Method for analyzing immune cell subpopulations in clinical test samples for infectious diseases
By adaptively adjusting the neighborhood size and density threshold and combining infection status information for clustering, the problem of inaccurate identification of rare immune cell subsets in existing technologies has been solved, achieving accurate immune cell subset analysis and improving the diagnostic and treatment monitoring capabilities for infectious diseases.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 南昌大学第一附属医院
- Filing Date
- 2026-03-18
- Publication Date
- 2026-06-09
Smart Images
Figure CN122177244A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of bioinformatics technology, specifically to a method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases. Background Technology
[0002] Infectious diseases are caused by pathogens such as bacteria and viruses. Their pathogenesis is closely related to the body's immune system response, making accurate monitoring of immune cell status crucial for disease diagnosis and treatment. In responding to infectious diseases, the body's immune system participates in infection response and immune regulation through different types of immune cell subsets. Different types of immune cells exhibit specific proportional changes, activation states, and functional transformations during the development stages of infectious diseases. Therefore, immune cell subset analysis has become an important direction for the clinical diagnosis, prognostic assessment, and efficacy monitoring of infectious diseases.
[0003] Traditional flow cytometry is typically limited to 3 to 5 colors, primarily identifying basic lymphocyte subsets. However, the development of multicolor flow cytometry has significantly increased the number of detection parameters, with advanced schemes employing 13, 18, or even 41 colors. These high-dimensional analytical methods allow for the design of biomarker combinations that can simultaneously resolve multiple immune cell subsets and their functional states.
[0004] In low-dimensional flow cytometry, gating provides a relatively intuitive definition of cell subpopulations. However, in high-dimensional space, it is difficult to accurately and repeatedly delineate a specific, rare immune cell subpopulation, requiring automated clustering algorithms for data analysis. In infectious diseases, however, key immune responses often manifest in rare cell subpopulations, such as antigen-specific T cells, unconventional T cells, and monocytes in specific functional states. Different algorithms, or even different parameters of the same algorithm, exhibit varying sensitivities to capturing rare populations. Existing automated clustering algorithms struggle to accurately capture this characteristic, resulting in insensitivity to rare subpopulations and consequently, low accuracy in cell classification. Summary of the Invention
[0005] To address the issue of low accuracy in cell classification using existing methods, this invention aims to provide a method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases. The specific technical solution employed is as follows: This invention provides a method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases, comprising the following steps: Obtain blood samples and their clinical infection status information; preprocess the blood samples to obtain several cells; and obtain the fluorescence intensity value of each cell. Based on the distribution of fluorescence intensity values of each cell, the neighborhood size of each cell is adaptively determined; and the basic density index of each cell is determined based on the neighborhood size of each cell. By combining the fluorescence intensity values of each cell, the corresponding core markers, auxiliary markers, clinical infection status information, and the aforementioned baseline density index, the local weighted density of each cell is obtained. A density threshold is determined based on the local weighted density of each cell; all cells are clustered based on the differences in fluorescence intensity values of different cells and the density threshold to obtain cell clusters.
[0006] Preferably, the adaptive determination of the neighborhood size of each cell based on the distribution of fluorescence intensity values of each cell includes: Based on the fluorescence intensity values of all detection channels of each cell, a high-dimensional single-cell expression matrix is constructed; the UMAP algorithm is used to reduce the dimensionality of the high-dimensional single-cell expression matrix. In the low-dimensional space, the Euclidean distance from each cell to its k-th nearest neighbor cell is obtained and recorded as the first distance for each cell, where k is a preset initial threshold. Obtain the median of the first distances corresponding to all cells; adjust the preset initial threshold using the ratio between the median and the first distances corresponding to each cell to obtain the neighborhood size of each cell.
[0007] Preferably, the step of adjusting the preset initial threshold using the ratio between the median and the first distance corresponding to each cell to obtain the neighborhood size of each cell includes: The neighborhood size of each cell is determined by rounding the product of the ratio and the preset initial threshold.
[0008] Preferably, determining the basic density index of each cell based on the neighborhood size of each cell includes: obtaining the number of cells in the neighborhood of each cell as the basic density index of each cell.
[0009] Preferably, the local weighted density of each cell is obtained by combining the fluorescence intensity value of each cell, the corresponding core marker, auxiliary marker, clinical infection status information, and the baseline density index, including: Infection status regulatory factors were obtained by combining C-reactive protein, procalcitonin, and pathogen type in blood samples; the clinical infection status information included C-reactive protein, procalcitonin, and pathogen type. The first average value of all normalized fluorescence intensity values corresponding to the core markers of the cells to be analyzed, and the second average value of all normalized fluorescence intensity values corresponding to the auxiliary markers of the cells to be analyzed are obtained respectively. The phenotypic pathological association score of the cells to be analyzed is obtained by weighted summation of the first average and the second average; wherein, when performing weighted summation, the weight of the first average is greater than the weight of the second average. By combining the baseline density index of the cells to be analyzed, the infection status regulator, and the phenotypic pathology correlation score, the local weighted density of the cells to be analyzed is obtained. The cell to be analyzed can be any cell.
[0010] Preferably, the infection status regulatory factors obtained from the combined blood sample's C-reactive protein, procalcitonin, and pathogen type include: The average of the normalized values of C-reactive protein concentration, procalcitonin concentration, and pathogen type parameter in blood samples was determined as the regulatory factor of infection status. If the pathogen type of the blood sample is a virus, then the pathogen type parameter is set to a preset first value; if the pathogen type of the blood sample is bacteria, then the pathogen type parameter is set to a preset second value; if the pathogen type of the blood sample is a parasite, then the pathogen type parameter is set to a preset third value. The preset first value, preset second value, and preset third value all have a range of (0, 1), and the preset first value is less than the preset second value, and the preset second value is less than the preset third value.
[0011] Preferably, the step of combining the baseline density index of the cells to be analyzed, the infection status regulator, and the phenotypic pathology correlation score to obtain the local weighted density of the cells to be analyzed includes: Calculate the first product of the association scores between the infection status regulator and the phenotypic pathology of the cells to be analyzed; calculate the sum of the first product and the constant 1; The product of the sum and the baseline density index of the cell to be analyzed is determined as the local weighted density of the cell to be analyzed.
[0012] Preferably, determining the density threshold based on the local weighted density of each cell includes: Obtain the median of the local weighted density of all cells; Cells with a local weighted density less than the median are selected as candidate cells; All candidate cells are sorted in descending order of local weighted density. The difference between the local weighted densities of two adjacent candidate cells is calculated. The local weighted density with the largest local weighted density in the two candidate cells corresponding to the two cells with the largest difference is selected as the basic threshold. The density threshold is obtained based on the base threshold and the infection status adjustment factor.
[0013] Preferably, obtaining the density threshold based on the base threshold and the infection status adjustment factor includes: Calculate the first difference between constant 1 and the infection status regulation factor; The product of the first difference and the base threshold is determined as the density threshold.
[0014] Preferably, the step of clustering all cells based on the differences in fluorescence intensity values of different cells and the density threshold to obtain cell clusters includes: In a low-dimensional space, the similarity between pairs of cells is evaluated based on the variance of the Euclidean distance between each pair of cells and the Euclidean distance between all cells; the Euclidean distance is used to characterize the difference between the fluorescence intensity values of the cells. Based on the relationship between the similarity and the density threshold, the DBSCAN algorithm is used to cluster all cells to obtain multiple cell clusters.
[0015] The present invention has at least the following beneficial effects: This invention first adaptively adjusts the neighborhood size of each cell based on the distribution of fluorescence intensity values in blood samples. It then determines the baseline density index of each cell based on this neighborhood information, capturing finer internal structures while avoiding excessive noise. Based on infection status perception features, it adaptively determines a density threshold using the baseline density index and performs adaptive clustering of all cells using the neighborhood information. Finally, it integrates the clustering results to analyze immune cell subpopulations in clinical laboratory samples from infectious diseases. This method accurately adapts to the dynamic characteristics of infectious disease immunophenotypes, adaptively determining the neighborhood size through infection status-driven parameters. It addresses the problems of poor adaptation to density heterogeneity and missed detection of rare subpopulations in general algorithms, achieving accurate identification of finely segmented subpopulations. This improves the accuracy and clinical relevance of the analysis results, providing a reliable basis for infection type identification, disease assessment, and treatment monitoring. Attached Figure Description
[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 A flowchart illustrating a method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases, provided in an embodiment of the present invention; Figure 2 This is a two-parameter scatter plot of flow cytometry provided in an embodiment of the present invention; Figure 3 This is a structural block diagram of an immune cell subset analysis system for clinical laboratory samples for infectious diseases, provided in an embodiment of the present invention. Detailed Implementation
[0018] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the following detailed description, in conjunction with the accompanying drawings and preferred embodiments, describes a method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases based on the present invention.
[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0020] The following description, in conjunction with the accompanying drawings, details the specific scheme of the method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases provided by this invention.
[0021] Example of an embodiment of a method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases: The specific scenario addressed in this embodiment is as follows: When analyzing immune cell subsets in clinical laboratory samples from infectious diseases, pathogens such as bacteria and viruses directly attack immune cells, leading to apoptosis and increased cell fragmentation. Furthermore, the type of anticoagulant and storage time after sample collection further exacerbate cell damage. These physiological changes introduce significant data noise. The physiological and pathological characteristics of clinical samples determine the sensitivity of different cell features to disease progression; these features are covariates in subsequent clinical weighting. Therefore, targeted preprocessing is necessary to first reduce damage at the sample level, then accurately classify cells in blood samples, grouping cells with similar characteristics into the same category, and finally performing subsequent analysis based on the cell classification results.
[0022] This embodiment proposes a method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases, such as... Figure 1 As shown, the method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases in this embodiment includes the following steps: Step S1: Obtain blood sample and its clinical infection status information; preprocess the blood sample to obtain several cells; and obtain the fluorescence intensity value of each cell.
[0023] First, peripheral blood was collected using EDTA anticoagulant tubes to prevent a decrease in monocyte activity during coagulation. Cell separation and loading were performed within 2 hours of collection to ensure that cell surface markers such as CD16 and CD62L were not affected by freezing or delayed activation. Then, the blood sample was treated with erythrocyte lysis buffer (e.g., ACK lysis buffer). After lysing the erythrocytes, they were washed repeatedly 2-3 times with PBS buffer to remove plasma proteins and soluble aggregates, preserving intact mononuclear cells and granulocyte populations. The sample was finally centrifuged at 500×g for 5 min to collect the cell pellet, which was then resuspended in PBS containing 1% BSA to prevent non-specific binding. An amine-reactive live / dead dye (e.g., Fixable Viability Dye eFluor™) was used. 506) Cells were stained, and dead cells were removed using flow cytometry. Samples with a live cell ratio ≥90% were then entered into the subsequent high-dimensional immunophenotyping analysis process. Furthermore, apoptotic cells and debris were removed using a pre-sorting module of the flow cytometry, retaining a stable FSC-A / SSC-A cell population without multiple peaks. This pre-processing of cells in the blood sample served as the basis for subsequent high-dimensional feature extraction. It should be noted that all cells mentioned below are those selected after pre-processing.
[0024] Two-parameter scatter plot of flow cytometry as shown Figure 2 As shown, the horizontal axis represents the macrophage colony-stimulating factor receptor (CD115, reflecting the cell's ability to respond to M-CSF signals), and the vertical axis represents CD14 (an important marker molecule for monocytes and some macrophages, involved in LPS recognition); blue → green → yellow → red indicates that the event density gradually increases, and the red area is the cell-dense area.
[0025] In high-dimensional streaming data acquisition, minute shifts in signal strength between different batches and channels can lead to spurious clusters or misclassifications in later stages of clustering algorithms. Therefore, a calibration and noise removal process needs to be established at the raw signal level to ensure the comparability of subsequent feature spaces and the stability of clustering.
[0026] The specific process of data quality control and artifact filtering is as follows: standard fluorescent microspheres are used to calibrate the voltage and compensation matrix of each channel to ensure the uniformity of fluorescence intensity signals across different detection batches. The calibration process achieves channel response consistency by fitting a standard curve. The fluorescent microspheres are monochromatic compensation microspheres that match the fluorescent dyes used in the experiment. During the calibration process, each monochromated sample is collected sequentially, and the compensation matrix is automatically calculated using flow cytometry software. The symmetry of the double-negative population after compensation is then verified.
[0027] After calibration, samples were collected using fixed voltage parameters to ensure the merging of data from multiple batches. A screening interval was set using FSC-A for each cell as the x-axis and SSC-A as the y-axis to exclude: cell debris (FSC-A < 1000) and cell aggregates (FSC-H / FSC-A deviation of 1 ± 0.1, representing non-single-cell events). Background thresholds were set for the signal intensity of all channels. Based on the fluorescence signal distribution of isotype control antibodies or FMO control samples, negative thresholds were set for each detection channel to exclude autofluorescence and non-specific binding signals. In this example, the threshold was set according to the 3σ principle: the threshold is the sum of the mean negative population fluorescence signal value of the FMO control for each channel plus three times the standard deviation, removing autofluorescence and non-specific staining signals below this threshold.
[0028] After the above processing and quality control, a multi-parameter fluorescence signal dataset at the single-cell level is obtained. Its core form is a high-dimensional single-cell expression matrix, with dimensions n×d, where n is the total number of cells and d is the number of detection channels (including scattering and fluorescence signals). Each detected cell is recorded as a row of data in the flow cytometer, and each detection channel (i.e., a fluorescent antibody label or scattering signal) corresponds to a column. The matrix element in the i-th row and j-th column represents the normalized fluorescence intensity value of the i-th cell for the j-th parameter. In this embodiment, the data normalization uses the maximum-minimum normalization method. In specific applications, implementers can also choose other existing data normalization methods for processing, depending on the specific circumstances. The maximum-minimum normalization method is existing technology and will not be elaborated further here.
[0029] Simultaneously, clinical infection status information is obtained, including C-reactive protein, procalcitonin, and pathogen type, which includes viruses, bacteria, and parasites.
[0030] Step S2: Based on the distribution of fluorescence intensity values of each cell, adaptively determine the neighborhood size of each cell; determine the basic density index of each cell based on the neighborhood size of each cell.
[0031] To address the issues of low sensitivity in capturing key rare subpopulations of infectious diseases and the inability to identify continuous changes in cellular functional states in high-dimensional flow cytometry data, and to achieve accurate clustering and dynamic state analysis, it is necessary to analyze the pathophysiological characteristics of infectious diseases and the characteristics of continuous changes in cellular functional states, and to construct a data clustering method optimized for the characteristics of immune cells in infectious diseases.
[0032] During the course of infectious diseases, the immune system undergoes dramatic dynamic changes. For example, at the peak of acute bacterial infection, neutrophils proliferate and activate in large numbers, resulting in extremely high local density of neutrophils in the phenotypic space; while in the late stage of chronic viral infection, antigen-specific T cells may be depleted and their numbers may decrease sharply, forming sparse, isolated populations.
[0033] If a globally fixed value of K is used to construct the neighborhood graph, in high-density regions, the K neighbors may all come from the same activation population, making it impossible to distinguish their internal heterogeneity; while in low-density regions, the K neighbors may contain a large number of irrelevant background cells, drowning out rare subpopulations. Therefore, the construction of the neighborhood needs to be aware of and respond to this local density change driven by the infection state.
[0034] The UMAP algorithm is used to reduce the dimensionality of high-dimensional single-cell expression matrices to obtain low-dimensional embedding matrices. Under natural conditions, the differences between immune cell phenotypes are not linearly changing, but rather a nonlinear manifold structure formed by the coupling of multiple factors such as cell lineage, activation pathways, and infection pressure. UMAP can reproduce this nonlinear immune lineage continuity in a low-dimensional space by preserving the local topological structure of the data.
[0035] In the low-dimensional space, the Euclidean distance from each cell to its k-th nearest neighbor is obtained and denoted as the first distance for each cell. The smaller the first distance, the denser the cell distribution, i.e., the higher the density of the local region. Here, k is a preset initial threshold. In this embodiment, k is 10. In specific applications, the implementer can set it according to the specific situation.
[0036] Obtain the median of the first distances corresponding to all cells; adjust the preset initial threshold using the ratio between this median and the first distance corresponding to each cell to obtain the neighborhood size of each cell. Determine the neighborhood size of each cell based on the rounded result of the product of this ratio and the preset initial threshold.
[0037] As a concrete example, the method for determining the neighborhood size is given. For any cell, its corresponding neighborhood size can be specifically represented as: ;
[0038] in, This indicates the K value corresponding to the cell. This represents the rounding function. This indicates a preset initial threshold. This represents the median of the first distances corresponding to all cells. This indicates the first distance corresponding to the cell.
[0039] The neighborhood size is adaptively adjusted by using the median of the first distances corresponding to all cells and the first distances corresponding to each cell. This ensures that the neighborhood range of high-density regions is increased to capture finer internal structures, while the neighborhood size of low-density regions is reduced to avoid introducing too much noise.
[0040] Understandably, by adaptively determining the K value corresponding to a single cell in the above manner, and taking the K nearest neighbor of each cell as its neighborhood size, the neighborhood size of each cell is adaptively obtained.
[0041] The above steps have captured the local structure of the phenotypic space through adaptive K-values, but they do not distinguish whether these structures are related to infection. For example, high-density clusters of neutrophils in acute bacterial infections are pathologically relevant, while aggregations caused by technical noise are irrelevant. By weighting local density with infection regulatory factors, the density signal of pathologically relevant structures can be enhanced, while irrelevant noise can be weakened, making density quantification more consistent with the laws of infection and immunity. The number of cells in the neighborhood of each cell is counted separately as the baseline density index for each cell.
[0042] Thus, the basic density index of each cell was determined through the above methods.
[0043] Step S3: Combine the fluorescence intensity value of each cell, the corresponding core markers, auxiliary markers, clinical infection status information, and the baseline density index to obtain the local weighted density of each cell.
[0044] The type and stage of infectious diseases profoundly shape the phenotypic landscape of immune cells. For example, acute viral infections may induce a strong type I interferon response, resulting in widespread overexpression of activation markers (such as CD38 and HLA-DR), leading to phenotypic spatial compression; while parasitic infections may induce Th2 polarization, causing CD4+ T cells to favor specific phenotypic regions. These pathological features lead to systematic shifts in the distribution morphology of cell populations (such as cluster density and diffusion). Current clustering algorithms cannot perform well in all these scenarios. Therefore, targeted optimization of the clustering process is needed.
[0045] Phenotypic shifts in infectious diseases are directly driven by infection type and inflammation severity: high inflammation corresponds to a diffuse distribution of widely activated immune cells, while pathogen type determines the direction of the phenotypic shift. If clustering parameters do not incorporate this information, it can lead to biased identification of pathology-related subgroups. Therefore, infection status indicators need to be integrated as moderating factors to quantify the strength of the influence of infection status on phenotypic distribution.
[0046] If the pathogen type of the blood sample is a virus, then the pathogen type parameter is set to a preset first value; if the pathogen type of the blood sample is bacteria, then the pathogen type parameter is set to a preset second value; if the pathogen type of the blood sample is a parasite, then the pathogen type parameter is set to a preset third value. The preset first, second, and third values all range from (0, 1), and the preset first value is less than the preset second value, and the preset second value is less than the preset third value. In this embodiment, the preset first value is 0.1, the preset second value is 0.3, and the preset third value is 0.6.
[0047] Furthermore, by combining C-reactive protein, procalcitonin, and pathogen type from blood samples, infection status regulatory factors are obtained. As a specific example, the normalized values of C-reactive protein concentration, procalcitonin concentration, and pathogen type parameter from blood samples are used as the infection status regulatory factor; the larger the infection status regulatory factor, the more significant the shift in infection status.
[0048] Based on the pathogen type (virus, bacteria, parasite) and pathological state (acute, chronic) of the blood sample, core biomarkers and auxiliary biomarkers are determined to form a standardized combination of blood samples. For example, the core biomarker for acute viral infection is CD38+HLA-DR+ (CD8+ T cells), and its auxiliary biomarker is CD69+Ki-67+ (CD8+ T cells). It should be noted that this process can be based on consulting clinical consensus and other materials.
[0049] Next, this embodiment will be illustrated using a single cell as an example. Other cells can be processed using the method provided in this embodiment.
[0050] Specifically, any cell is designated as the cell to be analyzed. The average value of all normalized fluorescence intensity values corresponding to the core marker of the cell to be analyzed is calculated and recorded as the first average value. The average value of all normalized fluorescence intensity values corresponding to the auxiliary marker of the cell to be analyzed is calculated and recorded as the second average value.
[0051] The phenotypic pathology association score of the cells to be analyzed is obtained by weighted summation of the first average and the second average. The phenotypic pathology association score is used to reflect the overall strength of cell expression of these specific markers. The more critical the marker expressed by the cell and the higher the expression level, the larger the value. In the weighted summation, the weight of the first average is greater than the weight of the second average. In this embodiment, the weight of the first average is 0.65 and the weight of the second average is 0.35.
[0052] Furthermore, by combining the baseline density index of the cells to be analyzed, the infection status regulator, and the phenotypic pathology association score, the local weighted density of the cells to be analyzed is obtained. As a specific example, the product of the infection status regulator and the phenotypic pathology association score of the cells to be analyzed is calculated, and this product is recorded as the first product; the sum of the first product and a constant 1 is calculated; this sum is then multiplied by the baseline density index of the cells to be analyzed, and this product is determined as the local weighted density of the cells to be analyzed. The stronger the pathological association and the more significant the infection status shift, the greater the amplification of the local weighted density, thus highlighting the density signal of the pathologically related population.
[0053] Understandably, the local weighted density of each cell can be obtained through the above methods.
[0054] Step S4: Determine the density threshold based on the local weighted density of each cell; cluster all cells according to the differences in fluorescence intensity values of different cells and the density threshold to obtain cell clusters.
[0055] The above steps obtained the local weighted density of each cell. Cells in different infection states require different density thresholds: in high-inflammatory states, cells are diffusely distributed, such as the activated population of acute viral infections, requiring a lower threshold to capture weakly correlated subpopulations; in chronic infections, rare subpopulations are densely distributed, such as antigen-specific T cell clones, requiring a higher threshold to filter background noise. Therefore, the density threshold needs to be negatively correlated with infection state regulators, while also referencing the overall level of weighted local density to avoid extreme values. The density threshold will then be determined based on the local weighted density and infection state regulators.
[0056] Specifically, the median of the local weighted density of all cells is obtained; cells with a local weighted density less than this median are selected as candidate cells.
[0057] All candidate cells are sorted in descending order of local weighted density. The difference in local weighted density between adjacent candidate cells is calculated, which is obtained by subtracting the local weighted density of the next candidate cell from the local weighted density of the previous candidate cell. The local weighted density with the largest difference is selected as the baseline threshold to ensure that the baseline threshold covers the medium-density population. The difference obtained by subtracting the infection status regulation factor from the constant 1 is recorded as the first difference. The product of the first difference and the baseline threshold is used as the density threshold. The larger the infection status regulation factor, the smaller the density threshold is needed to enhance sensitivity to the diffuse population; the smaller the infection status regulation factor, the larger the density threshold is needed to focus on the dense and rare subpopulation.
[0058] This embodiment has captured the local topological structure of the phenotypic space through UMAP dimensionality reduction and adaptive K-value. The density threshold is adapted to the infection-driven distribution characteristics. Based on the edges (cell similarity) of the neighborhood graph, the density threshold is combined to determine whether cells belong to the same subpopulation: in high-density pathologically related regions, a lower density threshold can distinguish the fine subpopulations within; in low-density rare regions, a higher density threshold can accurately locate closely related rare cells.
[0059] Furthermore, in the low-dimensional space, the similarity between pairs of cells is evaluated based on the variance of the Euclidean distance between each pair of cells and the Euclidean distance between all cells. The similarity between the i-th cell and the j-th cell can be expressed as: ; in, This represents the similarity between the i-th cell and the j-th cell. This represents the Euclidean distance between the i-th cell and the j-th cell. This represents the variance of the Euclidean distances between all pairs of cells. This represents an exponential function with the natural constant as its base.
[0060] Cell similarity was measured using Euclidean distance. It's important to note that if the denominator in the similarity calculation formula is 0, then the similarity between cells is set to 1. Next, cells are classified based on pairwise similarity and a density threshold.
[0061] Specifically, based on the similarity between pairs of cells and a density threshold, the DBSCAN algorithm is used to cluster all cells. If the similarity between any two cells is greater than or equal to the density threshold, then one cell is determined to be a density reachable point of the other cell, and cells that are density reachable from each other are grouped into the same cell cluster. This achieves adaptive clustering of all cells, resulting in multiple cell clusters. The DBSCAN algorithm is existing technology and will not be elaborated further here.
[0062] After obtaining cell clusters, physicians can then analyze the immune cell subsets in clinical laboratory samples for infectious diseases based on these clusters. They can also align and interpret the cell clusters with known immunological knowledge: for each cell cluster, its characteristics are matched with the typical phenotypes of known immune cell subsets and annotated; a comprehensive report is then generated based on the annotation results, including the proportion of major immune cell subsets and the abundance of rare subsets, providing a basis for clinical diagnosis and prognostic assessment.
[0063] Thus, the method provided in this embodiment has enabled the analysis of immune cell subsets in clinical laboratory samples of infectious diseases.
[0064] This embodiment first adaptively adjusts the neighborhood size of each cell based on the distribution of fluorescence intensity values in the blood sample. It then determines the baseline density index of each cell based on the neighborhood information, ensuring that more refined internal structures are captured while avoiding excessive noise. Based on infection status perception features, the density threshold is dynamically adjusted in conjunction with the baseline density index. All cells are then adaptively clustered based on their neighborhood. Finally, the clustering results are integrated to analyze immune cell subpopulations in clinical laboratory samples of infectious diseases. This method accurately adapts to the dynamic characteristics of infectious disease immunophenotypes, adaptively determining the neighborhood size through infection status-driven parameters. It addresses the problems of poor adaptation to density heterogeneity and missed detection of rare subpopulations in general algorithms, achieving accurate identification of finely segmented subpopulations. This improves the accuracy and clinical relevance of the analysis results, providing a reliable basis for infection type identification, disease assessment, and treatment monitoring.
[0065] Example of an immune cell subset analysis system for clinical laboratory samples for infectious diseases: A block diagram of a clinical laboratory sample immune cell subset analysis system for infectious diseases is shown below. Figure 3 As shown, the system may include a data acquisition module, a neighborhood analysis module, a local evaluation module, and a clustering module.
[0066] The data acquisition module is used to acquire blood samples and their clinical infection status information, preprocess the blood samples to obtain several cells, and acquire the fluorescence intensity value of each cell. The neighborhood analysis module is used to adaptively determine the neighborhood size of each cell based on the distribution of fluorescence intensity values of each cell; and to determine the basic density index of each cell based on the neighborhood size of each cell. The local evaluation module is used to combine the fluorescence intensity value of each cell, the corresponding core markers, auxiliary markers, clinical infection status information, and the baseline density index to obtain the local weighted density of each cell. The clustering module is used to determine a density threshold based on the local weighted density of each cell; and to cluster all cells based on the differences in fluorescence intensity values of different cells and the density threshold to obtain cell clusters.
[0067] It should be understood that the structural block diagram and modules of a clinical laboratory sample immune cell subset analysis system for infectious diseases can be implemented in various ways. For example, in some embodiments, the system and its modules can be implemented by hardware, software, or a combination of software and hardware. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by appropriate instructions, such as a microprocessor or dedicated hardware. Those skilled in the art will understand that the above-described methods and systems can be implemented using computer-executable instructions and / or included in processor control code, for example, on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The systems and modules of this specification can be implemented not only by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., but also by software, for example, executed by various types of processors, or by a combination of the above-described hardware circuits and software (e.g., firmware).
[0068] For more details about the above modules, please refer to other parts of this manual; they will not be repeated here.
[0069] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases, characterized in that, Includes the following steps: Obtain blood samples and their clinical infection status information; preprocess the blood samples to obtain several cells; and obtain the fluorescence intensity value of each cell. Based on the distribution of fluorescence intensity values of each cell, the neighborhood size of each cell is adaptively determined; and the basic density index of each cell is determined based on the neighborhood size of each cell. By combining the fluorescence intensity values of each cell, the corresponding core markers, auxiliary markers, clinical infection status information, and the aforementioned baseline density index, the local weighted density of each cell is obtained. A density threshold is determined based on the local weighted density of each cell; all cells are clustered based on the differences in fluorescence intensity values of different cells and the density threshold to obtain cell clusters.
2. The method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases according to claim 1, characterized in that, The adaptive determination of the neighborhood size of each cell based on the distribution of fluorescence intensity values includes: Based on the fluorescence intensity values of all detection channels of each cell, a high-dimensional single-cell expression matrix is constructed; the UMAP algorithm is used to reduce the dimensionality of the high-dimensional single-cell expression matrix. In the low-dimensional space, the Euclidean distance from each cell to its k-th nearest neighbor cell is obtained and recorded as the first distance for each cell, where k is a preset initial threshold. Obtain the median of the first distances corresponding to all cells; adjust the preset initial threshold using the ratio between the median and the first distances corresponding to each cell to obtain the neighborhood size of each cell.
3. The method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases according to claim 2, characterized in that, The step of adjusting a preset initial threshold using the ratio between the median and the first distance corresponding to each cell to obtain the neighborhood size of each cell includes: The neighborhood size of each cell is determined by rounding the product of the ratio and the preset initial threshold.
4. The method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases according to claim 1, characterized in that, The step of determining the basic density index of each cell based on the neighborhood size of each cell includes: obtaining the number of cells in the neighborhood of each cell as the basic density index of each cell.
5. The method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases according to claim 1, characterized in that, The local weighted density of each cell is obtained by combining the fluorescence intensity value of each cell, the corresponding core markers, auxiliary markers, clinical infection status information, and the baseline density index, including: Infection status regulatory factors were obtained by combining C-reactive protein, procalcitonin, and pathogen type in blood samples; the clinical infection status information included C-reactive protein, procalcitonin, and pathogen type. The first average value of all normalized fluorescence intensity values corresponding to the core markers of the cells to be analyzed, and the second average value of all normalized fluorescence intensity values corresponding to the auxiliary markers of the cells to be analyzed are obtained respectively. The phenotypic pathological association score of the cells to be analyzed is obtained by weighted summation of the first average and the second average; wherein, when performing weighted summation, the weight of the first average is greater than the weight of the second average. By combining the baseline density index of the cells to be analyzed, the infection status regulator, and the phenotypic pathology correlation score, the local weighted density of the cells to be analyzed is obtained. The cell to be analyzed can be any cell.
6. The method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases according to claim 5, characterized in that, The combined blood sample analysis of C-reactive protein, procalcitonin, and pathogen type yielded regulatory factors of infection status, including: The average of the normalized values of C-reactive protein concentration, procalcitonin concentration, and pathogen type parameter in blood samples was determined as the regulatory factor of infection status. If the pathogen type of the blood sample is a virus, then the pathogen type parameter is set to a preset first value; if the pathogen type of the blood sample is bacteria, then the pathogen type parameter is set to a preset second value; if the pathogen type of the blood sample is a parasite, then the pathogen type parameter is set to a preset third value. The preset first value, preset second value, and preset third value all have a range of (0, 1), and the preset first value is less than the preset second value, and the preset second value is less than the preset third value.
7. The method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases according to claim 5, characterized in that, The process of combining the baseline density index of the cells to be analyzed, the infection status regulator, and the phenotypic pathological correlation score to obtain the local weighted density of the cells to be analyzed includes: Calculate the first product of the association scores between the infection status regulator and the phenotypic pathology of the cells to be analyzed; calculate the sum of the first product and the constant 1; The product of the sum and the baseline density index of the cell to be analyzed is determined as the local weighted density of the cell to be analyzed.
8. The method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases according to claim 5, characterized in that, The determination of the density threshold based on the local weighted density of each cell includes: Obtain the median of the local weighted density of all cells; Cells with a local weighted density less than the median are selected as candidate cells; All candidate cells are sorted in descending order of local weighted density. The difference between the local weighted densities of two adjacent candidate cells is calculated. The local weighted density with the largest local weighted density in the two candidate cells corresponding to the two cells with the largest difference is selected as the basic threshold. The density threshold is obtained based on the base threshold and the infection status adjustment factor.
9. The method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases according to claim 8, characterized in that, The process of obtaining the density threshold based on the base threshold and the infection status adjustment factor includes: Calculate the first difference between constant 1 and the infection status regulation factor; The product of the first difference and the base threshold is determined as the density threshold.
10. The method for analyzing immune cell subsets in clinical laboratory samples for infectious diseases according to claim 2, characterized in that, The process of clustering all cells based on differences in fluorescence intensity values and a density threshold to obtain cell clusters includes: In a low-dimensional space, the similarity between pairs of cells is evaluated based on the variance of the Euclidean distance between each pair of cells and the Euclidean distance between all cells; the Euclidean distance is used to characterize the difference between the fluorescence intensity values of the cells. Based on the relationship between the similarity and the density threshold, the DBSCAN algorithm is used to cluster all cells to obtain multiple cell clusters.