Diabetes data processing and analyzing system based on machine learning

Through machine learning technology, the trend and seasonal terms of blood glucose timing data of diabetic patients are decomposed, normal indicators are calculated and reference data are screened, which solves the problem of insufficient reference data in the prior art and achieves more accurate treatment plan assistance.

CN118969314AActive Publication Date: 2024-11-15QIANJIANGYUE HEALTH IND (GUANGZHOU) CO LTD

Patent Information

Application Number
CN202411441072.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-16
Publication Date
2024-11-15
Estimated Expiration
2044-10-16

AI Technical Summary

Technical Problem

When screening reference data for diabetic patients, the reference data is not referenced only based on whether the blood sugar level is normal, and it is impossible to effectively assist doctors in formulating personalized treatment plans, which are greatly affected by individual differences and data indicators.

Method used

Using a machine learning-based diabetes data processing and analysis system, we can obtain and decompose blood glucose timing data, extract trend items and seasonal items, calculate initial and final normal indicators, and classify and screen medical record sample data based on these indicators, and build a reference database to Improve the reference value of reference data.

Benefits of technology

It improves the reference value of the screened reference data, can more accurately assist doctors in formulating targeted treatment plans, and enhances the personalization and effectiveness of treatment plans.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118969314A_ABST
    Figure CN118969314A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of medical data processing, in particular to a diabetes data processing and analyzing system based on machine learning. The system comprises a diabetes data acquisition module used for acquiring medical record sample data of a diabetic patient; the normal index analysis module is used for analyzing the change trend and the fluctuation condition of the blood glucose time sequence data in each piece of medical record sample data, and accurately quantifying the final normal index of each piece of medical record sample data in combination with numerical characteristics; and the reference database construction module is used for carrying out classification processing on all the medical record sample data based on multiple parameters in the medical record sample data, and screening the medical record sample data with high normal degree and high reference value in each category based on the obtained final normal indexes so as to construct a reference database. According to the method, the medical record sample data with relatively high reference value can be accurately screened, so that the constructed reference database can more accurately assist a doctor in making a treatment scheme.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, and in particular to a diabetes data processing and analysis system based on machine learning. Background Art

[0002] Diabetes is a chronic disease caused by absolute or relative insulin deficiency and utilization disorder, characterized by high blood sugar, and may cause a variety of complications. As a chronic disease, the treatment of diabetes is usually a long-term process. Doctors will adjust the treatment and medication plan according to the patient's specific situation to ensure the treatment effect. Over time, the scale of examination and treatment records of diabetic patients in hospital databases will gradually increase. When doctors determine the treatment plan for diabetic patients, they usually screen some medical records in the hospital database as reference data.

[0003] When selecting reference data, existing technologies usually screen based on whether blood sugar levels are at normal levels. However, due to individual differences and other factors, the changes in different medical record sample data and the various indicators in the medical record sample data will be different. Therefore, the reference value of different medical record sample data will also be different. Simply screening reference data based on whether blood sugar is at a normal level will result in the screened reference data being of low reference value and having a weak auxiliary effect on doctors formulating treatment plans. With the development of big data and machine learning technology, by using advanced data analysis tools to screen medical record sample data in hospital databases, more valuable reference data can be screened, thereby providing stronger support for doctors to determine treatment plans for diabetic patients. Summary of the invention

[0004] In order to solve the technical problem that due to individual differences, the changes in different medical record sample data and various indicators in the medical record sample data are different, so the reference value of different medical record sample data will also be different, and the reference data is only screened based on whether the blood sugar is at a normal level, which will lead to the reference of the screened reference data being weak and the auxiliary effect on doctors to formulate treatment plans is weak, the purpose of the present invention is to provide a diabetes data processing and analysis system based on machine learning, and the technical scheme adopted is as follows: The present invention proposes a diabetes data processing and analysis system based on machine learning, the system comprising: A diabetes data acquisition module is used to acquire medical record sample data of diabetic patients, wherein the medical record sample data includes multiple parameters and blood sugar time series data; The normal index analysis module is used to decompose the corresponding trend item time series data and season item time series data from each blood sugar time series data; comprehensively analyze the overall change trend of the trend item time series data and the change of the season item time series data corresponding to each blood sugar time series data, and determine the initial normal index corresponding to each medical record sample data; based on the local data fluctuation in each trend item time series data and the numerical characteristics of the data value, the initial normal index is corrected to determine the final normal index of each medical record sample data; The reference database construction module is used to classify all medical record sample data based on various parameters in the medical record sample data to obtain classification results; in each category, reference data is screened from the medical record sample data based on the final normal indicators, and a reference database is constructed based on the reference data.

[0005] Furthermore, the method for obtaining the initial normal index includes: In the trend item time series data corresponding to each blood glucose time series data, straight line fitting is performed on all data values, and the slope value of the fitting straight line is obtained, and the slope value is normalized and negatively correlated mapped to obtain a first normal factor; In the seasonal item time series data corresponding to each blood sugar time series data, all the minimum value points are obtained, and the data between two adjacent minimum value points is regarded as a change data segment; In all the changed data segments, a second normal factor is determined according to the changes between the data values ​​in the changed data segments; According to the first normal factor and the second normal factor corresponding to each blood sugar time series data, the initial normal index corresponding to each medical record sample data is calculated, and the first normal factor and the second normal factor are both positively correlated with the initial normal index.

[0006] Furthermore, the method for obtaining the second normal factor includes: In each change data segment, the difference between the maximum data value and the minimum data value is taken as the fluctuation amplitude of each change data segment; The fluctuation amplitudes of all the change data segments are arranged according to the time sequence of the change data segments to obtain a sorted sequence, and the values ​​in the sorted sequence are backward differentiated to obtain a first-order difference sequence, and the proportion of the number of positive numbers in the first-order difference sequence is used as the second normal factor.

[0007] Furthermore, the method for obtaining the final normal index includes: Based on the local data fluctuation in the trend item time series data of each blood glucose time series data, the initial normal index is preliminarily corrected to obtain the corrected normal index corresponding to each medical record sample data; Based on the numerical characteristics of the data values ​​in the trend item time series data of each blood glucose time series data, determine the abnormal factor corresponding to each medical record sample data; The abnormal factor corresponding to each medical record sample data is negatively correlated and mapped, and the normalized value is multiplied by the corresponding corrected normal index to obtain the final normal index of each medical record sample data.

[0008] Furthermore, the method for obtaining the corrected normal index includes: In each trend item time series data, the data between two adjacent minimum value points is regarded as a fluctuation data segment; Analyze the fluctuation continuity between all fluctuation data segments and determine the fluctuation continuity value; In each fluctuation data segment, the product of the difference between the maximum value and the minimum value and the length of each fluctuation data segment is used as the fluctuation factor; the fluctuation factors of all fluctuation data segments are averaged to obtain the fluctuation characteristic value; Calculate the correction factor corresponding to each medical record sample data according to the fluctuation continuous value and the fluctuation characteristic value, and the fluctuation continuous value and the fluctuation characteristic value are both negatively correlated with the correction factor; The value obtained by normalizing the product of the correction factor corresponding to each medical record sample data and the initial normal index is used as the corrected normal index of each medical record sample data.

[0009] Furthermore, the method for obtaining the fluctuation continuous value includes: The difference between the corresponding moments of the maximum values ​​of two adjacent fluctuation data segments is taken as the time difference factor. In all fluctuation data segments, the value after negative correlation mapping of the mean values ​​of all time difference factors is taken as the fluctuation continuous value.

[0010] Furthermore, the method for obtaining the abnormal factor includes: In the trend item time series data of each blood glucose time series data, the data value greater than the preset threshold is taken as the target value, and the continuous target values ​​in the time series are filtered out to obtain all subsequences; The proportion of the number of target values ​​in each trend item time series data is taken as the quantity factor, and the difference between the maximum value of all target values ​​and the preset threshold is taken as the deviation factor; the ratio of the maximum length of all subsequences to the minimum moment value of all target values ​​is taken as the time factor; The value obtained by normalizing the product of the quantity factor, the deviation factor and the time factor is used as the abnormal factor.

[0011] Furthermore, the method for obtaining the classification result includes: The parameters include at least age group, gender, presence or absence of complications, and type of diabetes; Among all the medical record sample data, the medical record sample data with the same parameters are regarded as one category, so as to obtain the classification results of all the medical record sample data.

[0012] Furthermore, the method for obtaining the reference data includes: In each category, all medical record sample data are arranged in descending order according to the final normal index to obtain a descending sequence; In the descending sequence corresponding to each category, the first preset number of medical record sample data are used as reference data.

[0013] Furthermore, decomposing corresponding trend item time series data and season item time series data from each blood sugar time series data includes: Each blood sugar time series data is decomposed based on the STL time series decomposition algorithm, so as to obtain the trend item time series data and the season item time series data corresponding to each blood sugar time series data.

[0014] The present invention has the following beneficial effects: First, obtain the medical record sample data and corresponding blood sugar time series data of diabetic patients in the hospital database to provide a data basis for subsequent data analysis. Since the medical record sample data selected as reference data should be medical record data with a higher degree of normality, the normal index analysis is performed on the blood sugar time series data corresponding to each medical record sample data. Under normal circumstances, when the treatment plan formulated by the doctor is effective, the blood sugar time series data of diabetic patients will gradually return to normal levels, so in the present invention, the overall trend and fluctuation of each blood sugar time series data can be analyzed. In view of the fact that the trend item can characterize the overall change trend of the data, and the season item can characterize the fluctuation and change pattern of the data, each blood sugar time series data is decomposed to analyze the overall change trend of the trend item time series data and the change of the season item time series data, and the normal situation of each medical record sample data is initially quantified to obtain the initial normal index. Furthermore, in view of the existence of individual differences, even if the blood sugar time series data of different medical record sample data have the same change trend as a whole, there will be different fluctuations in the local range; and the numerical characteristics of the data values ​​in the trend item time series data can most intuitively reflect the normality of the data, so for each trend item time series data, the local data fluctuation situation is combined with the numerical characteristics of the data value, so as to correct the initial normal index, and obtain the final normal index of each medical record sample data. The final normal index not only characterizes the normal situation of the medical record sample data, but also reflects the reference value of the medical record sample data. Finally, in order to facilitate the auxiliary doctors to formulate targeted plans for different categories of medical record sample data, in the present invention, the medical record sample data of all diabetic patients in the hospital database are classified to obtain classification results, so that in each category, the reference data is selected based on the final normal index to construct a reference database, so that the reference data in the reference database covers different categories and has a higher reference value, which can assist doctors in formulating more accurate treatment plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 It is a system block diagram of a diabetes data processing and analysis system based on machine learning provided by one embodiment of the present invention; Figure 2 It is a method flow chart of a method for obtaining an initial normal indicator provided by an embodiment of the present invention; Figure 3 It is a method flow chart of a method for obtaining a final normal index provided by an embodiment of the present invention; Figure 4 is a method flow chart of a method for obtaining a corrected normal index provided by an embodiment of the present invention; Figure 5 It is a system structure diagram of a diabetes data processing and analysis system based on machine learning provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the intelligent monitoring system and monitoring method of a building hanging basket proposed by the present invention, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0019] The specific scheme of a diabetes data processing and analysis system based on machine learning provided by the present invention is described in detail below with reference to the accompanying drawings.

[0020] See also Figure 1 , which shows a system block diagram of a diabetes data processing and analysis system based on machine learning provided by an embodiment of the present invention. The system includes: a diabetes data acquisition module 101, a normal indicator analysis module 102, and a reference database construction module 103.

[0021] The diabetes data acquisition module 101 is used to acquire medical record sample data of diabetic patients, and the medical record sample data includes multiple parameters and blood sugar time series data.

[0022] As a chronic disease, diabetes is usually a long-term process to treat. During the treatment process, blood sugar levels need to be controlled for a long time and checked regularly, which helps doctors adjust the treatment plan in time according to the specific situation to ensure the treatment effect. Over time, there will be a large number of medical record sample data of diabetic patients in the hospital database. In order to improve work efficiency, doctors usually refer to the treatment plan of the medical record data in the hospital database when treating diabetic patients. However, due to the huge amount of data, it is often necessary to filter out more representative reference data with higher reference value.

[0023] With the development of machine learning technology, the embodiments of the present invention can determine reference data with greater reference value by analyzing and screening the medical records of diabetic patients in the hospital database. By building a reference database with the screened reference data, doctors can be provided with more accurate and representative references, thereby assisting doctors in formulating treatment plans.

[0024] First, obtain the sample medical records of all diabetic patients in the hospital database. The sample medical records should include multiple parameters and the patient's blood glucose time series data. Blood glucose time series data can be measured and obtained at fixed time intervals; given that the types of diabetes include type 1, type 2, and gestational diabetes, the treatment plans for different types will vary; and diabetes may also cause complications, such as retinopathy, neuropathy, etc., and the manifestations of different age groups (0-12 years old, 13-19 years old, 20-39 years old, 40-59 years old, and 60 years old and above) will also be different, so the parameters in the sample medical records should at least include age group, gender, the presence or absence of complications, and the type of diabetes.

[0025] It should be noted that the types and number of parameters can be adjusted according to the implementation scenario and are not limited here; the measurement time interval of blood glucose time series data can be set to the time interval for follow-up visits of diabetic patients, and the specific time interval can be adjusted according to the implementation scenario and is not limited here.

[0026] The collection and acquisition of medical record sample data in the embodiments of the present invention are authorized by relevant users, and the process does not violate relevant laws and regulations and does not violate public order and good customs.

[0027] The normal index analysis module 102 is used to decompose the corresponding trend item time series data and season item time series data from each blood glucose time series data; comprehensively analyze the overall change trend of the trend item time series data corresponding to each blood glucose time series data and the change of the season item time series data, and determine the initial normal index corresponding to each medical record sample data; based on the local data fluctuations in each trend item time series data and the numerical characteristics of the data value, the initial normal index is corrected to determine the final normal index of each medical record sample data.

[0028] The reference data should be selected from the sample medical records of diabetic patients with better recovery conditions, and the recovery conditions of diabetic patients can be assisted by their blood glucose time series data. Therefore, for each blood glucose time series data, the embodiment of the present invention analyzes its normal condition and obtains the final normal index, which is used to evaluate the reference value of each medical record sample data. One of the criteria for evaluating the normal condition of blood glucose time series data is to judge the change trend and fluctuation of its data value. Given that the trend item can characterize the overall change trend of blood glucose level, the season item can characterize the periodic pattern and fluctuation of blood glucose time series data, and the data value in the trend item time series data can intuitively reflect the normal condition of the medical record sample data. Therefore, in the embodiment of the present invention, the blood glucose time series data is decomposed, and then the trend item time series data and the season item time series data are further analyzed and processed, and combined with the local data fluctuation in the trend item time series data of the blood glucose time series data and the numerical characteristics of the data value, the final normal index of each medical record sample data is determined to provide a basis for the subsequent construction of the reference database.

[0029] First, each blood sugar time series data is decomposed, specifically, each blood sugar time series data is decomposed based on the STL time series decomposition algorithm, so as to obtain the trend item time series data and the seasonal item time series data corresponding to each blood sugar time series data. It should be noted that the STL time series decomposition algorithm is a well-known technology, and the settings of various parameters such as the length of the seasonal smoother and the length of the trend smoother can be set according to the implementation scenario, which will not be described in detail here.

[0030] The trend item time series data reflects the long-term changing trend of blood sugar levels, which helps doctors evaluate the effectiveness of treatment plans for diabetic patients, that is, to evaluate the normal situation of medical record sample data. At the same time, the season item time series data can explain the detailed fluctuations in blood sugar levels, which is also helpful to evaluate the normal situation of medical record sample data. Therefore, the overall changing trend of the trend item time series data corresponding to each blood sugar time series data and the changes in the season item time series data are comprehensively analyzed to determine the initial normal indicators corresponding to each medical record sample data.

[0031] Preferably, in one embodiment of the present invention, the method for obtaining the initial normal index includes: See also Figure 2 , which shows a method flow chart of a method for obtaining an initial normal indicator in one embodiment of the present invention, the method comprising the following steps: Step S201: Analyze the overall data change trend in the trend item time series data corresponding to each blood glucose time series data, and quantify the first normal factor.

[0032] If the treatment plan is effective, then the data as a whole should show a downward trend in the trend item time series data corresponding to the blood sugar time series data. Therefore, in the trend item time series data corresponding to each blood sugar time series data, all data values ​​are fitted with a straight line, and the slope value of the fitted straight line is obtained. The slope value can intuitively represent the overall change trend of the data. A positive slope value indicates that the blood sugar level is rising, and the blood sugar time series data is more abnormal; a negative slope value indicates that the blood sugar level is falling, that is, the blood sugar time series data should be closer to the normal level. Then the slope value is normalized and negatively correlated to obtain the first normal factor. The formula model of the first normal factor includes: in, represents the first normal factor; represents the slope value; Represents the normalization function.

[0033] In the formula model of the first normal factor, since the slope value may be positive or negative, when the slope value is positive and the larger it is, the blood sugar level representing the blood sugar time series data is in an elevated state, that is, an abnormal state. On the contrary, when the slope value is negative and the smaller it is, the blood sugar level representing the blood sugar time series data is close to a normal state. In order to facilitate subsequent calculations, use The function normalizes the slope value. At this time, the smaller the normalized value is, the more it indicates that the blood glucose time series data is close to the normal state. Therefore, the normalized value is used Negative correlation mapping is performed, x represents the independent variable, and logical relationship correction is achieved to obtain the first normal factor. At this time, the larger the first normal factor is, the closer the blood sugar time series data is to the normal state, and the higher the normality of the medical record sample data corresponding to the blood sugar time series data is.

[0034] Step S202: In the seasonal item time series data corresponding to each blood glucose time series data, segmentation is performed based on the fluctuation of the data value to obtain the change data segment.

[0035] Seasonal item time series data usually reflects the patterns of data changes over time, which include periodic rises and falls. In order to better understand the changing patterns of data and thus quantify the normal conditions of medical record sample data, the seasonal item time series data corresponding to each blood sugar time series data can be segmented and analyzed.

[0036] In the seasonal item time series data corresponding to each blood sugar time series data, all minimum value points are obtained, and the data between two adjacent minimum value points is taken as a change data segment. At this time, each change data segment represents a change fluctuation in the seasonal item time series data.

[0037] Step S203: in all the changed data segments, determine the second normal factor according to the changes between the data values ​​in the changed data segments.

[0038] Under normal circumstances, the blood sugar level in the trend item time series data should show a downward trend, so the fluctuation in the seasonal item time series data should have a synchronous change feature, that is, the amplitude of the change will gradually decrease with the increase of time. Therefore, in each change data segment, the difference between the maximum data value and the minimum data value is used as the fluctuation amplitude of each change data segment. The larger the value of the fluctuation amplitude, the greater the fluctuation degree of the change data segment.

[0039] Then, the fluctuation amplitudes of all the changing data segments are arranged according to the time series of the changing data segments to obtain a sorted sequence, and the sorted sequence is backward differentiated to obtain a first-order difference sequence. At this time, the values ​​in the first-order difference sequence can reflect the changes in the fluctuation amplitude of the data over time. When the value is negative, it means that the fluctuation amplitude is increasing over time, and the data is more abnormal; conversely, when the value is positive, it means that the fluctuation amplitude is decreasing over time, and the possibility of normal data is higher; finally, the proportion of the number of positive numbers in the first-order difference sequence is taken as the second normal factor, that is, the ratio of the number of positive numbers in the first-order difference sequence to the total number of all values ​​in the first-order difference sequence is taken as the second normal factor. The larger the second normal factor, the higher the possibility of normality of the medical record sample data corresponding to the blood glucose time series data.

[0040] Step S204: The first normal factor and the second normal factor corresponding to each blood sugar time series data are integrated to obtain the initial normal index of the corresponding medical record sample data.

[0041] Based on the above analysis, it can be seen that the larger the first normal factor is, the closer the blood sugar time series data is to the normal state, and the higher the normality of the medical record sample data corresponding to the blood sugar time series data; the larger the second normal factor is, the higher the possibility of normality of the medical record sample data corresponding to the blood sugar time series data is, so the product of the first normal factor and the second normal factor of each blood sugar time series data is used as the initial normal indicator of the medical record sample data corresponding to each blood sugar time series data, denoted as CZ. The larger the initial normal indicator is, the higher the normality of the medical record sample data is, and the greater its reference value is.

[0042] So far, the blood glucose time series data of each medical record sample data has been preliminarily analyzed, and the initial normal index of each medical record sample data can be obtained.

[0043] When analyzing the trend of blood sugar levels in the process of obtaining the initial normal indicators, only the overall change trend of the trend item time series data corresponding to the blood sugar time series data is analyzed. However, due to individual differences, different medical record sample data may have consistent overall trends, but local change fluctuations may be different; at the same time, when the data fluctuates, some fluctuations may be abnormally drastic, and have a tendency to worsen compared to the initial or average state of blood sugar levels, that is, there are large data values ​​in the trend item time series data, which also indicates that the normality of the medical record sample data is low. Therefore, the initial normal indicators of the medical record sample data are corrected based on the local data fluctuations in the trend item time series data and the numerical characteristics of the data values, so as to determine the final normal indicators of each medical record sample data.

[0044] Preferably, in one embodiment of the present invention, the method for obtaining the final normal index includes: See also Figure 3 , which shows a method flow chart of a method for obtaining a final normal index in one embodiment of the present invention, the method comprising the following steps: Step S301: Based on the local data fluctuation in the trend item time series data of each blood glucose time series data pair, the initial normal index is preliminarily corrected to obtain the corrected normal index corresponding to each medical record sample data.

[0045] See also Figure 4 , which shows a method flow chart of a method for obtaining a corrected normal index in one embodiment of the present invention, the method comprising the following steps: Step S3011: Divide the trend item time series data corresponding to each blood sugar time series data to obtain fluctuation data segments.

[0046] In the trend item time series data, the data between two adjacent minimum value points is taken as a fluctuation data segment, and each fluctuation data segment corresponds to a data fluctuation in the trend item time series data.

[0047] Step S3012: Analyze the fluctuation continuity between all fluctuation data segments and determine the fluctuation continuity value.

[0048] When the fluctuation of data is more continuous, it is considered that the blood sugar level frequently rises and falls, and the abnormality is higher and the normality is lower. Therefore, the continuous fluctuation between the fluctuation data segments is considered here. It can be characterized by the time difference between adjacent fluctuation data segments: the difference between the corresponding moments of the maximum values ​​of two adjacent fluctuation data segments is used as the time difference factor. At this time, there is a time difference factor between each two adjacent fluctuation data segments in the time series, and the smaller the time difference factor, the more continuous the fluctuation. Therefore, the value after negative correlation mapping of the mean of all time difference factors is used as the fluctuation continuity value. The formula model of the fluctuation continuity value includes: in, Indicates a fluctuating continuous value; represents the number of time difference factors; represents the value of the i-th time difference factor; Indicates the preset first constant.

[0049] In the formula model of the continuous value of fluctuation, the smaller the time difference factor is, the larger the continuous value of fluctuation is. Therefore, the mean of the time difference factors between all fluctuation data segments is used as the denominator to achieve logical relationship correction and obtain the continuous value of fluctuation. The larger the continuous value of fluctuation is, the more continuous the fluctuation of the data is. The more frequent the increase and decrease of blood sugar level is, the lower the normality of the corresponding medical record sample data is.

[0050] It should be noted that the first constant is preset The function of is to prevent the denominator from being 0. The value can be 0.001. The specific value can be adjusted according to the implementation scenario and is not limited here.

[0051] In other embodiments of the present invention, the formula model of the fluctuation continuous value may also be: ;use Implement negative correlation mapping, where It represents an exponential function with the natural constant e as the base, and x represents the independent variable.

[0052] Step S3013: Determine the fluctuation characteristic value by comprehensively considering the differences between the data values ​​in each fluctuation data segment and the length of the fluctuation data segment.

[0053] Not only can the continuity of fluctuations be used as a criterion to measure whether the medical record sample data is normal or not, the amplitude and duration of fluctuations can also be used as an indicator to measure the normality of medical record sample data.

[0054] Therefore, in each fluctuation data segment, the product of the difference between the maximum value and the minimum value and the length of each fluctuation data segment is taken as the fluctuation factor. The larger the difference between the maximum value and the minimum value, the larger the amplitude of the fluctuation, which indicates that the degree of repeated blood sugar levels is more serious. At the same time, the longer the length of the fluctuation data segment, the longer the duration of the fluctuation. Therefore, the larger the fluctuation factor of the fluctuation data segment, the lower the normality of the corresponding medical record sample data.

[0055] Finally, the fluctuation factors of all fluctuation data segments are averaged to obtain the fluctuation characteristic value, denoted as BT. Similarly, the larger the fluctuation characteristic value, the lower the normality of the corresponding medical record sample data, and the lower the reference value.

[0056] Step S3014: Correct the initial normal index based on the fluctuation continuous value and the fluctuation characteristic value to obtain a corrected normal index.

[0057] Based on the analysis in the previous steps, it can be seen that the larger the fluctuation continuous value, the lower the normality of the corresponding medical record sample data; similarly, the larger the fluctuation characteristic value, the lower the normality of the corresponding medical record sample data, and the lower the reference value. Therefore, the correction factor corresponding to each medical record sample data is calculated based on the fluctuation continuous value and the fluctuation characteristic value, and the initial normal index is initially corrected. The corrected normal index of each medical record sample data is obtained. The formula model for correcting the normal index includes: in, Represents the corrected normal index of each medical record sample data; Represents the initial normal index of each medical record sample data; Indicates the fluctuating continuous value corresponding to each medical record sample data; Indicates the fluctuation characteristic value corresponding to each medical record sample data; It represents the exponential function with the natural constant e as the base; Represents the normalization function.

[0058] In the formula model for correcting normal indicators, when the initial normal indicator of the medical record sample data is larger, it means that the normality of the medical record sample data is higher, that is, the reference value is greater; and when the fluctuation continuous value corresponding to the medical record sample data is larger and the fluctuation characteristic value is larger, its normality will decrease and the reference value will also decrease. Therefore, the product of the fluctuation continuous value and the fluctuation characteristic value corresponding to the medical record sample data is negatively correlated and normalized to obtain the correction factor The smaller the correction factor, the lower the reference value of the medical record sample data should be. Therefore, the correction factor corresponding to each medical record sample data is multiplied by the initial normal index, and the resulting product is normalized to obtain the corrected normal index. The larger the corrected normal index, the higher the normality of the medical record sample data and the greater its reference value.

[0059] Step S302: Determine the abnormal factor corresponding to each medical record sample data based on the numerical characteristics of the data values ​​in the trend item time series data of each blood glucose time series data.

[0060] First, identify the abnormal data points. In the trend item time series data of each blood sugar time series data, the data value greater than the preset threshold is used as the target value (abnormal data point). The target value represents the data point that tends to be abnormal in the trend item time series data, and is the basis for measuring the abnormal factors of the medical record sample data. It should be noted that since the data value at the starting position in the trend item time series data can represent the starting state of the data, when screening abnormal data points, each data value is compared with the data value at the starting time, which can better reflect its abnormality. Therefore, in this embodiment of the present invention, the preset threshold is set to the data value at the starting position in the trend item time series data.

[0061] Then, the target values ​​that are continuous in time series are screened out to obtain all subsequences, which helps to identify continuous abnormal patterns rather than isolated abnormal data points. For example, the target values ​​correspond to the moments 4, 7, 8, and 10, respectively, and the data values ​​corresponding to moments 7 and 8 can form a subsequence. It should be noted that in actual applications, if there are no target values ​​that are continuous in time series, that is, if there is no subsequence with a length greater than 1, a single target value can also be used as a subsequence.

[0062] The number of target values ​​can be used as an indicator to measure data anomalies, so the proportion of the number of target values ​​in each trend item time series data is taken as the quantity factor, that is, the ratio of the total number of target values ​​to the total number of data values ​​in the trend item time series data. The larger the quantity factor, the more target values ​​there are in the trend item time series data, that is, the higher the degree of data anomaly. The difference between the maximum value of all target values ​​and the preset threshold is taken as the deviation factor. The larger the deviation factor, the higher the degree to which the maximum target value exceeds the preset threshold, and the higher the degree of anomaly. The length of the subsequence can represent the duration of the anomaly, so the ratio of the maximum length of all subsequences to the minimum value of all target values ​​is taken as the time factor. The minimum value of the target value represents the starting position of the initial data abnormal fluctuation. The earlier the abnormal fluctuation occurs, the more serious the data abnormality is. Therefore, the larger the time factor, the higher the degree of data anomaly.

[0063] Finally, the product of the quantity factor, deviation factor and time factor is normalized to obtain the value after normalization, which is taken as the abnormality factor, denoted as YY. The larger the abnormality factor, the higher the degree of abnormality of the data. It should be noted that the normalization method here can be used Normalization function.

[0064] Step S303: Calculate the final normal index by combining the abnormal factors corresponding to each medical record sample data and the corrected normal index.

[0065] The abnormal factors corresponding to each medical record sample data characterize the abnormal situation of the medical record sample data, so the abnormal factors are used to further correct the corrected normal indicators of the medical record sample data to obtain the final normal indicators. The formula model of the final normal indicators includes: in, Represents the final normal index of each medical record sample data; Represents the corrected normal index of each medical record sample data; Indicates the abnormal factor corresponding to each medical record sample data; It represents the exponential function with the natural constant e as the base; Represents the normalization function.

[0066] In the formula model of the final normal index, based on step S302, it can be seen that the larger the abnormal factor is, the higher the abnormal degree of the medical record sample data is. Therefore, the abnormal factor corresponding to each medical record sample data is negatively correlated and normalized. The reason for negative correlation mapping and normalization is to achieve logical relationship correction. At this time, the corrected value The larger the value is, the more normal the medical record sample data is. The data is then multiplied by the corresponding modified normal index, and the resulting product is normalized to obtain the final normal index for each medical record sample data. The larger the final normal index is, the more normal the medical record sample data is and the greater its reference value is.

[0067] At this point, after all the above steps, the final normal index of the medical record sample data of each diabetic patient in the hospital database can be obtained. The larger the final normal index, the closer the blood sugar level of the medical record sample data is to the normal level, and the treatment plan in the medical record sample data will have more reference value.

[0068] The reference database construction module 103 is used to classify all medical record sample data based on various parameters in the medical record sample data to obtain classification results; in each category, reference data is selected from the medical record sample data based on the final normal indicators, and a reference database is constructed based on the reference data.

[0069] Since there are a large number of medical record sample data in the hospital database, they are classified based on various parameters in the medical record sample data so that the medical record sample data in each category in the classification results have relatively similar characteristics. Then, in each category, medical record sample data with more reference value are screened out as reference data based on the final normal indicators. This can make the reference data more comprehensive, and ultimately the reference database constructed based on the reference data will be more representative and have a higher reference value.

[0070] Firstly, the medical record sample data is classified based on various parameters to obtain the classification results.

[0071] Preferably, in one embodiment of the present invention, the method for obtaining the classification result includes: In the diabetes data acquisition module 101, when acquiring medical record sample data, the medical record sample data includes multiple parameters, such as age group, gender, presence or absence of complications and diabetes type, and all medical record sample data are classified according to these parameters.

[0072] Among all the medical record sample data, the medical record sample data with the same parameters are classified as one category. For example, the age range of 0-12 years old is marked as 1, the age range of 13-19 years old is marked as 2, the age range of 20-39 years old is marked as 3, the age range of 40-59 years old is marked as 4, and the age range of 60 years old and above is marked as 5; the gender of male is marked as 1, and the gender of female is marked as 2; the presence of complications is marked as 1, and the absence of complications is marked as 2; type 1 diabetes is marked as 1, type 2 diabetes is marked as 2, and gestational diabetes is marked as 3; if the parameters in a medical record sample data are 60 years old and above, male, complications, and type 1 diabetes, then the medical record sample data corresponds to the sequence [5,1,1,1]. Based on this method, the sequence corresponding to each medical record sample data can be obtained, so that the medical record sample data corresponding to the same sequence can be classified into one category.

[0073] After obtaining the classification results, reference data can be filtered out in each category based on the final normal indicators.

[0074] Preferably, the method for obtaining reference data includes: The larger the normal index is, the more normal the medical record sample data is and the greater its reference value is. Therefore, in each category, all medical record sample data are arranged in descending order according to the final normal index to obtain a descending sequence.

[0075] Then, in the descending sequence corresponding to each category, the first preset number of medical record sample data is selected as reference data. It should be noted that the preset number is set to one-fourth of the total number of medical record sample data in the corresponding category, rounded up, and the specific value can be adjusted according to the implementation scenario and is not limited here.

[0076] After obtaining all the reference data, you can use all the reference data to build a reference database, which contains different categories of medical record sample data. The reference database supports classified search, and the search results can display the diabetes-related data and treatment plans of the reference data, which can help doctors refer to their treatment plans to develop new treatment plans.

[0077] In summary, firstly, the medical record sample data and the corresponding blood sugar time series data of the diabetic patients in the hospital database are obtained to provide a data basis for the subsequent data analysis. Since the medical record sample data selected as the reference data should be medical record data with a higher degree of normality, the normal index analysis is performed on the blood sugar time series data corresponding to each medical record sample data. Under normal circumstances, when the treatment plan formulated by the doctor is effective, the blood sugar time series data of the diabetic patient will gradually return to a normal level, so in the present invention, the overall trend and fluctuation of each blood sugar time series data can be analyzed. In view of the fact that the trend item can characterize the overall change trend of the data, and the season item can characterize the fluctuation and change pattern of the data, each blood sugar time series data is decomposed to analyze the overall change trend of the trend item time series data and the change of the season item time series data, and the normal situation of each medical record sample data is initially quantified to obtain the initial normal index. Furthermore, in view of the existence of individual differences, even if the blood sugar time series data of different medical record sample data have the same change trend as a whole, there will be different fluctuations in the local range; and the numerical characteristics of the data values ​​in the trend item time series data can most intuitively reflect the normality of the data, so for each trend item time series data, the local data fluctuation situation is combined with the numerical characteristics of the data value, so as to correct the initial normal index, and obtain the final normal index of each medical record sample data. The final normal index not only characterizes the normal situation of the medical record sample data, but also reflects the reference value of the medical record sample data. Finally, in order to facilitate the auxiliary doctors to formulate targeted plans for different categories of medical record sample data, in the present invention, the medical record sample data of all diabetic patients in the hospital database are classified to obtain classification results, so that in each category, the reference data is selected based on the final normal index to construct a reference database, so that the reference data in the reference database covers different categories and has a higher reference value, which can assist doctors in formulating more accurate treatment plans.

[0078] See also Figure 5, which shows a system structure diagram of a diabetes data processing and analysis system based on machine learning provided by an embodiment of the present invention, including a processor 500, a memory 501, a bus 502 and a communication interface 503, wherein the processor 500, the communication interface 503 and the memory 501 are connected via the bus 502; wherein the memory 501 may include a high-speed random access memory, the bus 502 may be an ISA bus, a PCI bus or an EISA bus, etc., and the processor 500 may be an integrated circuit chip having signal processing capabilities.

[0079] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0080] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0081] 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 in the protection scope of the present invention.

Claims

1. A diabetes data processing and analysis system based on machine learning, characterized in that: The system includes: A diabetes data acquisition module is used to acquire medical record sample data of diabetic patients, wherein the medical record sample data includes multiple parameters and blood sugar time series data; The normal index analysis module is used to decompose the corresponding trend item time series data and season item time series data from each blood sugar time series data; comprehensively analyze the overall change trend of the trend item time series data and the change of the season item time series data corresponding to each blood sugar time series data, and determine the initial normal index corresponding to each medical record sample data; based on the local data fluctuation in each trend item time series data and the numerical characteristics of the data value, the initial normal index is corrected to determine the final normal index of each medical record sample data; The reference database construction module is used to classify all medical record sample data based on various parameters in the medical record sample data to obtain classification results; in each category, reference data is selected from the medical record sample data based on the final normal indicators, and a reference database is constructed based on the reference data; The method for obtaining the initial normal index includes: In the trend item time series data corresponding to each blood glucose time series data, straight line fitting is performed on all data values, and the slope value of the fitting straight line is obtained, and the slope value is normalized and negatively correlated mapped to obtain a first normal factor; In the seasonal item time series data corresponding to each blood sugar time series data, all the minimum value points are obtained, and the data between two adjacent minimum value points is regarded as a change data segment; In all the changed data segments, a second normal factor is determined according to the changes between the data values ​​in the changed data segments; According to the first normal factor and the second normal factor corresponding to each blood sugar time series data, the initial normal index corresponding to each medical record sample data is calculated, and the first normal factor and the second normal factor are both positively correlated with the initial normal index; The method for obtaining the second normal factor includes: In each change data segment, the difference between the maximum data value and the minimum data value is taken as the fluctuation amplitude of each change data segment; The fluctuation amplitudes of all the change data segments are arranged according to the time sequence of the change data segments to obtain a sorted sequence, and the values ​​in the sorted sequence are backward differentiated to obtain a first-order difference sequence, and the proportion of the number of positive numbers in the first-order difference sequence is used as the second normal factor.

2. A diabetes data processing and analysis system based on machine learning according to claim 1, characterized in that: The method for obtaining the final normal index includes: Based on the local data fluctuation in the trend item time series data of each blood glucose time series data, the initial normal index is preliminarily corrected to obtain the corrected normal index corresponding to each medical record sample data; Based on the numerical characteristics of the data values ​​in the trend item time series data of each blood glucose time series data, determine the abnormal factor corresponding to each medical record sample data; The abnormal factor corresponding to each medical record sample data is negatively correlated and mapped, and the normalized value is multiplied by the corresponding corrected normal index to obtain the final normal index of each medical record sample data.

3. A diabetes data processing and analysis system based on machine learning according to claim 2, characterized in that: The method for obtaining the corrected normal index includes: In each trend item time series data, the data between two adjacent minimum value points is regarded as a fluctuation data segment; Analyze the fluctuation continuity between all fluctuation data segments and determine the fluctuation continuity value; In each fluctuation data segment, the product of the difference between the maximum value and the minimum value and the length of each fluctuation data segment is used as the fluctuation factor; the fluctuation factors of all fluctuation data segments are averaged to obtain the fluctuation characteristic value; Calculate the correction factor corresponding to each medical record sample data according to the fluctuation continuous value and the fluctuation characteristic value, and the fluctuation continuous value and the fluctuation characteristic value are both negatively correlated with the correction factor; The product of the correction factor corresponding to each medical record sample data and the initial normal index is normalized to a value, which is used as the corrected normal index of each medical record sample data.

4. A diabetes data processing and analysis system based on machine learning according to claim 3, characterized in that: The method for obtaining the fluctuation continuous value includes: The difference between the corresponding moments of the maximum values ​​of two adjacent fluctuation data segments is taken as the time difference factor. In all fluctuation data segments, the value after negative correlation mapping of the mean values ​​of all time difference factors is taken as the fluctuation continuous value.

5. A diabetes data processing and analysis system based on machine learning according to claim 2, characterized in that: The method for obtaining the abnormal factor includes: In the trend item time series data of each blood glucose time series data, the data value greater than the preset threshold is taken as the target value, and the continuous target values ​​in the time series are filtered out to obtain all subsequences; The proportion of the number of target values ​​in each trend item time series data is taken as the quantity factor, and the difference between the maximum value of all target values ​​and the preset threshold is taken as the deviation factor; the ratio of the maximum length of all subsequences to the minimum moment value of all target values ​​is taken as the time factor; The value obtained by normalizing the product of the quantity factor, the deviation factor and the time factor is used as the abnormal factor.

6. A diabetes data processing and analysis system based on machine learning according to claim 1, characterized in that: The method for obtaining the classification result includes: The parameters include at least age group, gender, presence or absence of complications, and type of diabetes; Among all the medical record sample data, the medical record sample data with the same parameters are regarded as one category, so as to obtain the classification results of all the medical record sample data.

7. The diabetes data processing and analysis system based on machine learning according to claim 1, characterized in that: The method for obtaining the reference data comprises: In each category, all medical record sample data are arranged in descending order according to the final normal index to obtain a descending sequence; In the descending sequence corresponding to each category, the first preset number of medical record sample data are used as reference data.

8. The diabetes data processing and analysis system based on machine learning according to claim 1, characterized in that: Decomposing corresponding trend item time series data and season item time series data from each blood sugar time series data includes: Each blood sugar time series data is decomposed based on the STL time series decomposition algorithm, so as to obtain the trend item time series data and the season item time series data corresponding to each blood sugar time series data.

Citation Information

Patent Citations

  • Diabetes digital health management system based on big data and artificial intelligence

    CN115274140A

  • Diabetes risk assessment system

    CN118737447A

  • Blood glucose prediction method and apparatus, and blood glucose level measurement system

    WO2023115751A1

Cited By

  • Medical examination medical record information arrangement method for cardiac surgery department

    CN119314694A

  • Information sharing method of pediatric clinical nursing system

    CN119400337A

  • Intelligent blood glucose monitoring method and system for nephrology department

    CN119851964A

  • Patient diet digital visual management method and system based on multi-source data

    CN120148766A

  • RTO real-time data processing and analyzing system fused with edge calculation

    CN120296507A