Comprehensive analysis and intelligent processing system for clinical data of neurosurgery department
By constructing a comprehensive analysis system for neurosurgery clinical data based on the semi-Markov chain, the problem of traditional models insufficient understanding of pathological mechanisms is solved, and the accurate prediction of prognostic status of patients with cerebral hemorrhage is achieved, and the prognostic effect is improved.
Patent Information
- Application Number
- CN202510583758.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-07
- Publication Date
- 2025-08-19
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The traditional cerebral hemorrhage prediction model lacks in-depth understanding of the pathological mechanism, ignores individual differences and dynamic clinical characteristics of patients, resulting in poor prognosis effect.
A comprehensive analysis and intelligent processing system for neurosurgery clinical data is adopted, including data acquisition module, patient weight analysis module, data screening module and prediction module. By constructing a prediction model based on the semi-Markov chain, analyzing the transfer time and state transfer probability, real and reliable clinical data are screened out, and an accurate cerebral hemorrhage prognosis model is constructed.
The prognostic effect of patients with cerebral hemorrhage is improved, and an accurate prognostic status prediction model is obtained by combining the pathological mechanism and the patient's state transfer process, which enhances the predictive ability of cerebral hemorrhage.
Smart Images
Figure CN120511069A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of early warning of cerebral hemorrhage in patients, and in particular to a comprehensive analysis and intelligent processing system for neurosurgery clinical data. Background Art
[0002] Neurosurgery is an important branch of surgery, primarily treating diseases of the brain, spinal cord, and other nervous systems caused by trauma, such as cerebrovascular disease, neurological trauma, and intracranial tumors. Cerebral hemorrhage is a common neurosurgery emergency and an acute cerebrovascular disease. Symptoms of this disease often have a serious impact on the patient's life and can even be fatal. Therefore, prognostic analysis of patients with cerebral hemorrhage is crucial. This can help predict the future course of a patient's condition and allow appropriate measures to be taken, greatly increasing the probability of improvement.
[0003] Based on factors such as clinical manifestations, bleeding site, and amount of bleeding, the condition of patients with cerebral hemorrhage can be divided into multiple levels, and the severity of each level varies greatly. In order to prognose the patient's condition, a commonly used method in medicine is to use predictive models to predict the patient's possible future condition and the corresponding probability based on the patient's current condition, thereby providing early warnings and taking treatment measures. However, in actual applications, the pathological mechanisms of cerebral hemorrhage are relatively complex, and traditional predictive models only rely on the state transition process during predictive analysis, lacking an in-depth understanding of these pathological mechanisms. Moreover, although traditional models are simple and easy to explain, they ignore individual differences in patients and dynamic clinical characteristics. Therefore, in many special scenarios, they are not accurate enough and have poor prognostic effects. Summary of the Invention
[0004] In order to solve the technical problems that traditional prediction models rely solely on state transition processes during prediction and analysis, lack in-depth understanding of pathological mechanisms, and easily ignore individual differences and dynamic clinical characteristics of patients, resulting in insufficient accuracy in prediction and analysis and poor prognosis, the purpose of the present invention is to provide a comprehensive analysis and intelligent processing system for neurosurgery clinical data. The technical solutions adopted are as follows:
[0005] A neurosurgery clinical data comprehensive analysis and intelligent processing system, the system comprising:
[0006] A data acquisition module is used to collect clinical data of all dimensions of each cerebral hemorrhage patient after prognosis at different sampling moments;
[0007] The patient weight analysis module is used to classify all clinical data into a preset number of patient status levels; the process of the disease status of a cerebral hemorrhage patient transferring between different patient status levels is regarded as a state transition process; the transfer time of the cerebral hemorrhage patient in each state transition process is obtained based on the time interval between different patient status levels; the state transition probability of each transfer state process is obtained based on the change characteristics of the number of patients in each patient status level; any one cerebral hemorrhage patient who has undergone prognosis is selected as a reference patient; the data authenticity of the reference patient is obtained based on the transfer time distribution and examination time interval of the reference patient in each state transition process; the data weight coefficient of the reference patient is obtained based on the difference in transfer time between the patient to be prognosed and the reference patient in the same state transition process, as well as the data authenticity of the reference patient;
[0008] A data screening module is configured to select clinical data of any sampling moment in one dimension of a reference patient as reference data; obtain the pseudo-state degree of the reference data based on the difference between the patient state levels of the reference patient after the state transition process, the duration of the patient state level of the reference data, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process; and screen all clinical data based on the pseudo-state possibility degree to obtain data to be fitted;
[0009] The prediction module is used to obtain a cerebral hemorrhage prognosis model for the patient to be prognosed based on all the to-be-fitted data, state transition probability and data weight coefficient of each cerebral hemorrhage patient who has undergone prognosis; and to perform state change analysis on the patient to be prognosed based on the cerebral hemorrhage prognosis model.
[0010] Furthermore, the method for obtaining the state transition probability includes:
[0011] The number of patients with ICH within each patient status level was counted;
[0012] Select one patient status level as the reference level, select another patient status level as the transfer level, and use the state transfer process from the reference level to the transfer level as the first transfer process; calculate the ratio between the number of cerebral hemorrhage patients from the reference level to the transfer level and the number of cerebral hemorrhage patients at the reference level as the state transfer probability of the first transfer process;
[0013] Traverse each state transition process and obtain the state transition probability of each transition state process.
[0014] Furthermore, the method for obtaining the transfer time includes:
[0015] The distribution function of the time intervals between different patient status levels of all patients with cerebral hemorrhage who have undergone prognosis is constructed to obtain the distribution function of the stay time of each cerebral hemorrhage patient at different patient status levels;
[0016] The transition time of each state transition process of the cerebral hemorrhage patient is obtained according to the residence time distribution function.
[0017] Furthermore, the method for obtaining the authenticity of the data includes:
[0018] Calculating the number of occurrences of the transition time length of each state transition process in all state transition processes as a first number, and taking the ratio between the first number and the number of all state transition processes as the distribution probability of the transition time of each state transition process in the residence time distribution function;
[0019] Count the time intervals between two adjacent examinations of the reference patient by relevant personnel during each state transition process;
[0020] The data authenticity is obtained according to the data authenticity calculation formula, which is as follows:
[0021]
[0022] Where ω represents the authenticity of the reference patient's data; K represents the number of state transition processes of the reference patient; H k H represents the distribution probability of the transfer time of the k-th state transfer process in the residence time distribution function; most Indicates the maximum distribution probability of all state transition processes in the residence time distribution function; ΔE k，0 Indicates the inspection time interval between two adjacent inspections during the k-th state transition process; It represents the mean of all inspection time intervals during the k-th state transition process; max() represents the maximum value function.
[0023] Furthermore, the method for obtaining the data weight coefficient includes:
[0024] The data weight coefficient is obtained according to the data weight coefficient calculation formula, which is as follows:
[0025]
[0026] In the formula, β represents the data weight coefficient of the reference patient; C0 represents the time from onset of the patient to be prognosed to the current patient status level; C1 represents the time from onset of the reference patient to the same patient status level; K represents the number of state transition processes of the reference patient; D k，0represents the transition time of the patient to be prognosed in the kth state transition process; D k，1 represents the transition time of the reference patient in the kth state transition process; ω represents the authenticity of the reference patient's data; || represents the absolute value function.
[0027] Furthermore, the method for obtaining the pseudo-state degree includes:
[0028] When the duration of the reference patient in the patient state level is greater than the transition time of the state transition process, the reference patient is in a possible pseudo-stable state; when the duration of the reference patient in the patient state level is less than the transition time of the state transition process, the reference patient is in a possible pseudo-changing state;
[0029] The duration of time that the reference patient remained in each patient status level was counted;
[0030] When the reference patient is in a possible pseudo-stable state, the pseudo-stability degree of the reference data is obtained according to the level difference between the patient state levels of the reference patient after the state transition process, the patient state level duration of the reference data, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process;
[0031] When the reference patient is in a state of possible pseudo-change, the pseudo-change degree of the reference data is obtained according to the level difference between the patient status levels of the reference patient after the state transition process, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process;
[0032] The pseudo-stability degree and the pseudo-change degree are collectively referred to as the pseudo-state degree.
[0033] Furthermore, the method for obtaining the pseudo stability includes:
[0034] The pseudo stability degree is obtained according to the pseudo stability degree calculation formula, and the pseudo stability degree calculation formula is as follows:
[0035]
[0036] Where f1 represents the pseudo stability of the reference data; L nc Indicates the next patient status level sequence number of the patient status level of the reference data after the next state transition process; L uc Indicates the patient status level number of the reference data; T nc T represents the transition time of the patient status level of the reference data after a state transition process; uc represents the duration of the reference patient in the patient status level of the reference data; I represents the number of other dimensions; X irepresents the clinical data of the i-th other dimension of the reference patient after the next state transition process; X i (Ithrs) represents the minimum clinical data of the reference patient in the i-th dimension after the next state transition process; ΔX i (thrs) represents the maximum difference in clinical data of the i-th other dimension of the reference patient after the next state transition process; e represents an exponential function with a natural constant as the base.
[0037] Furthermore, the method for obtaining the pseudo change degree includes:
[0038] The pseudo change degree is obtained according to the pseudo change degree calculation formula, and the pseudo change degree calculation formula is as follows:
[0039]
[0040] Where f2 represents the pseudo-change degree of the reference data; ΔL uc T represents the difference between the patient status level of the reference data and the patient status level after the next state transition process; uc，nc represents the time interval between the last state transition process and the next state transition process of the patient state level where the reference data is located; I represents the number of other dimensions; X i (hthrs) represents the maximum clinical data of the reference patient in the i-th dimension after the next state transition process; X i represents the clinical data of the i-th dimension of the reference patient after the next state transition process; ΔX i (thrs) represents the maximum difference in clinical data of the i-th other dimension of the reference patient after the next state transition process; e represents an exponential function with a natural constant as the base.
[0041] Furthermore, the method for obtaining the data to be fitted includes:
[0042] All clinical data with a pseudo-state degree greater than a preset first threshold are regarded as pseudo-state data;
[0043] All pseudo-state data were screened out, and other clinical data of all patients with cerebral hemorrhage who had undergone prognosis were used as data to be fitted.
[0044] A method for comprehensive analysis and intelligent processing of neurosurgery clinical data, comprising:
[0045] Collect all dimensions of clinical data of each ICH patient after prognosis at different sampling moments;
[0046] All clinical data are divided into a preset number of patient status levels; the process of the disease status of cerebral hemorrhage patients transferring at different patient status levels is regarded as the state transition process; according to the time interval between the cerebral hemorrhage patients at different patient status levels, the transition time of the cerebral hemorrhage patients in each state transition process is obtained; according to the change characteristics of the number of patients in each patient status level, the state transition probability of each transition state process is obtained; any cerebral hemorrhage patient who has undergone prognosis is selected as a reference patient; according to the transfer time distribution and examination time interval of the reference patient in each state transition process, the data authenticity of the reference patient is obtained; according to the difference in transfer time between the patient to be prognosed and the reference patient in the same state transition process, and the data authenticity of the reference patient, the data weight coefficient of the reference patient is obtained;
[0047] The clinical data of any sampling moment of one dimension of the reference patient is optionally used as reference data; the pseudo-state degree of the reference data is obtained based on the level difference between the patient state levels of the reference patient after the state transition process, the patient state level duration of the reference data, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process; all clinical data are screened according to the possible pseudo-state degree to obtain data to be fitted;
[0048] A cerebral hemorrhage prognosis model for the patient to be prognosed is obtained based on all the to-be-fitted data, state transition probability, and data weight coefficient of each cerebral hemorrhage patient who has undergone prognosis; and a state change analysis is performed on the patient to be prognosed based on the cerebral hemorrhage prognosis model.
[0049] The present invention has the following beneficial effects:
[0050] In order to facilitate the prognosis of cerebral hemorrhage patients by relevant personnel, the present invention obtains clinical data of multiple dimensions of different cerebral hemorrhage patients who have undergone prognosis at different sampling times; all clinical data are divided into a preset number of patient status levels; since the time of disease change of various states of cerebral hemorrhage patients is not the same, in reality, there are often cases where some stages take a long time and some stages change very quickly, so in order to fit such cases, a semi-Markov chain is used to construct a prediction model. Since the prediction model based on the semi-Markov chain needs to reflect the non-uniform changes in the disease state of cerebral hemorrhage patients by the stay time of cerebral hemorrhage patients in different disease stages, so as to enhance the prediction model's mining of the pathological mechanism of cerebral hemorrhage, the transition time of cerebral hemorrhage patients in each state transition process and the state transition probability of each state transition process are analyzed, and the cerebral hemorrhage pathology is summarized and analyzed by the transition time and state transition probability, so as to facilitate the subsequent analysis of the prognostic status of the patients to be prognosed; since there is often a large amount of low-quality data in the data samples obtained in actual application, in order to enable the prediction model to obtain accurate prognostic results, the data authenticity of each cerebral hemorrhage patient who has undergone prognosis is first analyzed, and the weights of different cerebral hemorrhage patients are assigned according to the data authenticity, and the data weight coefficient is used to assign weights to each The patients with cerebral hemorrhage who have undergone prognosis are limited; since the data used in constructing the prediction model are mostly intermittent detection data, there may be a situation where the patient's state changes cannot be reflected in time, and the data used in constructing the prediction model are mostly intermittent detection data, there may be a situation where the patient's cerebral hemorrhage is in a pseudo-state, which makes some clinical data not real and has a strong interference with the construction of the prediction model. Therefore, the pseudo-state degree of the clinical data is analyzed, and the clinical data in the pseudo-state is screened out by the pseudo-state degree to obtain the data to be fitted; according to all the data to be fitted, the state transition probability and the data weight coefficient of each cerebral hemorrhage patient who has undergone prognosis, the cerebral hemorrhage prognosis model of the patient to be prognosed is obtained; according to the cerebral hemorrhage prognosis model, the state change analysis of the patient to be prognosed is performed. The present invention combines the patient's cerebral hemorrhage state transition process and the pathological mechanism to obtain an accurate prognostic state prediction model, thereby improving the prognosis effect of the patient to be prognosed. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0052] Figure 1 A block diagram of a neurosurgery clinical data comprehensive analysis and intelligent processing system provided by one embodiment of the present invention. DETAILED DESCRIPTION
[0053] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of a comprehensive neurosurgical clinical data analysis and intelligent processing system proposed by the present invention. In the following description, references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
[0054] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0055] The specific scheme of the neurosurgery clinical data comprehensive analysis and intelligent processing system provided by the present invention is described in detail below with reference to the accompanying drawings.
[0056] See also Figure 1 , which shows a neurosurgery clinical data comprehensive analysis and intelligent processing system provided by one embodiment of the present invention. The system includes: a data acquisition module 101, a patient weight analysis module 102, a data screening module 103, and a prediction module 104, specifically including:
[0057] Data acquisition module 101: collects clinical data of all dimensions of each cerebral hemorrhage patient with prognosis at different sampling moments.
[0058] The embodiment of the present invention is mainly used in the scenario of predicting and analyzing the prognosis status of patients with cerebral hemorrhage, so that relevant personnel can perform prognostic treatment on patients with cerebral hemorrhage. Therefore, the embodiment of the present invention first obtains clinical data of multiple dimensions of different patients with cerebral hemorrhage who have undergone prognosis at different sampling times, and constructs a prediction model to perform prognostic status prediction analysis on patients to be prognosed.
[0059] In one embodiment of the present invention, the sampling time is set to 1 day, and the dimensions of the clinical data are set to blood glucose, hemoglobin, and intracranial pressure. It should be noted that the sampling time and clinical data dimensions can be set arbitrarily and are not limited here.
[0060] Patient weight analysis module 102: divide all clinical data into a preset number of patient status levels; regard the process of the disease status of the cerebral hemorrhage patient transferring between different patient status levels as the state transfer process; obtain the transfer time of the cerebral hemorrhage patient in each state transfer process according to the time interval between the cerebral hemorrhage patients at different patient status levels; obtain the state transfer probability of each transfer state process according to the change characteristics of the number of patients in each patient status level; select any cerebral hemorrhage patient who has undergone prognosis as a reference patient; obtain the data authenticity of the reference patient according to the transfer time distribution and examination time interval of the reference patient in each state transfer process; obtain the data weight coefficient of the reference patient according to the difference in transfer time between the patient to be prognosed and the reference patient in the same state transfer process, as well as the data authenticity of the reference patient.
[0061] The Glasgow Coma Scale is used to divide the degree of cerebral hemorrhage into five levels, namely grade I cerebral hemorrhage, grade II cerebral hemorrhage, grade III cerebral hemorrhage, grade IV cerebral hemorrhage and grade V cerebral hemorrhage, which together with the healthy state and death state constitute seven patient status levels, that is, the preset number is 7; according to the clinical data standards of each dimension within each patient status level, the clinical data of all dimensions are divided into each patient status level to facilitate the subsequent analysis of the physical state transfer of cerebral hemorrhage patients.
[0062] Since the time it takes for the various states of a cerebral hemorrhage patient to change is not the same, in reality, some stages often take a long time while others change very quickly. Therefore, in order to accommodate such situations, in an embodiment of the present invention, a semi-Markov chain is used to construct a prediction model. Since the prediction model based on the semi-Markov chain needs to reflect the non-uniform changes in the condition of the cerebral hemorrhage patient through the time the cerebral hemorrhage patient stays at different stages of the disease, in order to enhance the prediction model's mining of the pathological mechanism of cerebral hemorrhage. Therefore, in an embodiment of the present invention, the process of the cerebral hemorrhage patient's disease state transitioning at different patient state levels is first taken as the state transition process. Based on the time interval between the cerebral hemorrhage patient's different patient state levels, the transition time of the cerebral hemorrhage patient in each state transition process is obtained, and based on the change characteristics of the number of patients in each patient state level, the state transition probability of each transition state process is obtained. The cerebral hemorrhage pathology is summarized and analyzed through the transition time and state transition probability of each state transition process, so as to facilitate the subsequent analysis of the prognostic status of the patient to be treated.
[0063] Preferably, in one embodiment of the present invention, the method for obtaining the transfer time includes:
[0064] A Weibull distribution was used to construct a distribution function for the time intervals between different patient status levels for all patients with prognostic cerebral hemorrhage. The parameters of the Weibull distribution were obtained using a maximum likelihood estimation algorithm, and the residence time distribution function for each cerebral hemorrhage patient at different patient status levels was obtained. The transition time of each cerebral hemorrhage patient during each state transition was then obtained based on the residence time distribution function. It should be noted that the above process is a well-known technical approach by those skilled in the art and will not be elaborated upon here.
[0065] Preferably, in one embodiment of the present invention, the method for obtaining the state transition probability includes:
[0066] According to the Glasgow Coma Scale score of all patients with cerebral hemorrhage who have undergone prognosis by relevant institutions, the number of patients with cerebral hemorrhage in each patient status level was counted.
[0067] Any one patient status level is selected as the reference level, any other patient status level is selected as the transfer level, and the state transition process from the reference level to the transfer level is selected as the first transfer process; the ratio between the number of cerebral hemorrhage patients from the reference level to the transfer level and the number of cerebral hemorrhage patients at the reference level is calculated as the state transition probability of the first transfer process; each state transition process is traversed to obtain the state transition probability of each transfer state process.
[0068] In practice, the accuracy of the state transition probability depends on the quality of the data sample. The larger and more realistic the data, the more accurate the transition probability. However, in actual applications, the data samples obtained often contain a large amount of low-quality data. For example, the condition of some patients with cerebral hemorrhage may originally only transition between adjacent states, but due to untimely examination, the condition status shows a cross-stage transition. Therefore, in an embodiment of the present invention, in order to enable the prediction model to obtain accurate prognostic results, the authenticity of the data of each cerebral hemorrhage patient who has undergone prognosis is first analyzed. The authenticity of the data is then used to assign weights to different cerebral hemorrhage patients, and the data weight coefficient is used to limit each cerebral hemorrhage patient who has undergone prognosis.
[0069] Preferably, in one embodiment of the present invention, the method for obtaining the authenticity of data includes:
[0070] The number of occurrences of the transfer time length of each state transfer process in all state transfer processes is calculated as a first number, and the ratio between the first number and the number of all state transfer processes is used as the distribution probability of the transfer time of each state transfer process in the residence time distribution function.
[0071] The time interval between two adjacent examinations of the reference patient during each state transition by the relevant personnel can be directly obtained by the existing technology.
[0072] The data authenticity is obtained according to the data authenticity calculation formula. The data authenticity calculation formula is as follows:
[0073]
[0074] Where ω represents the authenticity of the reference patient's data; K represents the number of state transition processes of the reference patient; H k H represents the distribution probability of the transfer time of the k-th state transfer process in the residence time distribution function; most Indicates the maximum distribution probability of all state transition processes in the residence time distribution function; ΔE k，0 Indicates the inspection time interval between two adjacent inspections during the k-th state transition process; It represents the mean of all inspection time intervals during the k-th state transition process; max() represents the maximum value function.
[0075] In the data authenticity calculation formula, the greater the distribution probability of the kth state transition process, the greater the ratio of the maximum distribution probability of the residence time distribution function of all state transition processes. The larger it is, the more common the time length of the k-th state transition process is. Each state transition process is analyzed. The greater the distribution probability of each state transition process, the greater the authenticity of the reference patient's data. If the interval between two adjacent examination times in the k-th state transition process exceeds the mean of all examination time intervals, it means that the clinical data corresponding to the k-th state transition process is unreliable and the authenticity is worse. The max function is used to indicate that the interval between two adjacent examination times in the k-th state transition process is lower than the mean of the time interval. At this time, the clinical data of the k-th state transition process is considered to be more reliable. Returns 0.
[0076] Preferably, in one embodiment of the present invention, the method for obtaining the data weight coefficient includes:
[0077] In order to predict the prognostic status of patients with cerebral hemorrhage who have not undergone prognosis, we first select patients with cerebral hemorrhage who have undergone prognosis and have the same patient status level as the patients who have undergone prognosis to participate in the subsequent model construction. Therefore, in an embodiment of the present invention, the weight of the reference patient when constructing the prediction model is analyzed by comparing the state transition process differences between the reference patient and the patients to be prognosed who have the same patient status level, and then traverse all patients with cerebral hemorrhage who have undergone prognosis and have the same patient status level to obtain the data weight coefficient of each patient in the prediction model of the patients to be prognosed.
[0078] The data weight coefficient is obtained according to the data weight coefficient calculation formula. The data weight coefficient calculation formula is as follows:
[0079]
[0080] In the formula, β represents the data weight coefficient of the reference patient; C0 represents the time from onset of the patient to be prognosed to the current patient status level; C1 represents the time from onset of the reference patient to the same patient status level; K represents the number of state transition processes of the reference patient; D k，0 represents the transition time of the patient to be prognosed in the kth state transition process; D k，1 represents the transition time of the reference patient in the kth state transition process; ω represents the authenticity of the reference patient's data; || represents the absolute value function.
[0081] In the data weight coefficient calculation formula, the smaller the time difference |C0-C1| between the patient to be prognosed and the reference patient from onset to the current patient status level, the smaller the difference in the state transition process between the patient to be prognosed and the reference patient. At this time, the greater the weight of the reference patient in the prediction model, that is, the greater the data weight coefficient; the smaller the difference in the transition time between the patient to be prognosed and the reference patient in each state transition process, the smaller the difference in the transition time between the patient to be prognosed and the reference patient. The smaller it is, the closer the disease progression of the patient to be tested is to that of the reference patient. At this time, the greater the weight of the patient in the prediction model, that is, the greater the data weight coefficient; the greater the authenticity of the reference patient's data, that is, the more authentic the clinical data of the reference patient, the greater the weight of the reference patient in the prediction model, that is, the greater the data weight coefficient.
[0082] Data screening module 103: clinical data of any sampling moment of one dimension of the reference patient is selected as reference data; according to the level difference between the patient state levels of the reference patient after the state transition process, the patient state level duration of the reference data, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process, the pseudo-state degree of the reference data is obtained; all clinical data are screened according to the possible degree of pseudo-state to obtain the data to be fitted.
[0083] In reality, the clinical data of patients with cerebral hemorrhage are often unable to meet the requirements of real-time monitoring. Therefore, the data used in constructing the prediction model are mostly intermittent detection data, which may not be able to reflect the changes in the patient's condition in a timely manner. In addition, the judgment of the cerebral hemorrhage status is usually based on the fixed threshold of physiological indicators, rather than flexible judgment based on the patient's condition, which will cause the patient's apparent condition to be disconnected from the actual pathological progression. That is, at this time, there will be a situation where the cerebral hemorrhage patient is in a pseudo-state, which makes some clinical data not true, and has a strong interference with the construction of the prediction model. Therefore, in the embodiment of the present invention, the pseudo-state degree of the clinical data is first analyzed, and the clinical data in the pseudo-state is screened out according to the pseudo-state degree.
[0084] Preferably, in one embodiment of the present invention, the method for obtaining the pseudo state degree includes:
[0085] In reality, clinical data of different dimensions related to cerebral hemorrhage have a cumulative effect, that is, the physiological indicators of a certain dimension will continue to show a slight negative trend of change, but the clinical data at a single sampling moment does not reach the critical value, and eventually a cliff-like deterioration occurs over a period of time, resulting in a sudden change in the cerebral hemorrhage state. At this time, the reference patient is considered to be in a pseudo-stable state, that is, when the duration of the reference patient's patient status level is greater than the transfer time of the state transition process, the reference patient is in a possible pseudo-stable state.
[0086] When the reference patient receives intervention from relevant personnel or compensates on his own, the reference patient's condition temporarily improves. However, in fact, the clinical data in other dimensions change significantly, and the reference patient's condition does not improve. For example, after sodium supplementation, the patient's blood sodium level may quickly return to the normal range, but the osmotic pressure balance in the brain cells has not yet been rebuilt and may deteriorate to the original state again. At this time, the reference patient is in a pseudo-change state, that is, when the duration of the reference patient's patient status level is less than the transfer time of the state transition process, the reference patient is in a possible pseudo-change state.
[0087] The duration of the statistical reference patient in each patient status level can be directly obtained by existing technology.
[0088] When the reference patient is in a possible pseudo-stable state, the pseudo-stability of the reference data is obtained based on the difference in the patient state levels of the reference patient after the state transition process, the duration of the patient state level of the reference data, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process. The pseudo-stability calculation formula is as follows:
[0089]
[0090] Where f1 represents the pseudo stability of the reference data; L nc Indicates the next patient status level sequence number of the patient status level of the reference data after the next state transition process; L uc Indicates the patient status level number of the reference data; T nc T represents the transition time of the patient status level of the reference data after a state transition process; uc represents the duration of the reference patient in the patient status level of the reference data; I represents the number of other dimensions; X i represents the clinical data of the i-th other dimension of the reference patient after the next state transition process; X i(lthrs) represents the minimum clinical data of the reference patient in the i-th dimension after the next state transition process, that is, the minimum threshold of the physiological indicators of each dimension in the patient's state level, which can be directly obtained through existing technology; ΔX i (thrs) represents the maximum difference in clinical data of the i-th dimension of the reference patient after the next state transition process, that is, the difference between the maximum threshold and the minimum threshold of the physiological indicators of each dimension in the patient's state level; e represents an exponential function with a natural constant as the base.
[0091] In the pseudo-stability calculation formula, after a state transition process, the patient state level difference L of the reference patient nc -L uc The larger the value, the greater the degree of mutation of the patient's state. At this time, the reference data is more likely to be in a pseudo-stable state, and the greater the pseudo-stability of the reference data. The ratio of the transfer time to the duration of the reference patient in the patient state level of the reference data is The smaller it is, the greater the degree of mutation of the patient's state, the more likely the reference data is in a pseudo-stable state, and the greater the pseudo-stability of the reference data; the ratio of the difference between the clinical data of the other dimension of the i-th dimension after a state transition process and the minimum threshold of the patient's state level at this time to the maximum difference between the clinical data The larger the value is, the greater the fluctuation of the data in this dimension is. The clinical data of all other dimensions are analyzed. The larger the value is, the greater the fluctuation of the physiological indicators of the reference patient is, and the greater the pseudo-stability of the reference data is.
[0092] When the reference patient is in a state that may be subject to pseudo-change, the pseudo-change degree of the reference data is obtained based on the difference between the patient status levels of the reference patient after the state transition process, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process. The method for obtaining the pseudo-change degree includes:
[0093] The pseudo change degree is obtained according to the pseudo change degree calculation formula. The pseudo change degree calculation formula is as follows:
[0094]
[0095] Where f2 represents the pseudo-change degree of the reference data; ΔL uc T represents the difference between the patient status level of the reference data and the patient status level after the next state transition process; uc，nc represents the time interval between the last state transition process and the next state transition process of the patient state level where the reference data is located; I represents the number of other dimensions; Xi (hthrs) represents the maximum clinical data of the reference patient in the i-th dimension after the next state transition process, which is the maximum threshold of the physiological index of each dimension in the patient's state level and can be directly obtained through existing technology; X i represents the clinical data of the i-th dimension of the reference patient after the next state transition process; ΔX i (thrs) represents the maximum difference in clinical data of the i-th other dimension of the reference patient after the next state transition process; e represents an exponential function with a natural constant as the base.
[0096] In the pseudo-change degree calculation formula, after a state transition process, the patient state level difference ΔL of the reference patient is uc The smaller it is, the more likely the reference patient is in a pseudo-change state, that is, the greater the pseudo-change degree of the reference data; the smaller the time interval between two adjacent state transitions, the more frequent the changes in the reference patient's state. However, since the physiological state of the reference patient has not changed significantly, the more likely the reference patient is in a pseudo-change state, that is, the greater the pseudo-change degree of the reference data; the ratio of the difference between the maximum threshold of the clinical data of the i-th other dimension and the clinical data of the dimension after a state transition process to the maximum difference in the clinical data The larger the value is, the weaker the fluctuation of clinical data in the i-th dimension is. The clinical data of all other dimensions are analyzed. The larger it is, the smaller the state fluctuation of the reference patient is, which means that the reference patient is more likely to be in a pseudo-change state, that is, the greater the pseudo-change degree of the reference data.
[0097] The pseudo-stability degree and the pseudo-change degree are collectively referred to as the pseudo-state degree.
[0098] Preferably, in one embodiment of the present invention, the method for obtaining the data to be fitted includes:
[0099] All clinical data with a pseudo-state degree greater than a preset first threshold are regarded as pseudo-state data. In one embodiment of the present invention, the preset first threshold is set to 0.8. It should be noted that the preset first threshold can be set arbitrarily and is not limited here.
[0100] All pseudo-state data were screened out, and other clinical data of all patients with cerebral hemorrhage who had undergone prognosis were used as data to be fitted.
[0101] Prediction module 104: Obtain a cerebral hemorrhage prognosis model for each patient with cerebral hemorrhage who has undergone prognosis based on all the data to be fitted, the state transition probability, and the data weight coefficient; and perform state change analysis on the patient with cerebral hemorrhage according to the cerebral hemorrhage prognosis model.
[0102] In one embodiment of the present invention, the latest state transition probability for each state transition process is obtained based on the weight coefficients of the data sample to be fitted and the data of each cerebral hemorrhage patient. The specific calculation process has been described in detail in module 102 and is not repeated here. Based on the construction method of the prediction model based on the semi-Markov chain, a prognostic model for cerebral hemorrhage in the patient to be prognosed is obtained. It should be noted that this process is a technical means well known to those skilled in the art and is not described in detail here.
[0103] The model is used to analyze the state transitions of patients with a prognosis and determine their potential future disease progression. A progression is a combination of possible state changes and duration. Ultimately, this model will assist relevant personnel in the prognosis management of patients with cerebral hemorrhage.
[0104] The second object of the present invention is to provide a method for comprehensive analysis and intelligent processing of neurosurgery clinical data, which specifically includes:
[0105] Collect all dimensions of clinical data of each ICH patient after prognosis at different sampling moments;
[0106] All clinical data are divided into a preset number of patient status levels; the process of the disease status of cerebral hemorrhage patients transferring at different patient status levels is regarded as the state transition process; according to the time interval between the cerebral hemorrhage patients at different patient status levels, the transition time of the cerebral hemorrhage patients in each state transition process is obtained; according to the change characteristics of the number of patients in each patient status level, the state transition probability of each transition state process is obtained; any cerebral hemorrhage patient who has undergone prognosis is selected as a reference patient; according to the transfer time distribution and examination time interval of the reference patient in each state transition process, the data authenticity of the reference patient is obtained; according to the difference in transfer time between the patient to be prognosed and the reference patient in the same state transition process, and the data authenticity of the reference patient, the data weight coefficient of the reference patient is obtained;
[0107] The clinical data of any sampling moment of one dimension of the reference patient is selected as the reference data; the pseudo-state degree of the reference data is obtained based on the level difference between the patient state levels of the reference patient after the state transition process, the patient state level duration of the reference data, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process; all clinical data are screened according to the possible degree of pseudo-state to obtain the data to be fitted;
[0108] According to all the data to be fitted, the state transition probability and the data weight coefficient of each cerebral hemorrhage patient who has undergone prognosis, the cerebral hemorrhage prognostic model of the patient to be prognosed is obtained; and the state change analysis of the patient to be prognosed is performed according to the cerebral hemorrhage prognostic model.
[0109] In summary, the present invention collects clinical data of all dimensions of each cerebral hemorrhage patient who has undergone prognosis at different sampling moments; divides all clinical data into a preset number of patient status levels; regards the process of transferring the disease status of the cerebral hemorrhage patient at different patient status levels as a state transfer process; obtains the transfer time of the cerebral hemorrhage patient in each state transfer process according to the time interval between the cerebral hemorrhage patients at different patient status levels; obtains the state transfer probability of each transfer state process according to the change characteristics of the number of patients in each patient status level; selects any cerebral hemorrhage patient who has undergone prognosis as a reference patient; obtains the data authenticity of the reference patient according to the transfer time distribution and the inspection time interval of the reference patient in each state transfer process; obtains the data authenticity of the reference patient according to the time distribution of the patient to be prognosed and the reference patient in the same state transfer process The data weight coefficient of the reference patient is obtained based on the difference in transfer time and the authenticity of the data of the reference patient; the clinical data of any sampling moment of one dimension of the reference patient is selected as the reference data; the pseudo-state degree of the reference data is obtained based on the level difference between the patient state levels of the reference patient after the state transition process, the duration of the patient state level of the reference data, the transfer time of the state transition process, and the clinical data of other dimensions after the state transition process; all clinical data are screened according to the possible degree of pseudo-state to obtain the data to be fitted; the cerebral hemorrhage prognosis model of the patient to be prognosed is obtained based on all the data to be fitted, the state transition probability and the data weight coefficient of each cerebral hemorrhage patient who has undergone prognosis; and the state change analysis of the patient to be prognosed is performed according to the cerebral hemorrhage prognosis model.
[0110] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0111] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.
Claims
1. A neurosurgery clinical data comprehensive analysis and intelligent processing system, characterized by: The system comprises: A data acquisition module is used to collect clinical data of all dimensions of each cerebral hemorrhage patient after prognosis at different sampling moments; The patient weight analysis module is used to classify all clinical data into a preset number of patient status levels; the process of the disease status of a cerebral hemorrhage patient transferring between different patient status levels is regarded as a state transition process; the transfer time of the cerebral hemorrhage patient in each state transition process is obtained based on the time interval between different patient status levels; the state transition probability of each transfer state process is obtained based on the change characteristics of the number of patients in each patient status level; any one cerebral hemorrhage patient who has undergone prognosis is selected as a reference patient; the data authenticity of the reference patient is obtained based on the transfer time distribution and examination time interval of the reference patient in each state transition process; the data weight coefficient of the reference patient is obtained based on the difference in transfer time between the patient to be prognosed and the reference patient in the same state transition process, as well as the data authenticity of the reference patient; A data screening module is configured to select clinical data of any sampling moment in one dimension of a reference patient as reference data; obtain the pseudo-state degree of the reference data based on the difference between the patient state levels of the reference patient after the state transition process, the duration of the patient state level of the reference data, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process; and screen all clinical data based on the pseudo-state possibility degree to obtain data to be fitted; The prediction module is used to obtain a cerebral hemorrhage prognosis model for the patient to be prognosed based on all the to-be-fitted data, state transition probability and data weight coefficient of each cerebral hemorrhage patient who has undergone prognosis; and to perform state change analysis on the patient to be prognosed based on the cerebral hemorrhage prognosis model.
2. A neurosurgery clinical data comprehensive analysis and intelligent processing system according to claim 1, characterized in that: The method for obtaining the state transition probability includes: The number of patients with ICH within each patient status level was counted; Select one patient status level as the reference level, select another patient status level as the transfer level, and use the state transfer process from the reference level to the transfer level as the first transfer process; calculate the ratio between the number of cerebral hemorrhage patients from the reference level to the transfer level and the number of cerebral hemorrhage patients at the reference level as the state transfer probability of the first transfer process; Traverse each state transition process and obtain the state transition probability of each transition state process.
3. A neurosurgery clinical data comprehensive analysis and intelligent processing system according to claim 1, characterized in that: The method for obtaining the transfer time includes: The distribution function of the time intervals between different patient status levels of all patients with cerebral hemorrhage who have undergone prognosis is constructed to obtain the distribution function of the stay time of each cerebral hemorrhage patient at different patient status levels; The transition time of each state transition process of the cerebral hemorrhage patient is obtained according to the residence time distribution function.
4. A neurosurgery clinical data comprehensive analysis and intelligent processing system according to claim 1, characterized in that: The method for obtaining the authenticity of the data includes: Calculating the number of occurrences of the transition time length of each state transition process in all state transition processes as a first number, and taking the ratio between the first number and the number of all state transition processes as the distribution probability of the transition time of each state transition process in the residence time distribution function; Count the time intervals between two adjacent examinations of the reference patient by relevant personnel during each state transition process; The data authenticity is obtained according to the data authenticity calculation formula, which is as follows: Where ω represents the authenticity of the reference patient's data; K represents the number of state transition processes of the reference patient; H k Represents the distribution probability of the transfer time of the k-th state transfer process in the residence time distribution function; H most Indicates the maximum distribution probability of all state transition processes in the residence time distribution function; ΔE k，0 Indicates the inspection time interval between two adjacent inspections during the k-th state transition process; It represents the mean of all inspection time intervals during the k-th state transition process; max() represents the maximum value function.
5. A neurosurgery clinical data comprehensive analysis and intelligent processing system according to claim 1, characterized in that: The method for obtaining the data weight coefficient includes: The data weight coefficient is obtained according to the data weight coefficient calculation formula, which is as follows: In the formula, β represents the data weight coefficient of the reference patient; C0 represents the time from onset of the patient to be prognosed to the current patient status level; C1 represents the time from onset of the reference patient to the same patient status level; K represents the number of state transition processes of the reference patient; D k，0 represents the transition time of the patient to be prognosed in the kth state transition process; D k，1 represents the transition time of the reference patient in the kth state transition process; ω represents the authenticity of the reference patient's data; || represents the absolute value function.
6. A neurosurgery clinical data comprehensive analysis and intelligent processing system according to claim 1, characterized in that: The method for obtaining the pseudo-state degree includes: When the duration of the reference patient in the patient state level is greater than the transition time of the state transition process, the reference patient is in a possible pseudo-stable state; when the duration of the reference patient in the patient state level is less than the transition time of the state transition process, the reference patient is in a possible pseudo-changing state; The duration of time that the reference patient remained in each patient status level was counted; When the reference patient is in a possible pseudo-stable state, the pseudo-stability degree of the reference data is obtained according to the level difference between the patient state levels of the reference patient after the state transition process, the patient state level duration of the reference data, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process; When the reference patient is in a state of possible pseudo-change, the pseudo-change degree of the reference data is obtained according to the level difference between the patient status levels of the reference patient after the state transition process, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process; The pseudo-stability degree and the pseudo-change degree are collectively referred to as the pseudo-state degree.
7. A neurosurgery clinical data comprehensive analysis and intelligent processing system according to claim 6, characterized in that: The method for obtaining the pseudo stability degree includes: The pseudo stability degree is obtained according to the pseudo stability degree calculation formula, and the pseudo stability degree calculation formula is as follows: Where f1 represents the pseudo stability of the reference data; K nc Indicates the next patient status level sequence number of the patient status level of the reference data after the next state transition process; L uc Indicates the patient status level number of the reference data; T nc T represents the transition time of the patient status level of the reference data after a state transition process; uc represents the duration of the reference patient in the patient status level of the reference data; I represents the number of other dimensions; X i represents the clinical data of the i-th other dimension of the reference patient after the next state transition process; X i (lthrs) represents the minimum clinical data of the reference patient in the i-th dimension after the next state transition process; ΔX i (thrs) represents the maximum difference in clinical data of the i-th other dimension of the reference patient after the next state transition process; e represents an exponential function with a natural constant as the base.
8. A neurosurgery clinical data comprehensive analysis and intelligent processing system according to claim 1, characterized in that: The method for obtaining the pseudo change degree includes: The pseudo change degree is obtained according to the pseudo change degree calculation formula, and the pseudo change degree calculation formula is as follows: Where f2 represents the pseudo-change degree of the reference data; ΔL uc T represents the difference between the patient status level of the reference data and the patient status level after the next state transition process; uc，nc represents the time interval between the last state transition process and the next state transition process of the patient state level where the reference data is located; I represents the number of other dimensions; X i (hthrs) represents the maximum clinical data of the reference patient in the i-th dimension after the next state transition process; X i represents the clinical data of the reference patient in the i-th dimension after the next state transition process; ΔX i (thrs) represents the maximum difference in clinical data of the i-th other dimension of the reference patient after the next state transition process; e represents an exponential function with a natural constant as the base.
9. A neurosurgery clinical data comprehensive analysis and intelligent processing system according to claim 1, characterized in that: The method for obtaining the data to be fitted includes: All clinical data with a pseudo-state degree greater than a preset first threshold are regarded as pseudo-state data; All pseudo-state data were screened out, and other clinical data of all patients with cerebral hemorrhage who had undergone prognosis were used as data to be fitted.
10. A method for comprehensive analysis and intelligent processing of neurosurgery clinical data, characterized in that: The method comprises: Collect all dimensions of clinical data of each ICH patient after prognosis at different sampling moments; All clinical data are divided into a preset number of patient status levels; the process of the disease status of cerebral hemorrhage patients transferring at different patient status levels is regarded as the state transition process; according to the time interval between the cerebral hemorrhage patients at different patient status levels, the transition time of the cerebral hemorrhage patients in each state transition process is obtained; according to the change characteristics of the number of patients in each patient status level, the state transition probability of each transition state process is obtained; any cerebral hemorrhage patient who has undergone prognosis is selected as a reference patient; according to the transfer time distribution and examination time interval of the reference patient in each state transition process, the data authenticity of the reference patient is obtained; according to the difference in transfer time between the patient to be prognosed and the reference patient in the same state transition process, and the data authenticity of the reference patient, the data weight coefficient of the reference patient is obtained; The clinical data of any sampling moment of one dimension of the reference patient is optionally used as reference data; the pseudo-state degree of the reference data is obtained based on the level difference between the patient state levels of the reference patient after the state transition process, the patient state level duration of the reference data, the transition time of the state transition process, and the clinical data of other dimensions after the state transition process; all clinical data are screened according to the possible pseudo-state degree to obtain data to be fitted; A cerebral hemorrhage prognosis model for the patient to be prognosed is obtained based on all the to-be-fitted data, state transition probability, and data weight coefficient of each cerebral hemorrhage patient who has undergone prognosis; and a state change analysis is performed on the patient to be prognosed based on the cerebral hemorrhage prognosis model.