosis.

[0104] In some embodiments of the present application, the dynamic adjustment factor satisfies the following relationship:

[0105]

[0106] where m is the total number of specific factors, c k is the contribution coefficient of the kth specific factor, F k is the actual value of the kth specific factor, F k,max 、Fk,min They are respectively the minimum value and the maximum value of the k-th specific factor in the historical data.

[0107] In some embodiments of the present application, the diagnosis and warning module pre-sets a first index threshold and a second index threshold, and the first index threshold is less than the second index threshold;

[0108] Compare the HE index with the index threshold. When the HE index is less than the first index threshold, it is determined that the disease probability is low;

[0109] When the HE index is less than the second index threshold and greater than or equal to the first index threshold, it is determined that the disease probability is medium;

[0110] When the HE index is less than the third index threshold and greater than or equal to the second index threshold, it is determined that the disease probability is high.

[0111] It should be noted that the dynamic adjustment factor is adjusted according to the actual value of the specific factor, the minimum value and the maximum value of the historical data, ensuring that the HE index can be refined and corrected according to the individual differences of patients during the calculation process. This mechanism enables the diagnostic system to make personalized adjustments according to the specific conditions of patients, thereby improving the accuracy and personalization level of diagnosis.

[0112] By considering the contribution coefficients of different specific factors and specific historical data ranges, the dynamic adjustment factor enables the HE index to have strong adaptability when facing the clinical data of different patients. This flexibility can effectively cope with the changes in the patient's condition, different physiological characteristics, and clinical background differences, providing more accurate diagnosis and warning. The diagnosis and warning module sets multiple index thresholds, enabling the HE index to be classified into different warning levels (low, medium, high). This hierarchical management method can help medical staff more accurately judge the stage of the patient's condition, especially in the early stage, it can timely detect the potential risk of hepatic encephalopathy and carry out early intervention. By setting multiple threshold intervals (low, medium, high), the diagnostic system can provide multi-level warnings for patients, enabling the medical team to have a clearer judgment basis at different diagnostic stages. For example, low-risk patients and high-risk patients can receive different follow-ups and interventions to ensure the most effective allocation of medical resources. By comparing the HE index with multiple preset thresholds, the system can accurately evaluate the disease probability of the patient. The setting of multiple thresholds enables the system to refine the risk assessment, helping doctors better judge the degree of the patient's condition and providing support for subsequent treatment decisions.

[0113] The design of the dynamic adjustment factor enables the HE index to not only reflect the patient's physiological data but also, through personalized adjustment, better fit the patient's historical data and clinical background. This personalized correction mechanism enhances the accuracy of diagnosis, ensuring a high degree of consistency between the results and the patient's actual situation. The diagnostic warning module can evaluate different probabilities of illness (low, medium, high) based on the comparison between the HE index and the preset threshold. This graded warning system can effectively help doctors take different treatment measures at different stages of illness and optimize the treatment path. By calculating and comparing the HE index in real time, changes in the patient's condition can be detected promptly. Especially when the threshold is set reasonably, the system can quickly respond to fluctuations in the patient's state, reduce the lag in the symptoms of hepatic encephalopathy, and ensure that the patient receives timely diagnosis and treatment. Through the multi-level grading of the HE index, the system provides diversified decision-making support for medical staff. Doctors can reasonably select intervention measures based on the HE index level of the patient to ensure the timeliness and accuracy of treatment. This decision-making support system not only improves the diagnostic efficiency but also provides a more personalized treatment plan for the patient.

[0114] Refer to Figure 2 As shown, the embodiment of the present invention provides a detection method for assisting in the diagnosis of hepatic encephalopathy, including:

[0115] S1: Collect the comprehensive physical data of the patient and establish a disease impact dataset based on the comprehensive physical data. The comprehensive physical data includes: blood ammonia concentration, lactic acid concentration, electroencephalogram data, and behavioral data;

[0116] S2: According to the pathological characteristics of hepatic encephalopathy, construct a data importance model, output the influence value of each comprehensive physical data in the disease impact dataset on hepatic encephalopathy based on the data importance model, and select the disease impact dataset based on the influence value to obtain a core impact dataset and a supplementary impact dataset;

[0117] S3: Obtain the core impact dataset, extract the same comprehensive physical data in the core impact dataset and sort it, divide the area according to the preset core partition value, and count the core data quantity of the comprehensive physical data in each area;

[0118] Obtain the supplementary impact dataset and calculate the statistical supplementary data quantity;

[0119] S4: Calculate the HE index according to the core data quantity and the supplementary data quantity;

[0120] S5: Preset multiple index thresholds respectively, compare the HE index with the index thresholds, and divide the warning levels.

[0121] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific embodiments of the present invention, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A detection system for assisting the diagnosis of hepatic encephalopathy, characterized in that: include: The data collection module is configured to collect comprehensive physical data of the patient and establish a disease impact data set based on the comprehensive physical data, wherein the comprehensive physical data includes: blood ammonia concentration, lactic acid concentration, EEG data and behavioral data; a data selection module configured to construct a data importance model according to the pathological characteristics of hepatic encephalopathy, output the impact value of each comprehensive physical data in the disease impact data set on hepatic encephalopathy based on the data importance model, select the disease impact data set based on the impact value, and obtain a core impact data set and a supplementary impact data set; An indicator calculation module is configured to obtain the core impact data set, extract and sort the same comprehensive physical data in the core impact data set, divide the area according to a preset core partition value, and count the number of core data of the comprehensive physical data in each area; Obtaining the supplementary impact data set and calculating the number of statistical supplementary data; A HE index calculation module is configured to calculate the HE index according to the core data quantity and the supplementary data quantity; The diagnosis and early warning module is configured to respectively preset a plurality of index thresholds, compare the HE index with the index thresholds, and divide the early warning levels.

2. The detection system for assisting diagnosis of hepatic encephalopathy according to claim 1, characterized in that: The data selection module constructs a data importance model according to the pathological characteristics of hepatic encephalopathy, including: Wavelet transform or empirical mode decomposition is used to extract blood ammonia fluctuation rate and lactate peak value. EEG signal processing based on convolutional neural network captures the abnormal proportion of brain waves. Motion capture equipment and gait analysis model are used to extract step length fluctuation and standing stability indicators. A spatiotemporal correlation model of patient data is constructed based on a multi-scale graph neural network. The time series changes of each comprehensive physical data are aggregated to capture the relationship between short-term fluctuations and long-term trends, and a dynamic relationship diagram between patient characteristics is established. Initial weights were assigned based on the statistical characteristics of the patient's historical clinical data. A reinforcement learning agent was used to make an initial adjustment to the weight of each comprehensive physical data item with prediction accuracy as a reward signal. Patient-specific factors were introduced to make a secondary adjustment to the weight, and the weight after the secondary adjustment was used as the impact value.

3. The detection system for assisting diagnosis of hepatic encephalopathy according to claim 2, characterized in that: The specific factors include: patient age, number of previous attacks of hepatic encephalopathy and the course of related diseases; the course of the related diseases is the sum of the duration of diabetes, hyperlactatemia, cognitive impairment and dementia.

4. The detection system for assisting diagnosis of hepatic encephalopathy according to claim 3, characterized in that: The impact value is obtained by multiplying the primary adjustment by the secondary adjustment coefficient, and the secondary adjustment coefficient satisfies the following relationship: Where K is the quadratic adjustment coefficient; α is the quadratic adjustment global impact factor, which indicates the degree of influence of the quadratic adjustment on the impact value; N is the total number of patient-specific factors; C i is the contribution coefficient of the i-th specific factor; F i is the actual value of the i-th specificity factor; F i,min is the minimum value of the i-th specific factor in the historical data; F i,min is the maximum value of the i-th specific factor in the historical data.

5. The detection system for assisting diagnosis of hepatic encephalopathy according to claim 4, characterized in that: When the disease impact data set is selected based on the impact value to obtain a core impact data set and a supplementary impact data set, the method includes: An influence threshold is set in advance, and the influence value is compared with the influence threshold; when the influence value is greater than or equal to the influence threshold, the corresponding comprehensive physical data is used as the core influence data set, and when the influence value is less than the influence threshold, the corresponding comprehensive physical data is used as the supplementary influence data set.

6. The detection system for assisting diagnosis of hepatic encephalopathy according to claim 5, characterized in that: The indicator calculation module divides the area according to the preset core partition value, and counts the number of core data of comprehensive body data in each area, including: The core partition value includes a first core partition value and a second core partition value, and the first core partition value is smaller than the second core partition value; The index calculation module divides the same type of comprehensive physical data that is less than the first preset core partition value into a first partition; Classify the same type of comprehensive body data that is greater than the first preset core partition value and less than or equal to the second preset core partition value into a second partition; Classify the same type of comprehensive body data greater than the second preset core partition value into a third partition; Count the number of comprehensive physical data in the first partition, the second partition, and the third partition respectively, and record them as the number of core data; The amount of supplementary data is calculated in the same manner as the amount of core data.

7. The detection system for assisting diagnosis of hepatic encephalopathy according to claim 6, characterized in that: When calculating the HE index based on the core data and supplementary data, it includes: The HE index is calculated by the following relationship: Among them, n1 is the number of core data; N core,i is the number of core data of the i-th category; w i is the weight of the core data of the i-th category; n2 is the number of supplementary data; N supp,j is the number of supplementary data of the jth category; γ is the weight of the dynamic regulatory factor, which controls the role of specific factors in the HE index; D is the dynamic regulatory factor, which is calculated based on the patient's specific factors and corrects the HE index.

8. The detection system for assisting diagnosis of hepatic encephalopathy according to claim 7, characterized in that: The dynamic adjustment factor satisfies the following relationship: Where m is the total number of specific factors, c k is the contribution coefficient of the kth specific factor, F k is the actual value of the kth specificity factor, F k,max 、F k,min are the minimum and maximum values ​​of the kth specific factor in the historical data, respectively.

9. The detection system for assisting diagnosis of hepatic encephalopathy according to claim 8, characterized in that: The diagnosis and early warning module pre-sets a first index threshold and a second index threshold, and the first index threshold is smaller than the second index threshold; The HE index is compared with an index threshold, and when the HE index is less than a first index threshold, it is judged that the probability of the disease is low; When the HE index is less than the second index threshold and greater than or equal to the first index threshold, the probability of illness is judged to be medium; When the HE index is less than the third index threshold and greater than or equal to the second index threshold, the probability of illness is judged to be high.

10. A detection method for assisting diagnosis of hepatic encephalopathy, applied to the detection system for assisting diagnosis of hepatic encephalopathy according to any one of claims 1 to 9, characterized in that: include: S1: Collect the patient's comprehensive physical data and establish a disease impact data set based on the comprehensive physical data. The comprehensive physical data includes: blood ammonia concentration, lactate concentration, EEG data and behavioral data; S2: constructing a data importance model according to the pathological characteristics of hepatic encephalopathy, outputting the impact value of each comprehensive physical data in the disease impact data set on hepatic encephalopathy based on the data importance model, selecting the disease impact data set based on the impact value, and obtaining a core impact data set and a supplementary impact data set; S3: obtaining the core impact data set, extracting and sorting the same comprehensive physical data in the core impact data set, dividing the area according to the preset core partition value, and counting the number of core data of the comprehensive physical data in each area; Obtaining the supplementary impact data set and calculating the number of statistical supplementary data; S4: Calculating the HE index according to the core data quantity and the supplementary data quantity; S5: Preset a plurality of index thresholds respectively, compare the HE index with the index threshold, and divide the warning level.

Images