A chronic disease classification labeling method based on hash bucket neighborhood
By using a hash bucket-based classification model and adaptive diagnosis radius in chronic disease diagnosis, combined with distance voting rules, the problems of high time complexity and insufficient accuracy in chronic disease diagnosis are solved, and efficient and adaptive slowness are achieved Diagnosis and prediction of diseases.
Patent Information
- Application Number
- CN202111503616.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-12-09
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2041-12-09
AI Technical Summary
In the existing chronic disease diagnosis, the neighborhood calculation volume of the patient sample is large, resulting in high time complexity. The traditional method is not adaptable to different distributions of medical data sets. Most voting rules cannot distinguish the differences between patient samples, resulting in diagnostic efficiency. Low and insufficient accuracy.
Using a hash bucket-based chronic disease classification model, the patient's samples are divided into an ordered neighborhood diagnostic bucket, the hash function is used to narrow the search range, and the adaptive diagnostic radius and distance voting rules are introduced to calculate the patient's slowness. The probability and distance of the disease are marked by chronic diseases.
It significantly improves the efficiency and accuracy of chronic disease diagnosis, avoids artificial parameter settings in traditional methods, adapts to different data sets, and enhances the ability to distinguish different chronic disease labels.
Smart Images

Figure CN115101153B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of medical detection, and in particular relates to a chronic disease classification and labeling method based on hash bucket neighborhood. Background Art
[0002] Relevant surveys in medical big data show that 10 major chronic diseases, such as diabetes, coronary heart disease, lung cancer, gastric cancer, and esophageal cancer, account for more than 40% of the total disease burden of the entire population. Faced with multi-source heterogeneous and high-noise health and medical big data, the amount of data involved in chronic disease diagnosis is showing explosive growth, involving a large amount of information about patients, diseases, and doctors, which is often incomplete and inconsistent. Diagnosing and evaluating patients only through doctors' professional knowledge and medical experience will not only consume a lot of manpower and material resources, but also be inefficient and even unable to complete the diagnosis of the disease. Therefore, data mining methods such as neighborhood rough set models can be applied to the diagnosis and prediction of chronic diseases, quickly and effectively realize the integrated prediction and early warning of multiple chronic diseases, and build an interpretable chronic disease diagnosis and medical model.
[0003] For the numerical data widely existing in medical and health big data, the traditional rough set model must first discretize it when processing, which will lead to a large amount of medical information loss and cause great obstacles in the data processing of chronic disease medicine. Professor Hu Qinghua of Tianjin University introduced the neighborhood rough set model based on the spherical neighborhood of topological space, using neighborhood information granules to replace the equivalence relationship in the traditional rough set, so that it can support both discrete medical data and numerical medical data, and further extended the neighborhood rough set to classification and prediction applications, realizing a neighborhood classifier based on the majority principle. However, when the traditional neighborhood classifier classifies and predicts patient samples, it must first construct the patient's neighborhood information granules. This process requires traversing all training patient samples in the patient sample space, resulting in low time efficiency of classifier diagnosis. Correspondingly, in the neighborhood rough set attribute reduction mechanism, Professor Liu Yong of Zhejiang University proposed an attribute fast reduction algorithm based on hash bucket partitioning to reduce the neighborhood search space. The retrieval range of the patient sample neighborhood is significantly reduced through hash mapping, effectively solving the problem of high computational time complexity of the patient neighborhood.
[0004] Faced with the problems of large amount of information, incompleteness, uncertainty, and redundancy of patient samples in medical and health big data, the existing neighborhood classifiers have the following main deficiencies in the diagnosis and prediction of chronic diseases: 1) The neighborhood calculation of patient samples is large, resulting in high time complexity in constructing the neighborhood of patient samples; 2) The setting of the neighborhood radius has artificially set random parameters and cannot be applied to medical data sets with different distributions; 3) The majority voting rule in the classifier cannot distinguish the differences between patient samples in the neighborhood, so this principle cannot be applied to the diagnosis and prediction of all patient samples. Therefore, how to effectively use the health and medical big data of the population life cycle and combine data mining methods to realize the prediction and early warning mechanism of multiple chronic diseases is an urgent problem to be solved. Summary of the invention
[0005] In order to solve the above problems existing in the existing calculation, the present invention proposes a chronic disease classification and marking method based on hash bucket neighborhood, which includes: obtaining medical data of the user to be tested, and converting the obtained medical data into a patient decision information table, inputting the data in the patient decision information table into a chronic disease classification model based on hash bucket neighborhood, and obtaining the probability of each chronic disease of the user to be tested. If the probability of at least two chronic diseases in the user to be tested is the largest and the same, then selecting the user with the largest chronic disease distance in the neighborhood for marking;
[0006] The process of using the chronic disease classification model based on hash bucket neighborhood to process the medical information data of the user to be tested and mark the chronic diseases suffered by the patient includes:
[0007] S1: converting the personal medical data of the user to be tested into a chronic disease decision information table, normalizing the medical condition attribute values in the chronic disease decision information table, and obtaining a medical condition attribute set;
[0008] S2: Determine special patient samples and hash functions based on the medical condition attribute set, and use the hash function to divide the patient samples in the patient space into a series of ordered neighborhood diagnosis buckets in turn to obtain a hash bucket sequence based on neighborhood diagnosis of chronic diseases;
[0009] S3: Determine the diagnostic neighborhood of the user to be tested according to the hash bucket sequence based on the neighborhood diagnosis of chronic diseases;
[0010] S4: Calculate the probability of each user being predicted to suffer from various chronic diseases by majority voting according to the diagnostic neighborhood of the user to be tested, and calculate and obtain the maximum probability of disease; according to the chronic disease label corresponding to the maximum probability of disease, if the chronic disease label with the highest probability in the neighborhood of the user to be tested is unique, store the patient in the set of this category; if the disease label with the highest probability in the neighborhood of the user to be tested is not unique, execute step S5;
[0011] S5: According to the distance voting principle, if there are multiple chronic disease labels with the maximum probability that meet the majority voting principle in the neighborhood of the patient to be diagnosed, calculate the distance between the above chronic disease labels in the patient's neighborhood, and then select the chronic disease label with the largest distance and mark the patient as having this type of chronic disease.
[0012] Preferably, the personal medical attribute values of the user to be tested include: systolic blood pressure, cholesterol, vitamin D, hemoglobin concentration, lymphocyte percentage, C-reactive protein and urea.
[0013] Preferably, the formula for normalizing the data in the decision information table is:
[0014]
[0015] Among them, x ij represents the value of the jth medical attribute of the i-th patient sample, max(x j ) represents the maximum value of the jth medical attribute, min(x j ) represents the minimum value of the jth medical attribute.
[0016] Preferably, the process of obtaining a hash bucket sequence for diagnosing chronic diseases based on neighborhood includes:
[0017] Step 1: extract the minimum value of each medical condition attribute in the medical condition attribute set, and determine the special patient sample according to the minimum value;
[0018] Step 2: Calculate the Euclidean distance between the patient sample and the special sample; construct a neighborhood diagnosis bucket sequence based on the Euclidean distance between the patient sample and the special sample;
[0019] Step 3: Construct a hash function and use it to map patient samples to the neighborhood diagnosis bucket sequence to obtain a hash bucket sequence based on neighborhood diagnosis of chronic diseases.
[0020] Furthermore, the Euclidean distance formula for calculating the patient sample and the special sample is:
[0021]
[0022] Among them, x i represents the i-th patient sample, x j represents the jth patient sample, m represents the number of patient samples, x ik Represents the medical attribute value of the i-th patient sample under the j-th medical attribute.
[0023] Furthermore, the constructed neighborhood diagnostic bucket is:
[0024]
[0025] Among them, x irepresents the i-th patient sample, U represents the patient sample set, ∧ represents the AND operation, Δ(x 0 ,x i ) represents the patient sample x 0 To patient sample x i The Euclidean distance, wid represents the width of the hash bucket, w represents the hash key, Indicates rounding up.
[0026] Preferably, determining the diagnostic neighborhood of the user to be tested includes: determining an adaptive neighborhood diagnostic radius according to a hash bucket sequence based on neighborhood diagnosis of chronic diseases, and determining the diagnostic neighborhood of the user to be tested according to the adaptive neighborhood diagnostic radius.
[0027] Preferably, the formula for calculating the probability of each user to be tested being predicted to suffer from various chronic diseases is:
[0028]
[0029] Among them, D k represents the set of patients with the kth chronic disease in the patient neighborhood U, δ(x) represents the diagnostic neighborhood of the patients to be diagnosed, and ∩ represents the intersection operation.
[0030] Preferably, the formula for calculating the chronic disease label with the largest distance is:
[0031]
[0032]
[0033] in, Indicates chronic disease label d k The distance degree, d k represents the chronic disease decision label that satisfies the majority voting rule in the diagnostic neighborhood δ(x), x i represents the i-th patient sample, Represents patient sample x i The distance degree, d j Indicates chronic disease label, Indicates the chronic disease label d corresponding to the maximum value k .
[0034] Beneficial effects of the present invention:
[0035] 1) By adopting the method used in the present invention, medical personnel can assist in determining whether a patient has a chronic disease based on the prediction results, which can greatly improve the efficiency of chronic disease prediction and the quality of medical diagnosis;
[0036] 2) When the traditional neighborhood classifier constructs the diagnostic neighborhood of the patient to be diagnosed, it needs to calculate the distance to all patient samples in the patient sample space. As the number of patients increases, it will face extremely high time complexity. Based on the hash bucket sequence, the present invention maps the patient samples in the patient space into a series of ordered neighborhood diagnostic buckets in sequence, and constructs the diagnostic neighborhood based on the limited patient neighborhood search range, which greatly improves the efficiency of chronic disease diagnosis;
[0037] 3) Based on the constructed hash bucket sequence for neighborhood diagnosis of chronic diseases, an adaptive neighborhood diagnosis radius based on hash buckets is defined, eliminating the artificial parameters in the traditional neighborhood radius, and can generate different diagnostic neighborhood radii corresponding to different patient samples to be diagnosed, which is adaptive to neighborhood diagnosis bucket sequences and data sets with different distributions, effectively ensuring the rationality and adaptability of chronic disease diagnosis and prediction models;
[0038] 4) The traditional voting rules do not take into account the differences between patient samples, resulting in the indistinguishability of chronic disease decision labels in traditional models during chronic disease prediction and diagnosis. The distance voting rule defined in this model gives the chronic disease decision class a new evaluation indicator: distance, which improves the ability to distinguish patients with different chronic disease labels and effectively ensures the accuracy and rationality of chronic disease diagnosis and prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 It is the neighborhood diagnosis bucket sequence distribution diagram of the present invention;
[0040] Figure 2 A diagnostic neighborhood distribution map for a patient of the present invention;
[0041] Figure 3 This is a flow chart of the chronic disease classification method based on hash bucket neighborhood of the present invention. DETAILED DESCRIPTION
[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0043] A chronic disease classification labeling method based on hash bucket neighborhood, such as Figure 3As shown, the method includes: obtaining medical data of the user to be tested, and converting the obtained medical data into a patient decision information table, inputting the data in the patient decision information table into a chronic disease classification model based on a hash bucket neighborhood, and obtaining the probability of each chronic disease of the user to be tested; if the probability of at least two chronic diseases of the user to be tested is the largest and the same, then selecting the user with the largest chronic disease distance in the neighborhood for marking.
[0044] The process of using the chronic disease classification model based on hash bucket neighborhood to process the medical information data of the user to be tested and mark the chronic diseases suffered by the patient includes:
[0045] S1: converting the personal medical data of the user to be tested into a chronic disease decision information table, normalizing the medical condition attribute values in the chronic disease decision information table, and obtaining a medical condition attribute set;
[0046] S2: Determine special patient samples and hash functions based on the medical condition attribute set, and use the hash function to divide the patient samples in the patient space into a series of ordered neighborhood diagnosis buckets in turn to obtain a hash bucket sequence based on neighborhood diagnosis of chronic diseases;
[0047] S3: Determine the diagnostic neighborhood of the user to be tested according to the hash bucket sequence based on the neighborhood diagnosis of chronic diseases;
[0048] S4: Calculate the probability of each user being predicted to suffer from various chronic diseases by majority voting according to the diagnostic neighborhood of the user to be tested, and calculate and obtain the maximum probability of disease; according to the chronic disease label corresponding to the maximum probability of disease, if the chronic disease label with the highest probability in the neighborhood of the user to be tested is unique, store the patient in the set of this category; if the disease label with the highest probability in the neighborhood of the user to be tested is not unique, execute step S5;
[0049] S5: According to the distance voting principle, if there are multiple chronic disease labels with the maximum probability that meet the majority voting principle in the neighborhood of the patient to be diagnosed, calculate the distance between the above chronic disease labels in the patient's neighborhood, and then select the chronic disease label with the largest distance and mark the patient as having this type of chronic disease.
[0050] Obtaining the personal information data of the user to be tested and converting the personal information data into a decision-making information table includes: collecting personal information of chronic disease patients from the medical health database. The collected personal information of chronic disease patients has the following medical attributes: systolic blood pressure, cholesterol, vitamin D, hemoglobin concentration, lymphocyte percentage, C-reactive protein, urea, etc. Converting patient information into a chronic disease decision-making information table<U,A,D> , where U = {x 1 ,x 2 ,...,x n} is the patient sample set; A={a 1 ,a 2 ,...,am} is a medical condition attribute set, describing the patient's symptoms and diagnosis information; D is a chronic disease decision attribute set, marking the patient's chronic disease label. In order to facilitate the display and explanation of the problem, the present invention selects two columns of data from ten patient samples for analysis, as shown in Table 1, where U = {x 1 ,x 2 ,...,x 10}, A={a 1 ,a 2}, D = {d} = {1, 2}, and "1" represents that the patient has a chronic disease, "2" represents that the patient does not have a chronic disease, and "m" represents the patient's chronic disease label to be diagnosed. In order to eliminate the influence of different medical attribute dimensions on the diagnosis effect of the patient to be diagnosed, the medical data is normalized. The normalization method is as follows:
[0051]
[0052] Among them, x ij represents the value of the jth medical attribute of the i-th patient sample, max(x j ) represents the maximum value of the jth medical attribute, min(x j ) represents the minimum value of the jth medical attribute.
[0053] In the prediction and diagnosis tasks of chronic diseases, the present invention divides the patient sample set into a patient data set U for training. train and the patient dataset U for testing test . Among them, U train Used to train the diagnostic accuracy of chronic disease prediction models and find the optimal parameters; U test It is used to evaluate the chronic disease prediction model and predict and diagnose the chronic disease labels of the patients to be diagnosed. Therefore, in the present invention, the patient sample space U is divided into U train ={x 3 ,x 4 ,...,x 10 and U test ={x 1 ,x 2}, the chronic disease decision label “m” represents the patient to be diagnosed for testing, and the normalized patient data is shown in Table 1.
[0054] Table 1. Chronic disease decision-making information table
[0055]
[0056] The process of obtaining a hash bucket sequence based on neighborhood diagnosis of chronic diseases includes:
[0057] Step 1: extract the minimum value of each medical condition attribute in the medical condition attribute set, and determine the special patient sample according to the minimum value;
[0058] Step 2: Calculate the Euclidean distance between the patient sample and the special sample; construct a neighborhood diagnosis bucket sequence based on the Euclidean distance between the patient sample and the special sample;
[0059] Step 3: Construct a hash function and use it to map patient samples to the neighborhood diagnosis bucket sequence to obtain a hash bucket sequence based on neighborhood diagnosis of chronic diseases.
[0060] Specifically, extract the minimum value of each medical condition attribute and determine the special patient sample x 0 ,Right now a(x 0 )=min(a(x i )), x i ∈U; calculate the distance between each patient sample and the special patient sample in the patient sample space. The present invention uniformly adopts the Euclidean distance. x i With x j The Euclidean distance is defined as follows:
[0061]
[0062] Among them, x i represents the i-th patient sample, x j represents the jth patient sample, m represents the number of patient samples, x ik Represents the medical attribute value of the i-th patient sample under the j-th medical attribute.
[0063] Then, we construct a hash function, set wid to represent the width of the hash bucket, and use w as the hash key. Through the hash function, we map each patient sample in the patient sample space U to a series of ordered neighborhood diagnosis buckets CB. 1 ,CB 2 ,...,CB w In the neighborhood diagnosis bucket CB w The definition is as follows:
[0064]
[0065] Among them, x i represents the i-th patient sample, U represents the patient sample set, ∧ represents the AND operation, Δ(x 0 ,x i ) represents the patient sample x 0 To patient sample x i The Euclidean distance, wid represents the width of the hash bucket, w represents the hash key, Indicates rounding up.
[0066] A specific implementation method for determining a hash bucket sequence for diagnosing chronic diseases based on a neighborhood includes: taking Table 1 as an example, first determining a special patient sample x 0 , and its medical condition attribute value is C(x 0 )={a 1 =0.10,a 2 =0.20}. Then calculate each patient sample x in the patient sample space U i With special patient sample x 0 The distance between: Δ(x i ,x 0 ),i=1,2,.... Then construct the neighborhood diagnosis bucket: Set the hash bucket width wid = 0.08, w is the hash key, and set Δ(x i ,x 0 ) divided by wid and rounded up to get w. Figure 1 As shown in the figure, the chronic disease decision attribute value d = 1 for the patient sample with a gray label, indicating that the patient has a chronic disease; the chronic disease decision attribute value d = 2 for the patient sample with a black label, indicating that the patient does not have a chronic disease. Finally, the hash function is used to map each patient sample in the patient sample space U to the following ordered neighborhood diagnosis buckets: CB 1 ={x 3 ,x 6 ,x 8 ,x 9},CB 2 ={x 7},CB 3 ={x 4 ,x 5 ,x 10}.
[0067] Determining the diagnostic neighborhood of the user to be tested includes: determining an adaptive neighborhood diagnostic radius according to a hash bucket sequence based on the neighborhood diagnosis of chronic diseases, and determining the diagnostic neighborhood of the user to be tested according to the adaptive neighborhood diagnostic radius. Specifically, the neighborhood search range of each patient sample to be diagnosed is determined according to the hash bucket sequence based on the neighborhood diagnosis of chronic diseases. If there is a patient sample to be diagnosed x i ∈CB q (q=1,2,3,...,w), then x i The patient neighborhood search range will be limited to the bucket where the patient is located and the two buckets above and below it, that is, CB q-1 ,CB q ,CB q+1 Therefore, based on the limited neighborhood bucket search range, the number of patient samples involved in the calculation of the neighborhood of patients to be diagnosed is greatly reduced, and then for any patient sample x∈CB q (q=1,2,3,...,w), this model proposes a new adaptive neighborhood diagnostic radius based on hash bucket sequence:
[0068] δ=min(Δ(x,y i ))+wid
[0069] where y i ∈CB q-1 ∪CB q ∪CB q+1 For training patient samples, min(Δ(x,y i )) represents the distance to the nearest training patient in the neighborhood of the patient to be diagnosed x, and wid represents the width of the hash bucket. Then the diagnostic neighborhood of the patient to be diagnosed x is determined:
[0070] δ(x)={y∈U|Δ(x,y)≤δ}
[0071] like Figure 2 As shown, the hash bucket sequence CB of the obtained neighborhood diagnosis of chronic diseases 1 ={x 3 ,x 6 ,x 8 ,x 9}, CB 2 ={x 7},CB 3 ={x 4 ,x 5 ,x 10}, and then determine the neighborhood search range of each patient sample to be diagnosed. 1 ∈CB 1 , then δ(x 1 ) is limited to CB 1 ∪CB 2 , due to CB 0 does not exist. Calculate Δ(x i ,x 1 ), select patient x 1 The nearest training patient in the diagnostic neighborhood of : min(Δ(x i ,x 1 ))=Δ(x 6 ,x 1 )=0.036, and then calculate x 1 Patient neighborhood radius: δ 1 =min(Δ(x i ,x 1 ))+wid = 0.116, and finally we get the patient x to be diagnosed 1 Diagnostic neighborhood of δ(x 1 )={x 1 ,x 3 ,x 6 ,x 8 ,x9}, and similarly we can get the patient x to be diagnosed 2 Diagnostic neighborhood of δ(x 2 )={x 2 ,x 7 ,x 8 ,x 9 ,x 10}.
[0072] The process of calculating the probability of each user being predicted to be ill using the majority voting principle according to the diagnostic neighborhood of the user to be tested includes: based on the diagnostic neighborhood δ(x) of the patient to be diagnosed x obtained in S3, calculating the probability of each patient to be diagnosed x being predicted to suffer from the kth chronic disease:
[0073]
[0074] Among them, D k represents the set of patients with the kth chronic disease in the patient neighborhood U, δ(x) represents the diagnostic neighborhood of the patients to be diagnosed, and ∩ represents the intersection operation.
[0075]
[0076] That is, in the diagnostic neighborhood δ(x) of the patient to be diagnosed, the chronic disease label that occupies the largest number of neighborhood patient samples and has the largest probability value will be diagnosed as this type of chronic disease, and the model prediction ends. On the contrary, if there are multiple maximum probability chronic disease labels in the diagnostic neighborhood δ(x) of the patient to be diagnosed, that is, more than one chronic disease label satisfies the majority voting principle, then step S5 is entered.
[0077] Patients waiting for diagnosis x 1 ,x 2 The diagnostic neighborhood is δ(x 1 )={x 1 ,x 3 ,x 6 ,x 8 ,x 9}, as shown in the black circle, δ(x 2 )={x 2 ,x 7 ,x 8 ,x 9 ,x 10}, as shown in the gray circle. Now calculate the patient x to be diagnosed 1 Predicted probability of having a chronic disease: Patients waiting to be diagnosed x 1 The candidate set of disease labels is Similarly, we get the patient x 2 The candidate set of disease labels is like Figure 2 As shown, in the diagnostic neighborhood δ(x1 ), the training samples with gray diseased and black non-disease labels occupy the largest number, both corresponding to the maximum value of 2; similarly, in the diagnostic neighborhood δ(x 2 ) also contains two indistinguishable chronic disease decision classes. Therefore, according to the majority voting principle, the patient x 1 ,x 2 None of the conditions can be diagnosed or predicted.
[0078] The process of calculating the chronic disease label with the largest distance based on the maximum probability chronic disease label that satisfies the majority voting principle in the neighborhood of the patient to be diagnosed includes: for several maximum probability chronic disease labels that satisfy the majority voting principle in the neighborhood of the patient to be diagnosed, that is, given any patient sample x to be diagnosed, if there are multiple chronic disease decision classes D k , k=1,2,...,m satisfies the majority voting rule in the diagnostic neighborhood δ(x), and then adopts the distance voting rule to calculate the distances of the above multiple decision classes:
[0079]
[0080] where d k , k=1,2,...,p represents a number of chronic disease decision labels that satisfy the majority voting rule in the diagnostic neighborhood δ(x), representing x i The patient samples corresponding to the above chronic disease decision classes in δ(x), x i ∈δ(x), and d(x i ) represents the patient sample x i The chronic disease decision attribute value, d(x i ) = d k .
[0081] According to the distance voting rule, the one with the largest distance among the above chronic disease decision categories is selected, that is, the chronic disease label with the largest distance is found. Finally, the patient x to be diagnosed is predicted to be diagnosed with this type of chronic disease:
[0082]
[0083] According to the neighborhood of patients to be diagnosedδ(x 1 ),δ(x 2 ) that satisfies the majority voting principle Then use the distance voting principle: first calculate the patient x 1 In the diagnostic neighborhood of The distance degree, Since these two categories satisfy the majority voting principle at the same time, the chronic disease decision label with a larger distance is selected according to the distance voting rule, so x 1 The chronic disease prediction label is d(x 1 )=2, and similarly, we can predict x2 The label is d(x 2 ) = 2. Therefore, both patients to be diagnosed are predicted to have no chronic disease.
[0084] The above embodiments further illustrate the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above embodiments are only preferred implementation modes of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A chronic disease classification and labeling method based on hash bucket neighborhood, characterized in that: include: Obtain the medical data of the user to be tested, and convert the obtained medical data into a patient decision information table, input the data in the patient decision information table into the chronic disease classification model based on the hash bucket neighborhood, and obtain the probability of each chronic disease of the user to be tested. If the probability of at least two chronic diseases of the user to be tested is the largest and the same, select the user with the largest chronic disease distance in the neighborhood for marking; The process of using the chronic disease classification model based on hash bucket neighborhood to process the medical information data of the user to be tested and mark the chronic diseases suffered by the patient includes: S1: converting the personal medical data of the user to be tested into a chronic disease decision information table, normalizing the medical condition attribute values in the chronic disease decision information table, and obtaining a medical condition attribute set; the process of converting the acquired medical data into a patient decision information table includes: normalizing the medical condition attribute values of the patient in the medical data, and obtaining the patient decision information table, and the normalization formula is: Among them, x ij represents the value of the jth medical attribute of the i-th patient sample, max(x j ) represents the maximum value of the jth medical attribute, min(x j ) represents the minimum value of the jth medical attribute; S2: Determine special patient samples and hash functions according to the medical condition attribute set, use the hash function to divide the patient samples in the patient space into neighborhood diagnosis bucket sequences in turn, and obtain a hash bucket sequence based on neighborhood diagnosis of chronic diseases; specifically: Step 1: extract the minimum value of each medical condition attribute in the medical condition attribute set, and determine the special patient sample according to the minimum value; Step 2: Calculate the Euclidean distance between the patient sample and the special sample; construct a neighborhood diagnosis bucket sequence based on the Euclidean distance between the patient sample and the special sample; Step 3: Construct a hash function and use the hash function to map the patient samples to the neighborhood diagnosis bucket sequence to obtain a hash bucket sequence based on the neighborhood diagnosis of chronic diseases; S3: Determine the diagnostic neighborhood of the user to be tested according to the hash bucket sequence based on the neighborhood diagnosis of chronic diseases; S4: Calculate the probability of each user being predicted to suffer from various chronic diseases by majority voting according to the diagnostic neighborhood of the user to be tested; calculate the chronic disease label with the largest probability value in the neighborhood of the user to be tested, and if the chronic disease label with the largest probability in the neighborhood of the user to be tested is unique, store the patient in the set of chronic disease categories; if the disease label with the largest probability in the neighborhood of the user to be tested is not unique, execute step S5; S5: Classify and label the chronic diseases of the patients to be diagnosed according to the distance voting rule. For multiple maximum probability chronic disease labels that meet the majority voting principle in the neighborhood of the patients to be diagnosed, calculate the chronic disease label with the largest distance, classify the patients according to the chronic disease labels, and store the patients in the set of chronic disease categories.
2. According to claim 1, a chronic disease classification and labeling method based on hash bucket neighborhood is characterized in that: The personal medical condition attribute values of the user to be tested include: systolic blood pressure, cholesterol, vitamin D, hemoglobin concentration, lymphocyte percentage, C-reactive protein and urea.
3. According to claim 1, a chronic disease classification and labeling method based on hash bucket neighborhood is characterized in that: The formula for calculating the Euclidean distance between patient samples and special samples is: Among them, x i represents the i-th patient sample, x j represents the jth patient sample, m represents the number of patient samples, x ik Represents the medical attribute value of the i-th patient sample under the k-th medical attribute.
4. According to claim 1, a chronic disease classification and labeling method based on hash bucket neighborhood is characterized in that: The constructed neighborhood diagnostic buckets are: Among them, x i represents the i-th patient sample, U represents the patient sample set, ∧ represents the AND operation, Δ(x0,x i ) represents patient sample x0 to patient sample x i The Euclidean distance, wid represents the width of the hash bucket, w represents the hash key, Indicates rounding up.
5. According to claim 1, a chronic disease classification and labeling method based on hash bucket neighborhood is characterized in that: Determining the diagnostic neighborhood of the user to be tested includes: determining an adaptive neighborhood diagnostic radius according to a hash bucket sequence based on neighborhood diagnosis of chronic diseases, and determining the diagnostic neighborhood of the user to be tested according to the adaptive neighborhood diagnostic radius.
6. According to claim 1, a chronic disease classification and labeling method based on hash bucket neighborhood is characterized in that: The formula for calculating the probability of each user being predicted to suffer from various chronic diseases is: Among them, D k represents the set of patients with the kth chronic disease in the patient neighborhood U, δ(x) represents the diagnostic neighborhood of the patients to be diagnosed, and ∩ represents the intersection operation.
7. According to claim 1, a chronic disease classification and labeling method based on hash bucket neighborhood is characterized in that: The formula for calculating the chronic disease label with the largest distance is: in, Indicates chronic disease label d k The distance degree, d k represents the chronic disease decision label that satisfies the majority voting rule in the diagnostic neighborhood δ(x), x i represents the i-th patient sample, Represents patient sample x i The distance degree, d j Indicates chronic disease label, Indicates the chronic disease label d corresponding to the maximum value k .
Citation Information
Patent Citations
Method and system for microbiome-derived diagnostics and therapeutics
CN107075588A
Breast cancer data classification method based on multi-granularity evidence neighborhood rough set
CN112163133A
Large-scale unbalanced diabetes electronic medical record parallel classification neighborhood evidence Spark method
CN113012776A
Medical electronic medical record information management method and system based on blockchain
CN113130028A
