Co-disease trajectory classification method and system based on disease clustering

Abstract

The invention relates to a co-disease trajectory classification method and system based on disease clustering, belongs to the technical field of medical health, and solves the problem of inaccurate co-disease trajectory classification in the prior art. The method comprises the following steps: acquiring co-disease trajectory data, extracting all diseases in the co-disease trajectory data, and calculating correlation among the diseases; the attributes of the diseases are judged on the basis of the correlation between the diseases, bipartite nested clustering is conducted on all the diseases in the co-disease trajectory data on the basis of the attributes of the diseases, and the attributes of the diseases comprise hub diseases and non-hub diseases; and obtaining the type of each co-disease track according to a disease clustering result and the attribute of each disease. And accurate common disease track classification is realized.

Description

Technical Field

[0001] The present invention relates to the field of medical health technology, and in particular to a comorbidity trajectory classification method and system based on disease clustering. Background Art

[0002] In the field of medicine and health, comorbidity trajectory clustering technology has broad application prospects. Based on the data of disease occurrence and development, the technology uses a clustering algorithm to group disease trajectories according to similarities to reveal the characteristics of comorbidity occurrence and development patterns. This technology can help doctors better identify and manage patient groups with similar health problems and develop more effective treatment plans and preventive measures. In addition, this technology can also be used to study the pathogenesis and epidemic patterns of diseases, providing a scientific basis for the formulation of public health policies.

[0003] However, existing trajectory clustering methods have significant limitations for clustering disease trajectories. Due to the extensive and close association between some common diseases (such as hypertension) and many other diseases, existing clustering methods often mistakenly merge these disease trajectories with low correlation into the same group. Take hypertension as an example. It appears in the development paths of many diseases, but current clustering techniques often indiscriminately cluster disease trajectories containing hypertension, even if these trajectories are not highly correlated in clinical practice. Therefore, it is impossible to capture the characteristics of comorbidities, and the clustering accuracy for comorbidities is low, which in turn affects the accuracy of treatment plans and preventive measures. Summary of the invention

[0004] In view of the above analysis, the embodiments of the present invention aim to provide a comorbidity trajectory classification method and system based on disease clustering, so as to solve the problem of low accuracy of existing comorbidity trajectory classification.

[0005] On the one hand, an embodiment of the present invention provides a comorbidity trajectory classification method based on disease clustering, comprising the following steps:

[0006] Obtain comorbidity trajectory data, extract all diseases in the comorbidity trajectory data, and calculate the correlation between diseases;

[0007] Determine the attributes of the disease based on the correlation between the diseases, and perform binary nested clustering on all diseases in the comorbidity trajectory data based on the attributes of the disease, wherein the attributes of the disease include hub diseases and non-hub diseases;

[0008] The type of each comorbidity trajectory is obtained based on the disease clustering results and the attributes of each disease.

[0009] Based on the further improvement of the above method, the attributes of the disease are determined based on the correlation between the diseases, and binary nested clustering is performed on all diseases in the comorbidity trajectory data based on the attributes of the disease, including:

[0010] S20, perform binary clustering on all disease sets to obtain two sub-class disease sets;

[0011] S21. For each sub-category disease set, determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set; if the sub-category disease set has a hub disease and the number of non-hub diseases is greater than a first threshold, then treat the sub-category disease set as a subset to be clustered; otherwise, the sub-category disease set is a set in the clustering result;

[0012] S22, if there is a subset to be clustered, execute S23-S25, otherwise, end clustering;

[0013] S23. For each subset to be clustered, remove the diseases whose disease attributes are hub diseases in the subset to be clustered;

[0014] S24, performing binary classification clustering on each subset to be clustered after removing the hub disease to obtain two sub-class disease sets corresponding to each subset to be clustered;

[0015] S25. Return to step S21.

[0016] Based on the further improvement of the above method, the following method is used to determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set:

[0017] Calculate the q-quantile of the correlation between diseases in the sub-class disease set, and construct an undirected network corresponding to the sub-class disease set based on the q-quantile and the correlation between diseases;

[0018] The degree of each node in the undirected network is calculated, and the attribute of the disease corresponding to each node is determined based on the degree of each node.

[0019] Based on the further improvement of the above method, if the degree of the node d D >α×(|B|-1), the disease corresponding to the node is a hub disease, otherwise the disease corresponding to the node is a non-hub disease, α represents the proportion parameter, and |B| represents the number of diseases in the subclass disease set B.

[0020] Based on the further improvement of the above method, the type of each comorbidity trajectory is obtained in the following way according to the disease clustering results and the attributes of each disease:

[0021] For each comorbidity trajectory, if there is no non-hub disease in the comorbidity trajectory, then the comorbidity trajectory belongs to the hub comorbidity trajectory; otherwise:

[0022] If there is only one cluster type, and more than a first number of diseases in the comorbidity trajectory belong to the cluster type, then the type of the comorbidity trajectory is the cluster type;

[0023] If there are more than two cluster types, for each of the more than two cluster types, more than a first number of diseases in the comorbidity trajectory belong to the cluster type, then the cluster type to which the first non-hub disease in the comorbidity trajectory belongs is the type of the comorbidity trajectory;

[0024] Otherwise, the comorbidity trajectory belongs to the mixed comorbidity trajectory.

[0025] Based on the further improvement of the above method, the correlation between diseases is calculated in the following way:

[0026]

[0027] in, Denotes the i-th disease D i and the jth disease D j The correlation between Indicates that D is also included i and D j The number of comorbidity trajectories, Indicates that D is included i or D j The number of comorbidity trajectories.

[0028] On the other hand, an embodiment of the present invention provides a comorbidity trajectory classification system based on disease clustering, comprising the following modules:

[0029] A correlation calculation module is used to obtain comorbidity trajectory data, extract all diseases in the comorbidity trajectory data, and calculate the correlation between diseases;

[0030] A disease clustering module, used to determine the attributes of the disease based on the correlation between the diseases, and to perform binary nested clustering on all diseases in the comorbidity trajectory data based on the attributes of the disease, wherein the attributes of the disease include hub diseases and non-hub diseases;

[0031] The trajectory assignment module is used to obtain the type of each comorbidity trajectory based on the disease clustering results and the attributes of each disease.

[0032] Based on the further improvement of the above system, the attributes of the disease are determined based on the correlation between the diseases, and binary nested clustering is performed on all diseases in the comorbidity trajectory data based on the attributes of the disease. The attributes of the disease include hub diseases and non-hub diseases, including:

[0033] S20, perform binary clustering on all disease sets to obtain two sub-class disease sets;

[0034] S21. For each sub-category disease set, determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set; if the sub-category disease set has a hub disease and the number of non-hub diseases is greater than a first threshold, then treat the sub-category disease set as a subset to be clustered; otherwise, the sub-category disease set is a set in the clustering result;

[0035] S22, if there is a subset to be clustered, execute S23-S25, otherwise, end clustering;

[0036] S23. For each subset to be clustered, remove the diseases whose disease attributes are hub diseases in the subset to be clustered;

[0037] S24, performing binary classification clustering on each subset to be clustered after removing the hub disease to obtain two sub-class disease sets corresponding to each subset to be clustered;

[0038] S25. Return to step S21.

[0039] Based on the further improvement of the above system, the following method is used to determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set:

[0040] Calculate the q-quantile of the correlation between diseases in the subclass disease set, and based on the quantile, construct an undirected network corresponding to the subclass disease set with the diseases in the subclass disease set as nodes and the correlation between diseases as edges;

[0041] The degree of each node in the undirected network is calculated, and the attribute of the disease corresponding to each node is determined based on the degree of each node.

[0042] Based on the further improvement of the above system, if the degree of the node d D >α×(|B|-1), the disease corresponding to the node is a hub disease, otherwise the disease corresponding to the node is a non-hub disease, α represents the proportion parameter, and |B| represents the number of diseases in the subclass disease set B.

[0043] Compared with the prior art, the present invention obtains the comorbidity trajectory data, extracts the diseases covered by the trajectory and calculates the correlation between the diseases, then judges the disease attributes based on the correlation, and performs binary nested clustering on all the diseases covered by the trajectory based on the disease attributes, so as to identify the hierarchical relationship between the clusters. By identifying the hub diseases linking different clustering groups based on the correlation, these common diseases can be effectively separated, so as to accurately obtain the type of each comorbidity trajectory based on the disease clustering results and the attributes of each disease, thereby improving the accuracy of classification; at the same time, the comorbidity trajectory classification method provided in this embodiment overcomes the defects of traditional non-fuzzy processing that does not allow overlapping structures between groups and fuzzy clustering cannot form hub diseases belonging to multiple subclasses rather than hub nodes belonging to only a single subclass, and allows a certain degree of overlap between different groups, which makes the clustering results more in line with the actual situation, because the relationship between diseases is often not absolute, but there are multiple possibilities and intersections. By supporting overlapping structures and performing nested clustering, the trajectory types obtained by the present invention can more comprehensively reflect the complexity and diversity of diseases, and help doctors better understand the development paths and interrelationships of diseases, thereby formulating more accurate treatment plans and preventive measures. In addition, it can also be used to study the pathogenesis and epidemic patterns of diseases and provide a scientific basis for the formulation of public health policies.

[0044] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are only used for the purpose of illustrating specific embodiments and are not to be considered as limiting the present invention. In the entire drawings, the same reference symbols represent the same components;

[0046] Figure 1 This is a flow chart of a comorbidity trajectory classification method based on disease clustering according to an embodiment of the present invention;

[0047] Figure 2 A schematic diagram of an undirected network according to an embodiment of the present invention;

[0048] Figure 3 This is a schematic diagram of disease clustering results according to an embodiment of the present invention;

[0049] Figure 4 This is a block diagram of a comorbidity trajectory classification system based on disease clustering according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0051] Hub diseases are diseases that play a key role in the comorbidity network. Hub diseases link multiple comorbidity classes and belong to multiple clusters at the same time, while non-hub diseases belong to only one cluster.

[0052] Although the existing technology can identify key diseases within a single cluster group, it cannot accurately identify the hub diseases connecting different cluster groups. Therefore, it is difficult to effectively separate these common diseases during the clustering process. Secondly, when dealing with clustering, traditional non-fuzzy clustering methods do not allow overlapping structures between different groups, which to some extent limits the practicality and interpretability of clustering. Although fuzzy clustering methods allow overlap, they lack clear boundaries when defining the affiliation of members and subclasses, resulting in the inability to form a clustering scenario where hub diseases belong to multiple subclasses rather than hub nodes belonging to only a single subclass. Finally, although the existing technology can present a certain hierarchy when displaying the clustering hierarchy, there is a lack of actual node connections between the levels, making this hierarchy lack practical clinical guidance significance. How to effectively identify hub disease nodes based on the characteristics of comorbidity clustering, and how these nodes affect the clustering process to form a clustering mode with an overlapping structure, is still a problem to be solved.

[0053] Based on this, a specific embodiment of the present invention discloses a comorbidity trajectory classification method based on disease clustering, such as Figure 1 As shown, the following steps are included:

[0054] S1. Obtain comorbidity trajectory data, extract all diseases in the comorbidity trajectory data, and calculate the correlation between diseases;

[0055] S2. Determine the attributes of the disease based on the correlation between the diseases, and perform binary nested clustering on all diseases in the comorbidity trajectory data based on the attributes of the disease, wherein the attributes of the disease include hub diseases and non-hub diseases;

[0056] S3. Obtain the type of each comorbidity trajectory based on the disease clustering results and the attributes of each disease.

[0057] When implemented, the comorbidity trajectory is the sequence of disease occurrence, for example, D1→D2→D3, which means that disease D1 develops into disease D2 and then develops into disease D3.

[0058] Compared with the prior art, the comorbidity trajectory classification method based on disease clustering provided in this embodiment obtains the comorbidity trajectory data, extracts the diseases covered by the trajectory and calculates the correlation between the diseases, then judges the disease attributes based on the correlation, and performs binary nested clustering on all the diseases covered by the trajectory based on the disease attributes, so as to identify the hierarchical relationship between clusters. By identifying the hub diseases linking different clustering groups based on the correlation, the hub diseases can be effectively separated, so as to accurately obtain the clustering results of each comorbidity trajectory based on the disease clustering results and the attributes of each disease, thereby improving the accuracy of classification; at the same time, the comorbidity trajectory provided in this embodiment The classification method overcomes the defects of traditional non-fuzzy processing that does not allow overlapping structures between groups and fuzzy clustering cannot form hub diseases belonging to multiple subclasses rather than hub nodes belonging to only a single subclass. It allows a certain degree of overlap between different groups, which makes the clustering results more in line with the actual situation, because the relationship between diseases is often not absolute, but there are multiple possibilities and intersections. By supporting overlapping structures and performing nested clustering, the trajectory types obtained by the present invention can more comprehensively reflect the complexity and diversity of diseases, and help doctors better understand the development path and interrelationships of diseases, so as to formulate more accurate treatment plans and preventive measures. In addition, it can also be used to study the pathogenesis and epidemic laws of diseases, and provide a scientific basis for the formulation of public health policies.

[0059] During implementation, firstly, the comorbidity trajectory data is obtained, and all diseases in the comorbidity trajectory data are extracted. Exemplarily, the comorbidity trajectory is shown in Table 1, and the set consisting of all diseases in the trajectory data is V.

[0060] Table 1: Disease trajectory table

[0061] Track number disease disease disease 1 <![CDATA[D1]]> <![CDATA[D5]]> <![CDATA[D6]]> 2 <![CDATA[D2]]> <![CDATA[D5]]> <![CDATA[D6]]> 3 <![CDATA[D3]]> <![CDATA[D5]]> <![CDATA[D6]]> 4 <![CDATA[D4]]> <![CDATA[D5]]> <![CDATA[D6]]> 5 <![CDATA[D7]]> <![CDATA[D8]]> <![CDATA[D9]]> 6 <![CDATA[D7]]> <![CDATA[D 10 ]]> <![CDATA[D 11 ]]> 7 <![CDATA[D7]]> <![CDATA[D 12 ]]> <![CDATA[D 13 ]]> 8 <![CDATA[D7]]> <![CDATA[D 14 ]]> <![CDATA[D 15 ]]> 9 <![CDATA[D8]]> <![CDATA[D9]]> <![CDATA[D 16 ]]> 10 <![CDATA[D 10 ]]> <![CDATA[D 11 ]]> <![CDATA[D 16 ]]> 11 <![CDATA[D 12 ]]> <![CDATA[D 13 ]]> <![CDATA[D 17 ]]> 12 <![CDATA[D 14 ]]> <![CDATA[D 15 ]]> <![CDATA[D 17 ]]>

[0062] When implemented, the correlation between diseases is calculated in the following way:

[0063]

[0064] in, Denotes the i-th disease D i and the jth disease D j The correlation between Indicates that D is also included i and D j The number of comorbidity trajectories, Indicates that D is included i or D j The number of comorbidity trajectories.

[0065] The correlation between any two diseases can be characterized by calculating the correlation. During implementation, a correlation matrix between diseases can be constructed for subsequent use.

[0066] Then, binary nested clustering is performed on the disease set V. Specifically, step S2 includes:

[0067] S20, perform binary clustering on all disease sets to obtain two sub-class disease sets;

[0068] S21. For each sub-category disease set, determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set; if the sub-category disease set has a hub disease and the number of non-hub diseases is greater than a first threshold, then treat the sub-category disease set as a subset to be clustered; otherwise, the sub-category disease set is a set in the clustering result;

[0069] S22, if there is a subset to be clustered, execute S23-S25, otherwise, end clustering;

[0070] S23. For each subset to be clustered, remove the diseases whose disease attributes are hub diseases in the subset to be clustered;

[0071] S24, performing binary classification clustering on each subset to be clustered after removing the hub disease to obtain two sub-class disease sets corresponding to each subset to be clustered;

[0072] S25. Return to step S21.

[0073] Initially, for each disease in the set of all diseases, the correlation between the disease and each disease in the set of all diseases is used as the characteristic data of the disease, that is, as the input data for clustering, binary clustering is performed, and all disease sets are clustered into two categories to obtain two sub-class disease sets. During implementation, the k-means clustering algorithm can be used for binary clustering. For example, binary clustering is performed on all disease sets V in Table 1 to obtain two sub-class disease sets A 11 and A 12 , A 11 Including diseases D1 to D6, A 12 Including diseases D7~D 17 .

[0074] For each sub-category disease set obtained, search for the hub disease therein, that is, determine whether the attribute of the disease in the set is a hub disease or a non-hub disease.

[0075] Specifically, the following method is used to determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set:

[0076] S211, calculating the q-quantile of the correlation between diseases in the sub-class disease set, and constructing an undirected network corresponding to the sub-class disease set based on the q-quantile and the correlation between diseases;

[0077] S212. Calculate the degree of each node in the undirected network, and determine the attributes of the diseases corresponding to the nodes based on the degrees of the nodes.

[0078] During implementation, first calculate the q-quantile of the correlations between diseases in the subclass disease set, where 0 < q < 1. For example, q = 0.5, that is, calculate the 0.5 quantile of the correlations between diseases in the subclass disease set. The q-quantile is used to control the sparsity of the undirected network (i.e., the number of edges). The larger the value of q, the fewer the edges in the undirected network, the sparser the obtained network. Consequently, the degree of the nodes in step S212 is smaller, and thus the number of hub nodes obtained in step S21, that is, the number of hub diseases, is fewer.

[0079] After calculating the q-quantile, based on the quantile, construct an undirected network corresponding to the subclass disease set with diseases as nodes and the correlations between diseases as edges. During implementation, if the correlation between two diseases is greater than the q-quantile, there is a connecting edge between the nodes corresponding to the two diseases; otherwise, there is no connecting edge between the two diseases, thereby constructing the undirected network. Calculate the degree of each node in the network according to the constructed undirected network. The degree of a node is the number of edges connected to the node.

[0080] Then, determine the attributes of the diseases corresponding to the nodes according to the degrees of the nodes. Specifically, if the degree d of the node D > α × (|B| - 1), then the disease corresponding to the node is a hub disease; otherwise, the disease corresponding to the node is a non-hub disease. α represents the proportionality parameter, and |B| represents the number of diseases in the subclass disease set B. During implementation, for example, α can take 0.5. After identifying the hub diseases, the diseases in the subclass disease set that are not hub diseases are non-hub diseases.

[0081] Subclass disease set A 11 The 0.5 quantile is 0.25. Therefore, in the constructed undirected network, there is only an edge between nodes D5 and D6. Therefore, there are no hub diseases, and all are non-hub diseases. Subclass disease set A 12 The 0.5 quantile is 0. Therefore, the constructed undirected network is as Figure 2 shown. The degree of node D7 is 8. According to the above judgment formula, disease D7 is a hub disease, and the others are non-hub diseases.

[0082] If there are hub diseases in the subclass disease set, but the number of non-hub diseases is less than or equal to the first threshold, or there are no hub diseases, it indicates that the number of non-hub diseases is small, and there is no need to further analyze the possible more subclasses among them, or the non-hub diseases are grouped into one class not caused by hub diseases. Therefore, no further classification of this subclass disease set is performed to the next level, and this subclass disease set is a class set in the clustering result. For example, in subclass disease set A 11 there are no hub diseases. Therefore, subclass disease set A11 As a cluster in the clustering result, no further clustering is performed.

[0083] If there is a hub disease in the sub-class disease set and the number of non-hub diseases is greater than the first threshold τ, it means that the number of non-hub diseases in the set is still large, and it may be clustered into one category due to the existence of hub diseases. There may be more sub-classes in non-hub diseases. Therefore, it is necessary to continue to classify the sub-class disease set to the next level and treat the sub-class disease set as a subset to be clustered and continue clustering. Assuming τ = 5, the sub-class disease set A 12 If there is a hub disease and the number of non-hub diseases is greater than the first threshold τ, it is necessary to continue to perform binary clustering and cluster the sub-disease set A 12 as a subset to be clustered.

[0084] Since there are subsets to be clustered at this time, for each subset to be clustered, remove the hub disease in it, for example, set A 12 Disease D7 is removed from the cluster, and then the subset to be clustered without the hub disease is classified into two sub-class disease sets corresponding to the subset to be clustered. For example, the two sub-class disease sets obtained are A 121 and A 122 , A 121 Including diseases D8, D9, D 10 ,D 11 ,D 16 , A 122 Including diseases D 12 ,D 13 ,D 14 ,D 15 ,D 17 , then return to step S21 to determine the attributes of the disease in each sub-class disease set, since A 121 and A 122 There is no hub disease, so the sub-disease set is A 121 and A 122 As a cluster set in the clustering result, there is no subset to be clustered at this time, and the clustering is ended. The clustering result is as follows: Figure 3 shown.

[0085] Diseases are clustered into three categories, C1, C2 and C3, among which the second largest category C2 and the second largest category C3 are connected by the hub disease node D7.

[0086] The classification method of the present invention more accurately identifies the hub diseases connecting different clustering groups. These hub diseases frequently appear in multiple disease trajectories, but they do not represent high correlation in all cases. In addition to the initial clustering in step S20, the present invention accurately identifies the attributes of the disease based on the correlation of the disease before each binary clustering, and obtains the set to be clustered according to the attributes of the disease, so that it is possible to avoid erroneously merging unrelated disease trajectories together, thereby improving the accuracy and clinical guidance of clustering. In view of the limitation that the traditional non-fuzzy clustering method cannot handle overlapping structures, the clustering strategy of the present invention allows a certain degree of overlap between different groups, which makes the clustering results more in line with the actual situation, because the relationship between diseases is often not absolute, but there are multiple possibilities and intersections. By supporting overlapping structures, the present invention can more comprehensively reflect the complexity and diversity of diseases. And through binary nested clustering, the affiliation of each disease node in the clustering process can be made clearer, so as to construct a hierarchical structure with actual node connections, so that the clustering results not only show hierarchical nature, but also have clear clinical guidance significance between each level. This hierarchy helps doctors better understand the development pathways and interrelationships of diseases, thereby developing more precise treatment plans and preventive measures.

[0087] Specifically, the type of each comorbidity trajectory is obtained in the following way according to the disease clustering results and the attributes of each disease:

[0088] For each comorbidity trajectory, if there is no non-hub disease in the comorbidity trajectory, then the comorbidity trajectory belongs to the hub comorbidity trajectory; otherwise:

[0089] If there is only one cluster type, and more than a first number of diseases in the comorbidity trajectory belong to the cluster type, then the type of the comorbidity trajectory is the cluster type;

[0090] If there are more than two cluster types, for each of the more than two cluster types, more than a first number of diseases in the comorbidity trajectory belong to the cluster type, then the cluster type to which the first non-hub disease in the comorbidity trajectory belongs is the type of the comorbidity trajectory;

[0091] Otherwise, the comorbidity trajectory belongs to the mixed comorbidity trajectory.

[0092] The first number is half of the number of diseases in the comorbidity trajectory.

[0093] For example, for the trajectory D1→D5→D6 in Table 1, all three diseases belong to class C1, so the type of this trajectory is C1. For the trajectory D7→D8→D9, D8 and D9 are non-hub diseases and belong to class C2, while D7 is a hub disease and belongs to classes C2 and C3, so the type of this trajectory is C2.

[0094] For example, if there is a trajectory D7→D 10 →D 13 , D7 is a hub disease, belonging to C2 and C3 categories, D 10 and D 13 are non-hub diseases, belonging to categories C2 and C3 respectively. That is, category C2 contains two diseases, which exceeds half of the number of diseases in the trajectory, and category C3 also contains two diseases, which exceeds half of the number of diseases in the trajectory, but D 10 Before D 13 , therefore, the type of this trajectory is C2.

[0095] For all trajectories composed of hub diseases, a new class is formed separately, which is called the hub class. If a comorbidity trajectory does not have more than the first number of diseases belonging to the same cluster type, then this trajectory is regarded as a mixed class comorbidity trajectory.

[0096] Based on the type results of the comorbidity trajectory obtained, in-depth research can be conducted on the type group of each comorbidity trajectory. For example, all trajectories belonging to type C1 can be studied, and the interaction relationships such as the development time between diseases within this type can be analyzed. The comorbidity patterns, key disease development pathways and potential common risks unique to type C1 can be identified, so as to develop more personalized and effective intervention plans, such as recommendations on drug treatment, lifestyle intervention, health management, etc., to block or delay the development of comorbidities.

[0097] In addition, the individual disease trajectory of the patient can also be compared with the trajectory type results to find out the trajectory type to which the individual disease trajectory of the patient belongs. Referring to the common characteristics of the diseases within the trajectory type group, an integrated diagnosis and treatment plan for the trajectory type group can be formulated to replace the traditional diagnosis and treatment plan for a single disease. For example, if the individual disease trajectory of the patient is D7→D8, and the individual disease trajectory belongs to trajectory type C2, an integrated diagnosis and treatment plan can be formulated based on the common characteristics of the diseases within C2, thereby replacing the traditional diagnosis and treatment plan for a single disease, so as to block or delay the development of comorbidities.

[0098] The results of the present invention can provide more comprehensive and in-depth data support for policy makers. Based on the clustering results of the present invention, policy makers can understand the distribution characteristics, influencing factors and development trends of different comorbidity patterns, so as to formulate more targeted and effective public health strategies to cope with increasingly complex health challenges.

[0099] On the other hand, a specific embodiment of the present invention discloses a comorbidity trajectory classification system based on disease clustering, such as Figure 4 As shown, it includes the following modules:

[0100] A correlation calculation module is used to obtain comorbidity trajectory data, extract all diseases in the comorbidity trajectory data, and calculate the correlation between diseases;

[0101] A disease clustering module, used to determine the attributes of the disease based on the correlation between the diseases, and to perform binary nested clustering on all diseases in the comorbidity trajectory data based on the attributes of the disease, wherein the attributes of the disease include hub diseases and non-hub diseases;

[0102] The trajectory assignment module is used to obtain the type of each comorbidity trajectory based on the disease clustering results and the attributes of each disease.

[0103] Preferably, the attributes of the disease are determined based on the correlation between the diseases, and binary nested clustering is performed on all diseases in the comorbidity trajectory data based on the attributes of the disease, including:

[0104] S20, perform binary clustering on all disease sets to obtain two sub-class disease sets;

[0105] S21. For each sub-category disease set, determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set; if the sub-category disease set has a hub disease and the number of non-hub diseases is greater than a first threshold, then treat the sub-category disease set as a subset to be clustered; otherwise, the sub-category disease set is a set in the clustering result;

[0106] S22, if there is a subset to be clustered, execute S23-S25, otherwise, end clustering;

[0107] S23. For each subset to be clustered, remove the diseases whose disease attributes are hub diseases in the subset to be clustered;

[0108] S24, performing binary classification clustering on each subset to be clustered after removing the hub disease to obtain two sub-class disease sets corresponding to each subset to be clustered;

[0109] S25. Return to step S21.

[0110] Preferably, the following method is used to determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set:

[0111] Calculate the q-quantile of the correlation between diseases in the sub-class disease set, and construct an undirected network corresponding to the sub-class disease set based on the q-quantile and the correlation between diseases;

[0112] The degree of each node in the undirected network is calculated, and the attribute of the disease corresponding to each node is determined based on the degree of each node.

[0113] Preferably, if the node degree d D>α×(|B|-1), the disease corresponding to the node is a hub disease, otherwise the disease corresponding to the node is a non-hub disease, α represents the proportion parameter, and |B| represents the number of diseases in the subclass disease set B.

[0114] The above method embodiments and system embodiments are based on the same principle, and their related parts can be used for reference, and can achieve the same technical effect. The specific implementation process refers to the above embodiment, which will not be repeated here.

[0115] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0116] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A comorbidity trajectory classification method based on disease clustering, characterized in that: The following steps are involved: Obtain comorbidity trajectory data, extract all diseases in the comorbidity trajectory data, and calculate the correlation between diseases; Determine the attributes of the disease based on the correlation between the diseases, and perform binary nested clustering on all diseases in the comorbidity trajectory data based on the attributes of the disease, wherein the attributes of the disease include hub diseases and non-hub diseases; The type of each comorbidity trajectory is obtained based on the disease clustering results and the attributes of each disease.

2. The comorbidity trajectory classification method based on disease clustering according to claim 1, characterized in that: The attributes of the disease are determined based on the correlation between the diseases, and all diseases in the comorbidity trajectory data are subjected to binary nested clustering based on the attributes of the disease, including: S20, perform binary clustering on all disease sets to obtain two sub-class disease sets; S21. For each sub-category disease set, determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set; if the sub-category disease set has a hub disease and the number of non-hub diseases is greater than a first threshold, then treat the sub-category disease set as a subset to be clustered; otherwise, the sub-category disease set is a set in the clustering result; S22, if there is a subset to be clustered, execute S23-S25, otherwise, end clustering; S23. For each subset to be clustered, remove the diseases whose disease attributes are hub diseases in the subset to be clustered; S24, performing binary classification clustering on each subset to be clustered after removing the hub disease to obtain two sub-class disease sets corresponding to each subset to be clustered; S25. Return to step S21.

3. The comorbidity trajectory classification method based on disease clustering according to claim 2 is characterized in that: The following method is used to determine the attributes of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set: Calculate the q-quantile of the correlation between diseases in the sub-class disease set, and construct an undirected network corresponding to the sub-class disease set based on the q-quantile and the correlation between diseases; The degree of each node in the undirected network is calculated, and the attribute of the disease corresponding to each node is determined based on the degree of each node.

4. The comorbidity trajectory classification method based on disease clustering according to claim 3 is characterized in that: If the degree of a node is d D >α×(|B|-1), the disease corresponding to the node is a hub disease, otherwise the disease corresponding to the node is a non-hub disease, α represents the proportion parameter, and |B| represents the number of diseases in the subclass disease set B.

5. The comorbidity trajectory classification method based on disease clustering according to claim 1, characterized in that: According to the disease clustering results and the attributes of each disease, the type of each comorbidity trajectory is obtained in the following way: For each comorbidity trajectory, if there is no non-hub disease in the comorbidity trajectory, then the comorbidity trajectory belongs to the hub-type comorbidity trajectory; otherwise: If there is only one cluster type, and more than a first number of diseases in the comorbidity trajectory belong to the cluster type, then the type of the comorbidity trajectory is the cluster type; If there are more than two cluster types, for each of the more than two cluster types, more than a first number of diseases in the comorbidity trajectory belong to the cluster type, then the cluster type to which the first non-hub disease in the comorbidity trajectory belongs is the type of the comorbidity trajectory; Otherwise, the comorbidity trajectory belongs to the mixed comorbidity trajectory.

6. The comorbidity trajectory classification method based on disease clustering according to claim 1, characterized in that: The correlation between diseases was calculated as follows: in, Denotes the i-th disease D i and the jth disease D j The correlation between Indicates that D is also included i and D j The number of comorbidity trajectories, Indicates that D is included i or D j The number of comorbidity trajectories.

7. A comorbidity trajectory classification system based on disease clustering, characterized in that: Includes the following modules: A correlation calculation module is used to obtain comorbidity trajectory data, extract all diseases in the comorbidity trajectory data, and calculate the correlation between diseases; A disease clustering module, used to determine the attributes of the disease based on the correlation between the diseases, and to perform binary nested clustering on all diseases in the comorbidity trajectory data based on the attributes of the disease, wherein the attributes of the disease include hub diseases and non-hub diseases; The trajectory assignment module is used to obtain the type of each comorbidity trajectory based on the disease clustering results and the attributes of each disease.

8. The comorbidity trajectory classification system based on disease clustering according to claim 7, characterized in that: The attributes of the disease are determined based on the correlation between the diseases, and binary nested clustering is performed on all diseases in the comorbidity trajectory data based on the attributes of the disease. The attributes of the disease include hub diseases and non-hub diseases, including: S20, perform binary clustering on all disease sets to obtain two sub-class disease sets; S21. For each sub-category disease set, determine the attribute of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set; if the sub-category disease set has a hub disease and the number of non-hub diseases is greater than a first threshold, then treat the sub-category disease set as a subset to be clustered; otherwise, the sub-category disease set is a set in the clustering result; S22, if there is a subset to be clustered, execute S23-S25, otherwise, end clustering; S23. For each subset to be clustered, remove the diseases whose disease attributes are hub diseases in the subset to be clustered; S24, performing binary classification clustering on each subset to be clustered after removing the hub disease to obtain two sub-class disease sets corresponding to each subset to be clustered; S25. Return to step S21.

9. The comorbidity trajectory classification system based on disease clustering according to claim 8, characterized in that: The following method is used to determine the attributes of each disease in the sub-category disease set based on the correlation between the diseases in the sub-category disease set: Calculate the q-quantile of the correlation between diseases in the sub-class disease set, and construct an undirected network corresponding to the sub-class disease set based on the q-quantile and the correlation between diseases; The degree of each node in the undirected network is calculated, and the attribute of the disease corresponding to each node is determined based on the degree of each node.

10. The comorbidity trajectory classification system based on disease clustering according to claim 9, characterized in that: If the degree of a node is d D >α×(|B|-1), the disease corresponding to the node is a hub disease, otherwise the disease corresponding to the node is a non-hub disease, α represents the proportion parameter, and |B| represents the number of diseases in the subclass disease set B.

Images