Dynamic evaluation method and system for lung cancer based on immune monitoring

Abstract

The present invention relates to the technical field of medical data analysis, specifically a dynamic evaluation method and system for lung cancer based on immune monitoring, including the following steps: collecting immune index data obtained during the immune monitoring of patients, analyzing the mean change between adjacent sections of the time series of the immune index data, marking the abnormally fluctuating regions, and generating abnormally fluctuating section data. In the present invention, by dividing the patient's immune monitoring data into time series and accurately extracting the abnormally fluctuating sections and abnormally changing points through mean change analysis and the sliding window method, the significant fluctuations of the immune system can be captured more quickly and accurately, thereby improving the timeliness of dynamic monitoring. Further combining statistical feature analysis and similarity matrix calculation, the differential grouping of immune indexes among patients is refined, which helps to conduct personalized evaluations for different patient groups with different immune states.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data analysis, and in particular to a dynamic evaluation method and system for lung cancer based on immune monitoring. Background Art

[0002] The field of medical data analysis technology includes the collection, storage, processing, analysis and application of medical data. The core content of this technology field is to support medical activities such as disease diagnosis, health monitoring, treatment planning and disease prevention through systematic processing and in-depth analysis of various types of medical data. Medical data analysis usually includes the fusion and modeling of multimodal data to achieve accurate mining and use of medical information.

[0003] Among them, the dynamic assessment method for lung cancer refers to obtaining and analyzing the immune-related data of lung cancer patients to conduct real-time dynamic assessment of the patient's immune system status and its changes. It includes collecting the patient's immune parameters through immune monitoring and processing and analyzing these data in real time. This method is based on the results of immune monitoring, and through multiple measurements and data comparisons, analyzes the immune changes related to lung cancer, thereby achieving dynamic assessment and monitoring of the patient's disease status.

[0004] It is difficult for existing technologies to quickly capture local abnormal fluctuations in immune monitoring data. For example, when a patient's immune parameters suddenly rise or fall in a short period of time, it is usually impossible to accurately locate the key change area through simple data comparison, and it is easy to miss important immune response information. For example, when the patient's C-reactive protein concentration fluctuates significantly, existing methods often find it difficult to quickly mark these fluctuation areas. For the group differentiation analysis of patient immune data, existing technologies pay less attention to the quantification of specific similarities in characteristics between patients, resulting in the inability to accurately divide immune change patterns in patient groups. For the complexity assessment of immune data, existing technologies usually ignore the dynamic analysis of entropy changes and cannot distinguish between long-term trends and short-term fluctuations in patient immune data, which may lead to misjudgment or omission of key trend points. In terms of threshold adjustment, it is difficult for existing technologies to dynamically update evaluation parameters based on individual patient characteristics and time series data, which can easily lead to overly fixed evaluation criteria, such as failure to adjust the adaptive evaluation threshold in dynamic disease states, thereby affecting the accuracy and flexibility of the evaluation results. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a dynamic evaluation method and system for lung cancer based on immune monitoring.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a dynamic evaluation method for lung cancer based on immune monitoring, comprising the following steps:

[0007] S1: Collect immune index data obtained during the patient immune monitoring process, analyze the mean changes between adjacent segments of the immune index data time series, mark abnormal fluctuation areas, and generate abnormal fluctuation segment data;

[0008] S2: analyzing the abnormal fluctuation section data through a sliding window, selecting abnormal change points, and obtaining abnormal change point distribution data by determining the positions of the abnormal change points;

[0009] S3: extracting statistical features of the immune index data corresponding to the abnormal change point distribution data, and clustering the immune index data among patients to generate immune response patient grouping features;

[0010] S4: performing a distribution entropy analysis on the time series of the immune index data of each group of patients in the immune response patient grouping feature, calculating the local entropy value in the corresponding time window, accumulating the local entropy value into an entropy value change curve, marking the key feature points with reference to the slope change in the entropy value change curve, and obtaining the dynamic change trend feature of the immune parameter;

[0011] S5: The upper and lower limits of the change range of the key feature points in the dynamic change trend characteristics of the immune parameters are used as the initial fluctuation thresholds, and the initial fluctuation thresholds are matched and adjusted with reference to the change trends of the patient grouping characteristics. The dynamic changes of the patient's immune status are re-evaluated according to the adjustment results to obtain the dynamic evaluation results of lung cancer immunity.

[0012] As a further solution of the present invention, the step of acquiring the abnormal fluctuation section data is specifically as follows:

[0013] S111: Based on the immune index data collected during the patient's immune monitoring process, the acquired data is cleaned and converted into a time series, and divided into multiple segments according to a fixed time window to obtain the divided time series segment data;

[0014] S112: Based on the divided time series segment data, the formula is used:

[0015] ;

[0016] Calculate the The mean of immune index in each segment , integrating the information to obtain the mean of the immune index in each segment;

[0017] in, Indicates In the section immune indicator data points, Indicates The number of data points in a segment;

[0018] S113: Analyze the mean value change between adjacent segments according to the mean value of the immune index of each segment, compare it with the preset change threshold, mark the area where the change amplitude exceeds the change threshold, and generate abnormal fluctuation segment data.

[0019] As a further solution of the present invention, the step of acquiring the abnormal change point distribution data is specifically as follows:

[0020] S211: Based on the abnormal fluctuation section data, the formula is used:

[0021] ;

[0022] Calculate the change range of the mean value of the immune index of two adjacent windows in the sliding window , compare the mean change ranges of all window pairs to obtain the change range comparison results;

[0023] in, Indicates The average value of the immune index in the window, Indicates Next window Window index, It is a weight value set according to the historical immune fluctuation range, abnormal event ratio and individual characteristics of the patient. Indicates The standard deviation of the immune index data within the window;

[0024] S212: Select the largest point in the comparison result of the change amplitude as the abnormal change point, extract the time range of the abnormal change point and the corresponding immune index data, and generate abnormal change point distribution data.

[0025] As a further solution of the present invention, the step of obtaining the immune response patient grouping characteristics is specifically:

[0026] S311: Based on the abnormal change point distribution data, the formula is used:

[0027] ;

[0028] Calculation indicates patient and patients The similarity weight ,According to the similarity weights, a similarity matrix between patients is constructed;

[0029] in, and Indicates patient and patients The statistical eigenvector of Indicates patient and patients The Euclidean distance of the feature vector difference is calculated as: , and Patients and patients In the The value of a statistical feature, represents the dimension of the feature vector, represents the standard deviation of distance scaling;

[0030] S312: Based on the data in the inter-patient similarity matrix, the statistical feature vectors in the matrix are decomposed and analyzed, and the patients are grouped according to the similarity relationship to generate immune response patient grouping characteristics.

[0031] As a further solution of the present invention, the step of obtaining the local entropy value in the corresponding time window is specifically as follows:

[0032] S411: Based on the immune response patient grouping characteristics, the formula is used:

[0033] ;

[0034] Calculate the local entropy value within the corresponding time window ;

[0035] in, Indicates the number of intervals divided in the time window. Indicates the immune index data within the time window The probability of an interval is calculated as: , It is The frequency of immune index data in the interval, Representing probability The base 2 logarithm of ;

[0036] S412: Determine the distribution complexity of the patient's immune index by dividing the local entropy value range within the corresponding time window, and accumulate the local entropy value of each time window to obtain an entropy value change curve.

[0037] As a further solution of the present invention, the step of obtaining the dynamic change trend characteristics of the immune parameters is specifically:

[0038] S421: Based on the entropy value change curve, extract the slope value of each time point from the curve, and detect local extreme value points and mark them as key feature points by analyzing the change trend of the slope to obtain a set of key feature points;

[0039] S422: extracting the time position, slope value and cumulative entropy value information of each key feature point in the key feature point set, sorting the feature point data and establishing a feature point sequence, and obtaining the dynamic change trend characteristics of the immune parameter.

[0040] As a further solution of the present invention, the steps for obtaining the dynamic evaluation results of lung cancer immunity are specifically as follows:

[0041] S511: Based on the key feature points extracted from the dynamic change trend characteristics of the immune parameters, the upper and lower limits of the values ​​of each feature point are recorded according to the change range of each feature point in the time series, the recorded upper and lower limits are set as the initial fluctuation threshold range, and the change data of all immune parameters in the time series are screened and marked point by point to generate the initial fluctuation threshold range;

[0042] S512: Based on the initial fluctuation threshold range, referring to the changing trend of the patient grouping characteristics, the initial fluctuation threshold range and the grouping parameter changing trend range are matched and adjusted, and the dynamic changes of the immune status are re-evaluated by comparing the changing points of the time series data with the adjusted fluctuation threshold range to obtain the dynamic immune evaluation results of lung cancer.

[0043] A lung cancer dynamic assessment system based on immune monitoring, the lung cancer dynamic assessment system based on immune monitoring is used to execute the lung cancer dynamic assessment method based on immune monitoring, the system comprises:

[0044] The data collection and fluctuation marking module collects the immune index data obtained during the patient's immune monitoring process, analyzes the mean changes between adjacent segments of the immune index data time series, marks abnormal fluctuation areas, and generates abnormal fluctuation segment data;

[0045] The abnormal change point analysis module analyzes the abnormal fluctuation section data through a sliding window, selects abnormal change points, and generates abnormal change point distribution data by determining the positions of the abnormal change points;

[0046] The patient grouping feature extraction module extracts the statistical features of the immune index data corresponding to the abnormal change point distribution data, and clusters the immune index data among patients to generate immune response patient grouping features;

[0047] The entropy value analysis module performs a distribution entropy analysis on the time series of the immune index data of each group of patients in the immune response patient grouping characteristics, calculates the local entropy value in the corresponding time window, and accumulates the local entropy value into an entropy value change curve, and marks the key feature points with reference to the slope change in the entropy value change curve, thereby generating a dynamic change trend feature of the immune parameter;

[0048] The dynamic assessment module uses the upper and lower limits of the change range of the key feature points in the dynamic change trend characteristics of the immune parameters as the initial fluctuation thresholds, adjusts the initial fluctuation thresholds according to the change trends of the patient grouping characteristics, re-evaluates the dynamic changes of the patient's immune status according to the adjustment results, and generates dynamic assessment results of lung cancer immunity.

[0049] Compared with the prior art, the advantages and positive effects of the present invention are:

[0050] In the present invention, by dividing the patient's immune monitoring data into time series, and accurately extracting abnormal fluctuation sections and abnormal change points through mean change analysis and sliding window methods, it is possible to capture significant fluctuations in the immune system more quickly and accurately, thereby improving the timeliness of dynamic monitoring. Further combined with statistical feature analysis and similarity matrix calculation, the differentiated grouping of immune indicators between patients is refined, which is helpful for personalized evaluation of patient groups with different immune states. The local entropy value calculation and the cumulative analysis of the change curve can clearly reveal the complexity of the change in immune status, and mark key feature points by slope changes, thereby improving the ability to identify key change areas and avoiding confusion between long-term trends and short-term fluctuations in the data. By dynamically adjusting the matching of the immune parameter fluctuation threshold and the patient grouping characteristics, the evaluation of the immune status is combined with the actual patient characteristics, and a more targeted dynamic change analysis of the immune status is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0052] Figure 2 A flow chart of obtaining abnormal fluctuation section data for the present invention;

[0053] Figure 3 A flow chart for obtaining abnormal change point distribution data of the present invention;

[0054] Figure 4 Flow chart for obtaining immune response patient grouping characteristics for the present invention;

[0055] Figure 5 A flow chart of calculating the local entropy value within the corresponding time window of the present invention;

[0056] Figure 6 A flow chart of the present invention for obtaining the dynamic change trend characteristics of immune parameters;

[0057] Figure 7 The present invention is a flow chart for obtaining dynamic immune assessment results of lung cancer. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0059] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0060] See also Figure 1 The present invention provides a technical solution: a dynamic evaluation method for lung cancer based on immune monitoring, comprising the following steps:

[0061] S1: Collect immune index data obtained during the patient immune monitoring process, convert the immune index data into a time series, divide the time series into multiple segments according to a fixed window, obtain the mean of the immune index in each segment, analyze the mean change between adjacent segments, compare the mean change with the preset change threshold, mark the abnormal fluctuation area according to the comparison result, and generate abnormal fluctuation segment data;

[0062] S2: Analyze the abnormal fluctuation segment data through a sliding window, compare the mean changes of the immune index data in the previous and next windows in each sliding window, select the point with the largest mean change amplitude from the comparison results, regard the point as an abnormal change point, and obtain the abnormal change point distribution data by determining the position of the abnormal change point;

[0063] S3: extracting the statistical features of the immune index data corresponding to the abnormal change point distribution data, constructing a similarity matrix between patients based on the statistical features, decomposing the statistical feature vectors in the similarity matrix between patients, clustering and grouping the immune index data between patients, and generating immune response patient grouping features;

[0064] S4: Performing distribution entropy analysis on the time series of immune index data of each group of patients in the immune response patient grouping characteristics, calculating the local entropy value in the corresponding time window, accumulating the local entropy value into an entropy value change curve, marking the key feature points with reference to the slope change in the entropy value change curve, and obtaining the dynamic change trend characteristics of the immune parameters;

[0065] S5: taking the upper and lower limits of the change range of the key characteristic points in the dynamic change trend characteristics of the immune parameters as the initial fluctuation threshold, adjusting the initial fluctuation threshold according to the change trend of the patient grouping characteristics, re-evaluating the dynamic changes of the patient's immune status according to the adjustment results, and obtaining the dynamic evaluation results of lung cancer immunity;

[0066] The abnormal fluctuation segment data specifically include the area where the mean change of immune indicators is significant, the area where the mean change exceeds the change threshold, and the time segment marked as abnormal. The abnormal change point distribution data specifically include the point position with the largest mean change, the marked abnormal change point time series, and the immune indicator value corresponding to the abnormal change point. The immune response patient grouping characteristics include patient grouping results, patient grouping feature vectors, and consistency of patient immune indicator change trends. The dynamic change trend characteristics of immune parameters specifically include the local entropy value within the time window, the key feature points of the entropy change curve, and the entropy slope change area. The dynamic immune assessment results of lung cancer include the dynamic changes of immune status, the adjusted fluctuation threshold, and the dynamic change analysis results of patient groups.

[0067] See also Figure 2 , the specific steps for obtaining abnormal fluctuation segment data are:

[0068] S111: Based on the immune index data collected during the patient's immune monitoring process, the acquired data is cleaned and converted into a time series, and divided into multiple segments according to a fixed time window to obtain the divided time series segment data;

[0069] The immune index data of patients are collected through blood analyzers, cell counters and other equipment, including white blood cell counts, T cell ratios, C-reactive protein concentrations, etc. The acquired immune index data are processed using data cleaning methods. The integrity and consistency of the data are ensured by detecting and eliminating null values, duplicate values ​​and data points that do not meet the standard range. The cleaned immune index data are converted into time series data according to the timestamp information, and the time series data are divided according to fixed time windows. The length of each window is set to 1 hour. The immune index data in each window is independently marked and divided into segments to generate segment data of the time series.

[0070] S112: Based on the divided time series segment data, the formula is used:

[0071] ;

[0072] Calculate the The mean of immune index in each segment ,Immune indicators may include white blood cell count, T cell ratio or C-reactive protein concentration. Specific immune indicators are directly obtained through experimental detection instruments (such as blood analyzers), and the information is integrated to obtain the mean value of immune indicators in each segment;

[0073] in, Indicates In the segment, from the first data point to the The cumulative result of the data points is Indicates In the section An immune indicator data point is a single observation value of an immune indicator, such as the observation result of each white blood cell count. This data is directly obtained through experimental data collected during the immune monitoring process, such as using a blood analyzer to collect the patient's white blood cell count value. Indicates The number of data points in a segment, that is, the number of sampling times for immune monitoring in the segment. For example, if the white blood cell count is tested every hour in a day, the number of data points in 5 hours is , sampling times It can be determined by dividing the time series into segments.

[0074] Taking the patient's white blood cell count as an example, if the 5 white blood cell count values ​​collected in a certain time series segment are , , , , (Unit: 10^9 / L), the number of data points in the segment is The calculation steps are as follows:

[0075] Add up all the data points: ;

[0076] ;

[0077] Calculate the mean: ;

[0078] This result shows that the current The mean white blood cell count in each segment was 6.28 (unit: 10^9 / L).

[0079] S113: Analyze the mean change between adjacent segments according to the mean value of the immune index of each segment, compare it with the preset change threshold, mark the area where the change amplitude exceeds the change threshold, and generate abnormal fluctuation segment data;

[0080] Based on the mean of each segment’s immune index, for example Mean white blood cell count for the segment and Mean white blood cell count for the segment , calculate the absolute difference of the mean change of adjacent segments, the formula is: , Indicates The mean of the immune index of each segment is calculated, and the change range of the mean is calculated. With preset change threshold To compare, for example, to set a threshold .because , so mark the and The time range between the segments is the abnormal fluctuation area, and the white blood cell count data within this time period are extracted for subsequent analysis. In addition, combined with other immune indicators, such as T cell ratio and C-reactive protein concentration, it is assumed that the corresponding Section and The segment means are: T cell proportion: , , mean change . C-reactive protein concentration: , , mean change These changes are compared with the thresholds of the corresponding indicators. For example, the threshold for the change in the proportion of T cells is , the threshold of change in C-reactive protein concentration is For the T cell ratio, , marked as abnormal fluctuation. For C-reactive protein concentration, , marked as abnormal fluctuations. Finally, the abnormal fluctuation areas marked in all immune indicators are extracted, their time ranges and corresponding immune indicator data are integrated, and abnormal fluctuation segment data are generated, providing a basis for subsequent abnormal change point analysis.

[0081] See also Figure 3 , the specific steps for obtaining the abnormal change point distribution data are:

[0082] S211: Based on the abnormal fluctuation segment data, the formula is used:

[0083] ;

[0084] Calculate the change range of the mean value of the immune index of two adjacent windows in the sliding window , compare the mean change ranges of all window pairs to obtain the change range comparison results;

[0085] in, Indicates The average value of the immune index in the window, Indicates Next window Window index, It is a weight value set according to the historical immune fluctuation range, abnormal event ratio and individual characteristics of the patient. Indicates The standard deviation of the immune indicator data within the window is used to normalize the fluctuation characteristics of the mean change amplitude. The calculation formula is: , It is The first immune indicator data points, such as white blood cell counts, Indicates The total number of data points within the window, represents the square of the deviation of the data point from the mean, It represents the cumulative value of the square deviation of all data points in the window. The standard deviation data is used to measure the volatility within the window. It is obtained by analyzing the deviation range of the data within the window. For example, the regular fluctuation range of white blood cell count is provided by historical health data. Represents the absolute value of the mean change difference, multiplied by the weight parameter To adjust the impact of the change, the absolute value operation eliminates the interference of positive and negative signs, and the weight parameter The setting can be determined by combining the immune fluctuation range in the patient's historical immune monitoring data and the characteristics of the specific time period. For example, in a patient, during the 7-day continuous monitoring, the fluctuation range of the white blood cell count was counted and it was found that the fluctuation range of the immune index at night (22:00-06:00) was concentrated in Unit (10^9 / L), the fluctuation range of immune indicators during the day (06:00-22:00) is concentrated in Unit (10^9 / L). Combined with the preset range, it is determined that the fluctuation range of night data is greater than that of daytime data (specifically, the average fluctuation value at night is 2.2, and the average fluctuation value during the day is 1.0). At the same time, through abnormal event analysis, the proportion of abnormal fluctuations in white blood cell counts during the night time period is 60%, while the proportion of abnormal fluctuations during the day time period is 30%. Based on the above fluctuation range and event ratio, a higher weight value can be set for night data, for example , and set a lower weight value for daytime data, such as In addition, in the immune monitoring during the treatment phase, if the patient's abnormal fluctuation frequency increases by 20% overall (for example, from 60% at night to 72%, and from 30% during the day to 36%), the weight values ​​of all time periods need to be dynamically adjusted based on this increase. For example, the night weight increases from Increase to , daytime weight from Increase to This weight setting process combines the immune fluctuation range, event proportion and treatment dynamic adjustment to ensure that the weight value is reasonable and consistent with the patient's immune status characteristics.

[0086] Taking the white blood cell count (unit: 10^9 / L) as an example, the sliding window length is set to 1 hour. The data points in the window are , No. The data points in the window are The calculation process is as follows: Window mean and standard deviation:

[0087] ;

[0088] ;

[0089] Calculate the Window mean and weights:

[0090] ;

[0091] Set weight , through historical monitoring data, it is determined that the patient's nighttime immunity fluctuates greatly, and the normalized mean change range is calculated: ;

[0092] The results show that the normalized mean change is , by introducing the weight parameter and standard deviation adjustment parameters , can more flexibly reflect the impact of different windows on mean changes, and use standard deviation parameters to normalize data fluctuations, reduce the interference of extreme values ​​on anomaly detection, and improve the accuracy and sensitivity of calculations. By comparing the mean change of immune indicator data in the front and back sliding windows, the change characteristics of immune indicators between each pair of adjacent windows can be clarified. For example, by calculating the first and Normalized mean change of the window , it is found that it exceeds the preset abnormal fluctuation threshold (for example, 3.5), indicating that there is a significant fluctuation in the immune index change between this pair of windows. At the same time, for the mean change of other window pairs, the same method can be used to calculate the normalized change amplitude in turn and compare it with the threshold, and the window pairs with significant changes are marked as abnormal fluctuation areas, which is convenient for further analysis of the dynamic characteristics of their immune indicators.

[0093] S212: Select the point with the largest change amplitude comparison result as the abnormal change point, extract the time range of the abnormal change point and the corresponding immune index data, and generate abnormal change point distribution data;

[0094] By comparing the change amplitudes of the normalized mean values ​​of all sliding window pairs, the point with the largest change amplitude is selected as the abnormal change point. , No. , No. , No. The normalized mean changes of the window pairs are: and Window Pairs: , No. and Window Pairs: , No. and Window Pairs: In the above comparison results, it is found that is the maximum value of the change in the normalized mean value of all windows, so we select and The mean change of the window is the abnormal change point. Record the location of the abnormal change point and extract the immune index data corresponding to the window Finally, all marked abnormal change points are integrated into abnormal change point distribution data to form abnormal change point distribution data.

[0095] See also Figure 4 , the specific steps for obtaining the immune response patient grouping characteristics are:

[0096] S311: Based on the abnormal change point distribution data, the formula is used:

[0097] ;

[0098] Calculation indicates patient and patients The similarity weight , the range is The larger the value, the higher the similarity between patients. According to the similarity weight, the similarity matrix between patients is constructed;

[0099] in, and Indicates patient and patients Each vector contains the statistical characteristics of multiple immune indicators, such as: mean value: reflects the overall level of immune indicators within a given time range, such as the mean of white blood cell count; variance: measures the volatility of immune indicators within a time range, such as the fluctuation range of patients' white blood cell counts over time; rate of change: measures the dynamic change amplitude of immune indicators in time series, such as the hourly change trend of white blood cell counts. The hourly data of white blood cell counts are collected by a blood analyzer and its statistical characteristics are calculated. Indicates patient and patients The Euclidean distance of the feature vector difference is used to measure the statistical feature difference between two patients. The calculation formula is: , and Patients and patients In the The value of a statistical feature, Represents the dimension of the feature vector. For example, for three statistical features, , ' represents the standard deviation of distance scaling, controlling the influence of Euclidean distance on similarity, standard deviation The value of is obtained by analyzing the overall heterogeneity of the patient population, for example, setting It is the standard deviation of the statistical characteristics of the group of patients, determined through group-wide analysis combined with the dynamic changes of immune indicators over time.

[0100] If the patient and patients The statistical eigenvectors of are: , where the first value is the average white blood cell count (unit: ), the second value is the variance of the white blood cell count, and the third value is the rate of change of the white blood cell count (unit: Hourly). Calculate the Euclidean distance:

[0101] ;

[0102] Calculating similarity weights ,set up , according to the formula:

[0103] ;

[0104] The results showed that patients and patients The similarity weight is . Indicates that the patient and patients The similarity in statistical features is high but not completely consistent. The range of similarity weight is , among which, when When , it means that the statistical characteristics between patients are highly similar, indicating that the change trends of the immune indicators of the two are highly consistent; when When , it means that the similarity of statistical characteristics between patients is low, indicating that there are significant differences in the changing trends of immune indicators between the two groups; when In When the difference is between 0 and 1, it means that the similarity of statistical characteristics between patients is moderate, there is partial consistency but the difference is still significant.

[0105] The process of constructing the similarity matrix is ​​as follows: First, the similarity weights between each pair of patients are calculated. For example, for four patients , respectively calculate , and the following weight values ​​are obtained: , , , , , Next, these weight values ​​are arranged into a symmetric matrix according to the order of the patients : , the diagonal elements of the matrix are 0, indicating that the similarity between each patient and itself does not participate in the weight calculation. The similarity matrix constructed It directly expresses the statistical similarity relationship between each pair of patients and provides a basis for the subsequent analysis of immune response patient groups. In the matrix, the patient pairs with higher similarity weight values ​​indicate that they are more consistent in the trend of changes in immune indicators. For example, Indicates patient and patients The similarity is higher, while Indicates patient and patients The similarity is low.

[0106] S312: Based on the data in the similarity matrix between patients, the statistical feature vectors in the matrix are decomposed and analyzed, and the patients are grouped according to the similarity relationship to generate immune response patient grouping features;

[0107] Based on the similarity matrix constructed , first of all, the matrix Perform eigenvalue decomposition and extract the main statistical eigenvectors for clustering of patient immune indicators. Perform eigenvalue decomposition and calculate the eigenvalue and eigenvector. The eigenvalue is , the corresponding eigenvector is: , , , , select the maximum eigenvalue The corresponding eigenvector , and use it as the main basis for clustering. Map the elements of the feature vector to a low-dimensional space for grouping analysis. For example, for the feature vector , we can divide patients into different groups according to the value of the eigenvector: the eigenvalues ​​of the first and second patients are close (0.5 and 0.5), so they can be divided into the first group; the eigenvalue of the third patient is 0.6, which is slightly different from the first two, but still belongs to the same trend, so he is divided into the first group; the eigenvalue of the fourth patient is 0.1, which is quite different from the other patients, so he is divided into the second group alone. -Mean clustering method was used to further refine the patient grouping. , input the eigenvector value into - Mean algorithm, clustering results are: Group 1 patients (similar immune response trends): patients ; Group 2 patients (significantly different immune response trends): patients According to the clustering results, the immune response patient grouping characteristics are generated, and the number of each group of patients is bound to its corresponding immune index characteristics. For example, the immune index of the first group of patients shows a higher trend of change, the mean of the white blood cell count is higher and the fluctuation range is larger; the immune index of the second group of patients shows a lower trend of change, and the immune system performance is relatively stable.

[0108] See also Figure 5 , the specific steps for calculating the local entropy value in the corresponding time window are:

[0109] S411: Based on the characteristics of the immune response patient groups, the formula is used:

[0110] ;

[0111] Calculate the local entropy value within the corresponding time window ;

[0112] in, Indicates the number of intervals divided in a time window, for example, a time window is divided into The specific number of intervals is set based on the time series data of immune indicators and the density of sampling frequency. The number of intervals is determined through experimental design to make the distribution analysis more statistically significant. Indicates the immune index data within the time window The probability of an interval is calculated as: , It is The frequency of immune index data within an interval, the probability distribution is obtained by statistically analyzing the data distribution within the time window. For example, the hourly data of white blood cell count can be collected by the instrument and the interval frequency statistics can be performed. Representing probability The base-2 logarithm of the information entropy is used to measure the "information content" of a single interval, and the value range is determined by the probability The distribution of Express The probability value of the interval Perform the accumulation operation, where the summation symbol is associated It is the index of each interval distributed in the time window. The sum of the probabilities of all intervals is 1, which represents the complete probability distribution.

[0113] Assume that within a time window, the interval of immune index data is divided into intervals, and the frequencies of data points in each interval are . Calculate the probability distribution of each interval: , , calculate the probability The logarithm of and its product :

[0114] ;

[0115] ;

[0116] ;

[0117] ;

[0118] ;

[0119] Summarize the amount of information in each interval and calculate the total entropy value: ;

[0120] This result shows that the local entropy It reflects the complexity of the distribution of immune index data within the time window. Through the pre-set entropy value range, the entropy value can be divided into the following three situations: : Indicates that the distribution of immune index data within this time window is relatively concentrated and the volatility is low, indicating that the change trend of the patient's immune index is relatively stable. : Indicates that the distribution complexity of the immune index data within this time window is moderate, with a certain degree of volatility but not reaching a significant level. : Indicates that the immune index data distribution complexity in this time window is high, and there may be significant volatility or abnormal changes, indicating the possibility of abnormal fluctuations in the patient's immune index. Current calculation results Located in the range This indicates that the distribution complexity of immune indicators in this time window is moderate, with certain volatility, but it has not reached the level of abnormal or significant fluctuation.

[0121] S412: Determine the distribution complexity of the patient's immune index by dividing the local entropy value range within the corresponding time window, and accumulate the local entropy value of each time window to obtain an entropy value change curve;

[0122] Input the local entropy value data in each time window into the data processing software (such as MATLAB or Python's NumPy library), and perform accumulation operations on the local entropy values ​​in chronological order. For example, suppose the local entropy values ​​in 5 time windows are The accumulation process is as follows: The accumulation value at the first time point is , the cumulative value at the second time point is , the cumulative value at the third time point is , the cumulative value at the fourth time point is , the cumulative value at the fifth time point is The final entropy value change curve is: ,Data processing software can be used for fast cumulative calculation and visualization to generate curve graphs. For example, using the "cumsum" function of MATLAB or the "cumsum" method of the NumPy library in Python, the local entropy value array can be input into the function to calculate the cumulative value. The drawn curve reflects the cumulative change of entropy value over time.

[0123] See also Figure 6 ,The specific steps for obtaining the dynamic change trend characteristics of immune parameters are:

[0124] S421: based on the entropy value change curve, extract the slope value of each time point from the curve, and by analyzing the change trend of the slope, detect the local extreme value points and mark them as key feature points to obtain a set of key feature points;

[0125] The key feature points are identified by analyzing the slope changes in the curve. The specific process is as follows: first, the slope value corresponding to each time point is extracted from the entropy cumulative change curve, and the slope is calculated using the numerical difference method. For example, for the entropy change curve , the slope is calculated as follows: the first time point has no previous value, and the slope is defined as the null value; the slope at the second time point is ; The slope at the third time point is ; The slope at the fourth time point is ; The slope at the fifth time point is .in, : represents the time point index in the time series, in units of time (such as seconds, minutes, hours, etc.), and the specific unit depends on the sampling frequency of the immune monitoring data. For example, if data is collected 10 times in one minute, the time point index Respectively represent the time position every 6 seconds. : represents the time interval, the unit is consistent with the time point, and the specific time interval can be directly obtained through the sampling frequency. For example, if the sampling frequency is 1 second, then seconds; if the sampling frequency is 10 seconds, then Then, according to the slope change trend, detect its local extreme point, for example, the slope at the third time point is the local maximum , are marked as key feature points, and the marked feature points are bound to the positions in the time series to eventually form a set of key feature points.

[0126] S422: extracting the time position, slope value and cumulative entropy value information of each key feature point in the key feature point set, sorting the feature point data and establishing a feature point sequence, and obtaining the dynamic change trend characteristics of the immune parameter;

[0127] First, the information of each key feature point is extracted and classified. For example, for the key feature point at the third time point, its corresponding time position, slope value and cumulative entropy value are extracted and recorded as a feature point set. , sort the data of all feature points and establish a feature point sequence to obtain the dynamic change trend characteristics of the immune parameters.

[0128] See also Figure 7 The specific steps for obtaining the dynamic evaluation results of lung cancer immunity are as follows:

[0129] S511: Based on the key feature points extracted from the dynamic change trend characteristics of the immune parameters, the upper and lower limits of the values ​​of each feature point are recorded according to the change range of each feature point in the time series, the recorded upper and lower limits are set as the initial fluctuation threshold range, and the change data of all immune parameters in the time series are screened and marked point by point to generate the initial fluctuation threshold range;

[0130] Record the upper and lower limits of each feature point according to its range of variation in the time series. For example, by analyzing the key feature points in the time series , determine the upper and lower limits of its variation range to be 2.1 and 5.0 respectively, and set this range as the initial fluctuation threshold range ,Then the change data of all immune parameters in the time series are screened point by point, and only the data points that fall within the initial fluctuation threshold range are retained. During the screening process, the time position of each data point needs to be marked, and the immune parameter type and its corresponding trend characteristics of each data point are recorded at the same time, and the association between the key feature points and the initial fluctuation threshold range is completed, providing the initial fluctuation threshold range for the subsequent adjustment steps.

[0131] S512: Based on the initial fluctuation threshold range, referring to the change trend of the patient grouping characteristics, the initial fluctuation threshold range is matched and adjusted with the change trend range of the grouping parameter, and the dynamic change of the immune status is re-evaluated by comparing the change points of the time series data with the adjusted fluctuation threshold range to obtain the dynamic evaluation result of lung cancer immunity;

[0132] The obtained initial fluctuation threshold range is matched and adjusted. First, the change trend range of immune parameters in the patient grouping characteristics is extracted. For example, the change trend range of the parameter in group 1 is , the parameter variation trend range of group 2 is , the initial fluctuation threshold range Match it with the parameter change trend range of each group, and determine the new adjusted fluctuation threshold range by calculating the intersection of the initial fluctuation threshold range and the group parameter change trend range. Then, the dynamic changes of the patient's immune status are re-evaluated according to the adjusted fluctuation threshold range. The specific operation is to compare the change points of all immune parameters in the time series with the adjusted fluctuation threshold range one by one, and mark the data points that fall within the adjusted fluctuation threshold range as normal fluctuations, and mark the data points that exceed the adjusted fluctuation threshold range as abnormal fluctuations. Finally, the dynamic evaluation results of lung cancer immunity are obtained through the adjusted marked data.

[0133] A dynamic evaluation system for lung cancer based on immune monitoring, which is used to execute the above-mentioned dynamic evaluation method for lung cancer based on immune monitoring, comprises:

[0134] The data collection and fluctuation marking module collects the immune index data obtained during the patient's immune monitoring process, analyzes the mean changes between adjacent segments of the immune index data time series, marks abnormal fluctuation areas, and generates abnormal fluctuation segment data;

[0135] The abnormal change point analysis module analyzes the abnormal fluctuation segment data through a sliding window, selects the abnormal change point, and generates the abnormal change point distribution data by determining the location of the abnormal change point;

[0136] The patient grouping feature extraction module extracts the statistical features of the immune index data corresponding to the abnormal change point distribution data, and clusters the immune index data among patients to generate immune response patient grouping features;

[0137] The entropy analysis module performs distribution entropy analysis on the time series of immune index data of each group of patients in the immune response patient grouping characteristics, calculates the local entropy value in the corresponding time window, accumulates the local entropy value into an entropy change curve, marks the key feature points with reference to the slope change in the entropy change curve, and generates the dynamic change trend characteristics of immune parameters;

[0138] The dynamic assessment module uses the upper and lower limits of the change range of key feature points in the dynamic change trend characteristics of immune parameters as the initial fluctuation threshold, and adjusts the initial fluctuation threshold according to the change trend of patient grouping characteristics. According to the adjustment results, the dynamic changes of the patient's immune status are re-evaluated to generate dynamic assessment results of lung cancer immunity.

[0139] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A dynamic assessment method for lung cancer based on immune monitoring, characterized in that: The following steps are involved: S1: Collect immune index data obtained during the patient immune monitoring process, analyze the mean changes between adjacent segments of the immune index data time series, mark abnormal fluctuation areas, and generate abnormal fluctuation segment data; S2: analyzing the abnormal fluctuation section data through a sliding window, selecting abnormal change points, and obtaining abnormal change point distribution data by determining the positions of the abnormal change points; S3: extracting statistical features of the immune index data corresponding to the abnormal change point distribution data, and clustering the immune index data among patients to generate immune response patient grouping features; S4: performing a distribution entropy analysis on the time series of the immune index data of each group of patients in the immune response patient grouping feature, calculating the local entropy value in the corresponding time window, accumulating the local entropy value into an entropy value change curve, marking the key feature points with reference to the slope change in the entropy value change curve, and obtaining the dynamic change trend feature of the immune parameter; S5: taking the upper and lower limits of the change range of the key characteristic points in the dynamic change trend characteristics of the immune parameters as the initial fluctuation threshold, matching and adjusting the initial fluctuation threshold with reference to the change trend of the patient grouping characteristics, and re-evaluating the dynamic changes of the patient's immune status according to the adjustment results to obtain the dynamic evaluation results of lung cancer immunity; The steps for obtaining the lung cancer immune dynamics assessment results are specifically as follows: S511: Based on the key feature points extracted from the dynamic change trend characteristics of the immune parameters, the upper and lower limits of the values ​​of each feature point are recorded according to the change range of each feature point in the time series, the recorded upper and lower limits are set as the initial fluctuation threshold range, and the change data of all immune parameters in the time series are screened and marked point by point to generate the initial fluctuation threshold range; S512: Based on the initial fluctuation threshold range, referring to the changing trend of the patient grouping characteristics, the initial fluctuation threshold range and the grouping parameter changing trend range are matched and adjusted, and the dynamic changes of the immune status are re-evaluated by comparing the changing points of the time series data with the adjusted fluctuation threshold range to obtain the dynamic immune evaluation results of lung cancer.

2. The dynamic assessment method for lung cancer based on immune monitoring according to claim 1, characterized in that: The steps for obtaining the abnormal fluctuation section data are specifically as follows: S111: Based on the immune index data collected during the patient's immune monitoring process, the acquired data is cleaned and converted into a time series, and divided into multiple segments according to a fixed time window to obtain the divided time series segment data; S112: Based on the divided time series segment data, the formula is used: ; Calculate the The mean of immune index in each segment , integrating the information to obtain the mean value of the immune index in each segment; in, Indicates In the section immune indicator data points, Indicates The number of data points in a segment; S113: Analyze the mean value change between adjacent segments according to the mean value of the immune index of each segment, compare it with the preset change threshold, mark the area where the change amplitude exceeds the change threshold, and generate abnormal fluctuation segment data.

3. The dynamic assessment method for lung cancer based on immune monitoring according to claim 2, characterized in that: The steps for obtaining the abnormal change point distribution data are specifically as follows: S211: Based on the abnormal fluctuation section data, the formula is used: ; Calculate the change range of the mean value of the immune index of two adjacent windows in the sliding window , compare the mean change ranges of all window pairs to obtain the change range comparison results; in, Indicates The average value of the immune index in the window, Indicates Next window Window index, It is a weight value set according to the historical immune fluctuation range, abnormal event ratio and individual characteristics of the patient. Indicates The standard deviation of the immune index data within the window; S212: Select the largest point in the comparison result of the change amplitude as the abnormal change point, extract the time range of the abnormal change point and the corresponding immune index data, and generate abnormal change point distribution data.

4. The dynamic assessment method for lung cancer based on immune monitoring according to claim 3, characterized in that: The steps for obtaining the immune response patient grouping characteristics are specifically as follows: S311: Based on the abnormal change point distribution data, the formula is used: ; Calculation indicates patient and patients The similarity weight ,According to the similarity weights, a similarity matrix between patients is constructed; in, and Indicates patient and patients The statistical eigenvector of Indicates patient and patients The Euclidean distance of the feature vector difference is calculated as: , and Patients and patients In the The value of a statistical feature, represents the dimension of the feature vector, ' represents the standard deviation of distance scaling; S312: Based on the data in the inter-patient similarity matrix, the statistical feature vectors in the matrix are decomposed and analyzed, and the patients are grouped according to the similarity relationship to generate immune response patient grouping characteristics.

5. The method for dynamic assessment of lung cancer based on immune monitoring according to claim 4, characterized in that: The step of obtaining the local entropy value in the corresponding time window is specifically as follows: S411: Based on the immune response patient grouping characteristics, the formula is used: ; Calculate the local entropy value within the corresponding time window ; in, Indicates the number of intervals divided in the time window. Indicates the immune index data within the time window The probability of an interval is calculated as: , It is The frequency of immune index data in the interval, Representing probability The base 2 logarithm of ; S412: Determine the distribution complexity of the patient's immune index by dividing the local entropy value range within the corresponding time window, and accumulate the local entropy value of each time window to obtain an entropy value change curve.

6. The method for dynamic assessment of lung cancer based on immune monitoring according to claim 5, characterized in that: The steps for obtaining the dynamic change trend characteristics of the immune parameters are specifically as follows: S421: Based on the entropy value change curve, extract the slope value of each time point from the curve, and detect local extreme value points and mark them as key feature points by analyzing the change trend of the slope to obtain a set of key feature points; S422: extracting the time position, slope value and cumulative entropy value information of each key feature point in the key feature point set, sorting the feature point data and establishing a feature point sequence, and obtaining the dynamic change trend characteristics of the immune parameter.

7. A dynamic evaluation system for lung cancer based on immune monitoring, characterized in that: According to any one of claims 1 to 6, the method for dynamic assessment of lung cancer based on immune monitoring comprises: The data collection and fluctuation marking module collects the immune index data obtained during the patient's immune monitoring process, analyzes the mean changes between adjacent segments of the immune index data time series, marks abnormal fluctuation areas, and generates abnormal fluctuation segment data; The abnormal change point analysis module analyzes the abnormal fluctuation section data through a sliding window, selects abnormal change points, and generates abnormal change point distribution data by determining the positions of the abnormal change points; The patient grouping feature extraction module extracts the statistical features of the immune index data corresponding to the abnormal change point distribution data, and clusters the immune index data among patients to generate immune response patient grouping features; The entropy value analysis module performs a distribution entropy analysis on the time series of the immune index data of each group of patients in the immune response patient grouping characteristics, calculates the local entropy value in the corresponding time window, and accumulates the local entropy value into an entropy value change curve, and marks the key feature points with reference to the slope change in the entropy value change curve, thereby generating a dynamic change trend feature of the immune parameter; The dynamic assessment module uses the upper and lower limits of the change range of the key feature points in the dynamic change trend characteristics of the immune parameters as the initial fluctuation thresholds, adjusts the initial fluctuation thresholds according to the change trends of the patient grouping characteristics, re-evaluates the dynamic changes of the patient's immune status according to the adjustment results, and generates dynamic assessment results of lung cancer immunity.

Images