WO2020107909A1 - Procédé, appareil et dispositif de détermination de frais de traitement anormaux, et support de stockage informatique - Google Patents

Procédé, appareil et dispositif de détermination de frais de traitement anormaux, et support de stockage informatique Download PDF

Info

Publication number
WO2020107909A1
WO2020107909A1 PCT/CN2019/097447 CN2019097447W WO2020107909A1 WO 2020107909 A1 WO2020107909 A1 WO 2020107909A1 CN 2019097447 W CN2019097447 W CN 2019097447W WO 2020107909 A1 WO2020107909 A1 WO 2020107909A1
Authority
WO
WIPO (PCT)
Prior art keywords
group
data
current
cost
medical
Prior art date
Application number
PCT/CN2019/097447
Other languages
English (en)
Chinese (zh)
Inventor
黄越
陈明东
Original Assignee
平安医疗健康管理股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安医疗健康管理股份有限公司 filed Critical 平安医疗健康管理股份有限公司
Publication of WO2020107909A1 publication Critical patent/WO2020107909A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/08—Insurance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/018—Certifying business or products
    • G06Q30/0185—Product, service or business identity fraud
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0283—Price estimation or determination
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
    • G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records

Definitions

  • This application mainly relates to the technical field of medical systems, and in particular, to a method, device, equipment, and computer storage medium for judging the cost of abnormal medical treatment.
  • the main purpose of the present application is to provide a method, device, equipment and computer storage medium for judging the cost of abnormal medical treatment, aiming to solve the problem of lack of an effective judgment mechanism for the cost of abnormal medical treatment in the prior art.
  • the present application provides a method for judging the cost of abnormal medical treatment, which includes the following steps:
  • each of the historical medical data to multiple preset classification models for classification, and generate each classification result, wherein the multiple preset classification models are set according to different clustering algorithms, And each of the classification results is generated by the plurality of preset classification models based on the patient information, the consultation information, and the historical consultation fee in each of the historical consultation data;
  • the current medical consultation data is transmitted to the target classification model, and it is judged whether the current medical consultation fee in the current medical consultation data is abnormal.
  • the present application also proposes a device for judging abnormal medical expenses, the device for judging abnormal medical expenses includes:
  • the generating module is used to obtain multiple pieces of historical medical data, and transmit each of the historical medical data to a plurality of preset classification models for classification to generate each classification result, wherein the plurality of preset classification models are based on different A clustering algorithm is set, and each of the classification results is generated by the plurality of preset classification models based on patient information, consultation information, and historical consultation fees in each of the historical consultation data;
  • a determining module configured to detect each of the classification results, determine a target classification result, and determine the preset classification model that generates the target classification result as a target classification model;
  • the judging module is used to transmit the current medical treatment data to the target classification model when receiving the current medical treatment data regularly sent by the medical institution, and determine whether the current medical treatment fee in the current medical treatment data is abnormal.
  • the present application also proposes a device for judging abnormal medical expenses, the device for judging abnormal medical expenses includes: a memory, a processor, a communication bus, and judgment of abnormal medical expenses stored on the memory program;
  • the communication bus is used to implement connection communication between the processor and the memory
  • the processor is used to execute the procedure for judging the abnormal medical expenses, so as to realize the steps of the method for judging the abnormal medical expenses.
  • the present application also provides a computer storage medium that stores one or more programs, and the one or more programs can be executed by one or more processors as described above. The steps of the method for judging the cost of a doctor.
  • a large amount of historical visit data is used as sample data of each preset classification model, and transmitted to the preset classification model for classification to obtain the classification result classified by each preset classification model;
  • Each classification result is tested to determine the most accurate target classification result, and the preset classification model that generates the target classification result is determined as the target classification model; then the received current medical data is transmitted to the target classification model, The abnormality of the current consultation cost in the current consultation data is judged by the target classification result therein.
  • the target classification result in the target classification model is the classification of the corresponding relationship between various disease information, treatment options and cost information; it is generated from a large amount of real and effective historical visit data, with high accuracy, making it The abnormality judgment is more accurate and effective, which improves the accuracy of abnormal medical expenses as the basis for determining malicious use of medical insurance.
  • FIG. 1 is a schematic flowchart of a first embodiment of a method for judging abnormal medical expenses of the present application
  • FIG. 2 is a schematic diagram of a functional module of the first embodiment of the apparatus for judging abnormal medical expenses of the present application
  • FIG. 3 is a schematic diagram of the device structure of the hardware operating environment involved in the method of the embodiment of the present application.
  • This application provides a method for judging the cost of abnormal medical treatment.
  • FIG. 1 is a schematic flowchart of a first embodiment of a method for judging an abnormal treatment fee of an application.
  • the method for judging the abnormal medical expenses includes:
  • Step S10 Acquire multiple pieces of historical consultation data, and transmit each of the historical consultation data to multiple preset classification models for classification to generate various classification results, wherein multiple preset classification models are based on different clusters An algorithm is set, and each of the classification results is generated by a plurality of the preset classification models based on the patient information, the consultation information, and the historical consultation fee in each of the historical consultation data;
  • the method for judging the abnormal medical expenses of this application is applied to the server, and it is suitable for judging the abnormality of the medical expenses in medical institutions through the server; the medical expenses are the expenses for the patients to go to the medical institutions for certain diseases.
  • Institutions include but are not limited to general hospitals, traditional Chinese medicine hospitals, specialized hospitals and other types of hospitals, as well as clinics, health centers, pharmacies, etc.
  • the types of diseases suffered by patients with different diseases are different, and the treatment options for the same disease are also very different, which makes the cost of medical treatment different when treating various diseases with various treatment options; but for For the same type of illness and similar treatment plan, the cost of the consultation is similar; thus, the current real-time consultation cost can be judged by the usual general consultation cost of various types of diseases with various treatment plans.
  • Sex for example, when treating patients with hypertensive heart disease and heart failure under the plan A, 90% of the patients' cost of treatment was between a1 and a2, so that the cost of the consultation can be used as a reference to determine the real-time cost of treatment Is it abnormal?
  • the server sends a request to each medical institution to obtain historical visit data. After receiving the request, each medical institution transmits multiple copies of the historical visit data stored in it to the server. One of the visit data corresponds to one visit Historical visit data.
  • a plurality of preset classification models for classifying historical visit data are preset in the server. Because different preset classification models have different classification methods, the accuracy of classification results varies; in order to make the classification results more accurate For accuracy, multiple historical classification data are simultaneously classified through multiple preset classification models to obtain multiple classification results; then the classification results with the highest accuracy are determined from the multiple classification results.
  • a preset classification model is used to judge the abnormality of real-time medical expenses to ensure the accuracy of judgment.
  • the preset classification model is set according to the clustering algorithm, and the difference of the preset classification model is formed according to the difference of the clustering algorithm; such as the corresponding clustering algorithm K-Means (K-means) clustering, mean drift clustering 3.
  • K-Means K-means
  • each historical visit data is transmitted to multiple preset classification models for classification, and the steps of generating each classification result include:
  • Step S11 Transmit each of the historical visit data to a plurality of preset classification models, and read the patient information and visit information in each of the historical visit data by each of the preset classification models to transfer the patient
  • Each historical visit data whose similarity of the information and the visit information is higher than a preset value is divided into the same group category, and the historical visit cost of the historical visit data in each of the group categories is read;
  • each patient may use different treatment plans due to differences in age, gender, etc., that is, the treatment plan varies according to the patient information; thus, each historical data is transmitted to the server
  • the patient information and consultation information are read by each preset classification model;
  • the patient information is personal information that may affect the treatment plan such as the patient's age, gender, past medical history, family medical history, etc.
  • the visit information is various treatment-related data generated by the patient during the visit, such as surgical treatment or drug treatment, type of disease, type of medication, amount of medication, treatment course, etc.
  • each preset classification model compares the patient information and the consultation information between each piece of historical consultation data, and judges the similarity between each piece of historical consultation data between the patient information and the consultation information;
  • the degree of similarity between information and medical information is whether the age of each patient is within a certain range, whether the gender is the same, whether the family medical history is similar, etc.;
  • the similarity between the medical information is whether the symptoms of each patient are the same, whether the treatment is the same, Whether the type of medication is the same, whether the amount of medication is within a certain range, etc.
  • the similarity between the patient information and the similarity between the visit information represent the similarity of the treatment plan between the patients.
  • the ratio between the amount of similar information in patient information and consultation information and the total amount of judgment information can be used to determine the similarity of patient information and the similarity of consultation information; for example, the total amount of judgment information in patient information includes age, gender, There are three items of family medical history, and the amount of similar information is age and gender, the similarity is 2/3.
  • a preset value is set in advance, and the similarity between each piece of patient information obtained by comparison is compared with the preset value to determine whether the similarity of patient information is greater than the preset value. If it is larger, it means that the compared patient information has similarity, otherwise it has no similarity. At the same time, compare the similarity between the obtained medical information and the preset value to determine whether the similarity of the medical information is greater than the preset value, if it is greater, it means that the compared medical information is similar, otherwise it is not similar Sex.
  • the historical medical data are divided into the same group, that is, the historical medical data in the same group It looks like patient information and medical information.
  • the historical visit data B includes patient information b1, visit information b2, the historical visit data C includes patient information c1, visit information c2, and the historical visit data D includes patient information d1, visit information d2; It is not similar to d1, but b2 is similar to c2, and c2 is similar to d2. Due to the similar patient information b1 and c1, and the similar visit information b2 and c2, they are derived from the same historical visit data B and C , Then divide B and C into the same group.
  • each historical consultation data also includes the consultation cost of the patients.
  • the historical consultation cost of each historical consultation data in each group is read , To characterize the cost of each historical visit that can be used for reference in each group of similar patient information and visit information.
  • Step S12 Read each of the group categories divided by each of the preset classification models, and the historical medical expenses corresponding to each of the group categories to form each model group category set, and determine each model group category set as The classification result of each of the preset classification models.
  • each preset classification model divides each historical consultation data into different group categories and reads each historical consultation fee in each of the divided group categories
  • the server divides each preset classification model into Each group class and the historical visit fee corresponding to each group class are read, and a model group class set is formed between each group class corresponding to the preset classification model and each historical visit fee corresponding to each group class.
  • the model group class set is used as the classification result of the preset classification model, and one model group class set corresponds to the classification result divided by the preset classification model.
  • Step S20 detecting each of the classification results, determining a target classification result, and determining the preset classification model that generates the target classification result as a target classification model;
  • the server needs to determine the most accurate classification result from each classification result. Because the classification results are based on the patient information and medical information generated by each group category, the accuracy of the classification results and the difference between the patient information and medical information between the different group categories, and the same group of patient information and medical information Similarity is related; the greater the difference between the patient information and the consultation information between different groups, and the higher the similarity between the patient information and the consultation information in the same group category, the better the accuracy of the classification results.
  • the server detects the difference between different groups in each classification result and the similarity in the same group, and determines the most accurate target classification result in each classification result.
  • the preset classification model that generates the target classification result is determined as the target classification model, so that the target classification result in the target classification model can be used later to perform anomalies on the cost of treatment Judgment.
  • Step S30 When receiving the current medical consultation data regularly sent by the medical institution, the current medical consultation data is transmitted to the target classification model, and it is judged whether the current medical treatment fee in the current medical consultation data is abnormal.
  • the server sends an acquisition request to the medical institution to request the current medical treatment data requiring abnormality determination; or the medical institution may periodically send the required diagnosis data to the server The current medical data of abnormality judgment.
  • the server receives the current medical data, it will transmit the current medical data to the target classification model.
  • the current medical data includes the characteristic data of the medical patient's medical treatment and the current medical cost; the target classification model looks for and The group corresponding to the medical characteristic data, and then the historical medical cost for reference in the corresponding group type, to determine whether the current medical cost in the current medical data is abnormal.
  • each historical visit data includes historical visit cost
  • the numerical interval formed between the historical visit cost with the smallest value and the historical visit cost with the largest value is used as the reference interval . If the current consultation cost is within the reference interval, it means that the current consultation cost matches each historical consultation cost, and it can be determined that the current consultation cost is normal; and if the current consultation cost is not within the reference interval, it means that the current consultation cost and each historical consultation The costs do not match, and it is determined that the current cost of medical treatment is abnormal.
  • a large amount of historical visit data is used as sample data of each preset classification model, and transmitted to the preset classification model for classification to obtain the classification result classified by each preset classification model;
  • Each classification result is tested to determine the most accurate target classification result, and the preset classification model that generates the target classification result is determined as the target classification model; then the received current medical data is transmitted to the target classification model, The abnormality of the current consultation cost in the current consultation data is judged by the target classification result therein.
  • the target classification result in the target classification model is the classification of the corresponding relationship between various disease information, treatment options and cost information; it is generated from a large amount of real and effective historical visit data, with high accuracy, making it The abnormality judgment is more accurate and effective, which improves the accuracy of abnormal medical expenses as the basis for determining malicious use of medical insurance.
  • the step of detecting each of the classification results and determining the target classification results includes:
  • Step S21 Compare the inter-group spacing between each of the model group class sets to generate the inter-group comparison result of each model group class set, and determine each institute according to the inter-group comparison result The first model group class set with the largest spacing between the groups;
  • the classification results generated by each preset classification model are tested for differences between different groups and similarities in the same group, to determine the most accurate target classification result among the classification results, the The difference between different group categories is used as the inter-group spacing between each group category, reflecting the difference between the group category and the group category; the larger the interval between the groups, the greater the difference, the more the classification of each group category accurate.
  • the classification results divided by the preset classification model include three groups of M, N, and K. If the difference between M and N, M and K, and N and K is greater, the distance between groups is greater; It means that the more obvious the boundary between the three groups is, the more accurate the classification is.
  • each model group class set as a result of each classification group comparisons are performed in units of two non-repeating group classes to generate the group-to-group spacing between each group class; for the above group class M, For N and K, group comparisons are made in units of M and N, M and K, and N and K, respectively, and the spacing between the generated groups is m, n, and k, respectively.
  • group comparisons are performed in units of two non-repeating group classes to generate the group-to-group spacing between each group class; for the above group class M, For N and K, group comparisons are made in units of M and N, M and K, and N and K, respectively, and the spacing between the generated groups is m, n, and k, respectively.
  • both M and N include age and gender data, in which the age boundaries in M and N overlap, M represents the disease patients between 20 and 30, and N represents 30 ⁇ Patients with diseases between 40; and genders in M and N are different, M represents female patients with disease, and N represents male patients with disease; because the difference between the number types of "sex" in M and N is 100 %, and there is an overlap of 30 between the "age” data types, which may affect the classification, and the difference between ages is used as the interval m between the groups of M and N.
  • each model group class set After each model group class set generates the inter-group spacing between each group class, that is, after each model group class group set each group spacing, compare the spacing between each group to determine the size relationship between each group spacing ;
  • the size relationship is the generated comparison result between each group, in order to characterize the accuracy of the classification of each model group set by each model group class set by the comparison result between each group. Because the larger the inter-group spacing, the more accurate the classification.
  • the largest inter-group spacing is read from the comparison results between the groups that characterize the size relationship between the groups, and the model group class set that generates the largest inter-group spacing is used as the first model.
  • the first model group class set is the most accurate classification result among the classification results for classifying data between each group class.
  • Step S22 comparing the intra-group spacing of each group in each of the model group class sets to generate an in-group comparison result of each model group class set, and determining each of the groups according to the in-group comparison result The second model group class set with the smallest inner distance;
  • the similarity between group categories is used as the intra-group spacing in each group category, reflecting the similarity of data in the same group category; the smaller the intra-group spacing, the greater the similarity and the more accurate the classification of the data in each group category.
  • the data similarity in M, M, and K is greater, that is, the smaller the intra-group spacing; The more similar the data classification in the class, the more accurate the classification.
  • each model group class set involves multiple group classes, the data in each group class is inconsistent, so it is necessary to first determine the intra-group spacing of each group class in the same model group class set, that is, each group The similarity of the data in the class, and then the intra-group distance of the model group class set is determined by the intra-group distance of each group class.
  • group categories M, N, and K compare the data in M, N, and K, respectively, and generate the inner-group spacings of p, w, and s, respectively.
  • similarity comparison is performed on various types of data in each group, and the minimum value of the compared similarity is used as the intra-group spacing.
  • M includes 20 items of age-type data, of which 10 items are 22 years old, and 2 items are 25 years old, 8 items are 29 years old, because the 8 items of data are 29 years old and
  • the similarity between the 22-year-olds is smaller than the similarity between the two data at the age of 25 and the 22-year-old, so the similarity between the 29-year-old and the 22-year-old is regarded as the M within-group distance.
  • the intra-group spacing between each model group class set After generating the intra-group spacing for the multiple groups involved in each model group class set, compare the intra-group spacing between each model group class set to compare the maximum intra-group distance, That is, the intra-group distance of the smallest similarity group class is used as the intra-group distance of each group class in the model group class set, and represents the degree of similarity of the data of each group class divided in the model group class set.
  • each model group class set After each model group class set generates an intra-group spacing of each group class, that is, after each model group class group within each group spacing, compare the intra-group spacing to determine the size relationship between each group spacing;
  • the size relationship is the generated comparison result within each group, and the comparison result within each group is used to characterize the accuracy of the classification of each model group set by each model group class set. Because the smaller the intra-group spacing, the more accurate the classification.
  • the minimum intra-group spacing is read from the intra-group comparison results that characterize the relationship between the size of each group, and the set of model group classes that generate the smallest intra-group spacing is used as the second model Group class set, the second model group class set is the classification result with the most accurate classification of the data in each group class in the classification result.
  • Step S23 Determine whether the first model group class set and the second model group class set are the same model group class set, if they are the same model group class set, the same model The set of groups is determined as the target classification result.
  • the first model group class set is the most accurate classification result in the data classification among the group classes in the classification result
  • the second model group class set is the most accurate classification data in the group results in the classification result Classification results
  • the preset classification model that generates the same model group class set is used to classify the historical visit data
  • the data between the divided groups has a high difference, and the data in each group has a high similarity, and the classification accuracy is high.
  • the step of determining whether the first model group class set and the second model group class set represent the same model group class set includes:
  • Step S24 if the first model group class set and the second model group class set are not the same model group class set, then search for a group between the group comparison results that is greater than a first preset distance The result value, and the result value in the group that is greater than the second preset distance among the comparison results in each of the groups;
  • the server is preset with a first preset pitch for determining the size of the spacing between groups and a second preset pitch for determining the size of the spacing within the group.
  • first model group class set and the second model group class set are not When characterizing the same set of model groups, compare the comparison results between the groups that characterize the size relationship between the groups with the first preset spacing to find out the results between the groups that are larger than the first preset spacing Value; the result value between the groups represents the distance between the groups that is greater than the first preset distance in the comparison results between the groups, and the data classification between the groups has a large difference.
  • the comparison results between the groups that characterize the relationship between the distances between the groups and the second preset distance are compared to find out the result values between the groups that are greater than the second preset distance from the comparison results between the groups;
  • the interval between the groups is greater than the second preset interval, and the data classification in each group has a large similarity.
  • Step S25 when the inter-group result value and the intra-group result value are derived from the same model group class set, the same model group class set is determined as the target classification result.
  • result values between groups greater than the first preset distance and within the group values greater than the second preset distance may be derived from different sets of model group classes; thus, the result values between each group and each group Compare the internal result values to determine whether there are inter-group result values and intra-group result values from the same model group class set among the inter-group result values and the intra-group result values.
  • the classification model divides the historical visit data into groups, the data between the divided groups has a high difference, and the data in each group has a high similarity, and its classification accuracy is more high.
  • the same model group class set is determined as the target classification result, and the preset classification model corresponding to the target classification result with high representation classification accuracy is used to judge the abnormality of the current medical data to improve the accuracy of the judgment.
  • the target classification result involves multiple group categories, and each group category has multiple pieces of historical consultation data, and each piece of historical consultation data has corresponding historical consultation fees; in order to make the historical consultation fees of each group category more Accurately reflect the abnormality of the current medical data, after determining the target classification result, integrate the historical medical expenses corresponding to each group in the target classification result for reference.
  • the step of determining the target classification result includes:
  • Step S26 Generate an average historical cost of each group in the target classification result according to each historical medical expense corresponding to each group in the target classification result;
  • the same group of classes represents similar disease information and treatment plan information
  • the historical cost of each historical consultation data in the same group of classes may be different, and only the difference between the minimum historical cost and the maximum historical cost
  • the reference interval is used to judge the abnormality of the current consultation cost, which may make the judgment inaccurate due to the large range of the reference interval; therefore, the historical consultation costs corresponding to each group in the target classification result are averaged.
  • the number of historical consultation data in each group and the various historical consultation costs of each historical consultation data are counted; Add up to get the addition result, and then use the addition result and the number of historical consultation data in each group to make a ratio; the obtained ratio result is the historical cost average of each group, which represents the corresponding Reference cost of illness and treatment plan.
  • step S27 the target classification result is updated according to each historical cost average value, so as to determine whether the current consultation cost in the current consultation data is abnormal based on the updated target classification result in the target classification model.
  • each classification result reads the historical medical expenses of the group class in the generation process, and does not involve the historical cost The average value; so that the generated target classification result does not involve the average historical cost of each group. Therefore, after the historical cost average value is generated for each group in the target classification result, the target classification result is updated with the historical cost average to form a correspondence between each group class in the target classification result and the historical cost average ; In order to follow up based on the target classification results updated in the target classification model, that is, the average value of the historical costs corresponding to each group of classes, to determine whether the current cost of treatment in the current consultation data is abnormal.
  • the current medical data is transferred to the target classification model, and it is judged whether the current medical expenses in the current medical data are abnormal
  • the steps include:
  • Step S31 Transmit the current medical treatment data to the target classification model, read the medical characteristic data in the current medical treatment data from the target classification model, and combine the medical characteristic data and the target classification result Compare each group in the group to determine the target group corresponding to the medical characteristic data and the average value of the target historical costs in the target group;
  • the server transmits the current medical data to the target classification model with the most accurate classification.
  • the current consultation data includes various types of data such as the patient's age, gender, medical history, consultation time, treatment duration, disease information, medication type, and dosage;
  • the target classification model reads the consultation characteristic data from the current consultation data, and the diagnosis characteristic data is characterized Personalized data related to the treatment plan during the patient's treatment, such as the patient's personal information such as the patient's age, gender, and medical history, as well as disease information, medication type, and medication amount.
  • the target classification model divides the group according to the patient information and the consultation information in each historical consultation data, one group category corresponds to one type of patient information and consultation information; and the consultation feature data represents the patient information in the current consultation data And the visit information, so that you can compare the visit feature data and the target classification results to determine the target group category corresponding to the visit feature data.
  • the target group class has the patient information, visit information, and visit feature data The characterized patient information and the consultation information are consistent, and the average historical cost of the target group can be used to judge the abnormality of the current consultation cost in the current consultation characteristic data. Therefore, after the target group type is determined, the historical cost average value corresponding to the target group type is determined, and the average historical cost value is used as the target historical cost average value to judge the abnormality of the current medical cost.
  • Step S32 Read the cost identifier generated by the target classification model, and determine whether the current cost of the current consultation in the current medical consultation data is abnormal according to the cost identifier, where the cost identifier is determined by the target classification model The average value of the historical cost of the target and the current cost of the visit are generated by comparison.
  • the target classification model determines the target group class corresponding to the visit feature data in the current visit data and the target average historical cost of the target; the target classification model calls the target historical average cost and uses the target historical average cost Compare with the current cost of the current visit in the current visit data to determine whether the current cost of the visit is consistent with the average value of the target historical cost for reference. Consistency is characterized by the floating range of the average value of the target historical cost.
  • the floating range of the average value of the target historical cost is set to plus or minus 10; that is, when the judgment is based on the average value of the target historical cost, the current cost of treatment is at the target historical cost
  • the range between the average minus 10 and the increase of 10 is normal; otherwise, it is abnormal to ensure the accuracy of the judgment of the current cost of treatment.
  • the comparison between the average value of the target historical cost and the current cost of medical treatment will generate a judgment result of whether the current cost of the medical treatment is within the floating range of the average value of the target historical cost, and the method of adding a cost identifier to characterize the abnormality of the current cost of medical treatment; namely When the result of the judgment is that the current cost of medical treatment is not within the floating range of the average value of the target historical cost, it means that the current cost of medical treatment is abnormal, and a cost identifier characterizing the abnormality is added to the current cost of medical treatment.
  • the server reads the cost identifier generated by the comparison of the target cost model and judges whether the cost identifier is an identifier representing abnormality, that is, whether the current cost of the current consultation in the current consultation data is abnormal according to the cost identifier.
  • the step of judging whether the current consultation fee in the current consultation data is abnormal according to the fee identifier includes:
  • Step S321 judging whether the cost identifier is an abnormal identifier, if the cost identifier is an abnormal identifier, it is determined that the current cost of medical treatment in the current medical data is abnormal;
  • the server After reading the fee identifier, the server compares the fee identifier with the anomaly identifier to determine whether the fee identifier is an anomaly identifier; if the fee identifier If the symbol is an abnormal identifier, it means that the current consultation cost in the current consultation data is not within the floating range of the average value of the target historical cost, and the current consultation cost is abnormal.
  • Step S322 if the fee identifier is not an abnormal identifier, it is determined that the current fee for the current visit in the current visit data is normal.
  • the cost identifier and the abnormal identifier are compared and it is determined that the cost identifier is not an abnormal identifier, it means that the current medical cost in the current medical data is within the floating range of the average value of the target historical cost, and the current medical cost is normal.
  • the present application provides a device for judging abnormal medical expenses.
  • the device for judging abnormal medical expenses includes:
  • the generating module 10 is used to obtain multiple pieces of historical medical consultation data, and transmit each of the historical medical consultation data to a plurality of preset classification models for classification to generate each classification result, wherein the plurality of preset classification models are based on different
  • the clustering algorithm is set, and each of the classification results is generated by a plurality of the preset classification models based on the patient information, the consultation information, and the historical consultation fee in each of the historical consultation data;
  • the determining module 20 is configured to detect each of the classification results, determine a target classification result, and determine the preset classification model that generates the target classification result as the target classification model;
  • the judging module 30 is configured to transmit the current medical treatment data to the target classification model when receiving the current medical treatment data regularly sent by the medical institution, and determine whether the current medical treatment fee in the current medical treatment data is abnormal.
  • the generating module 10 uses a large amount of historical medical data as sample data of each preset classification model, and transmits it to the preset classification model for classification to obtain the classification result classified by each preset classification model
  • the determination module 20 detects each classification result to determine the most accurate target classification result, and determines the preset classification model that generates the target classification result as the target classification model; then the determination module 30 will receive the current medical treatment received
  • the data is transferred to the target classification model, and the abnormality of the current cost of treatment in the current consultation data is judged by the target classification result.
  • the target classification result in the target classification model is the classification of the corresponding relationship between various disease information, treatment options and cost information; it is generated from a large amount of real and effective historical visit data, with high accuracy, making it The abnormality judgment is more accurate and effective, which improves the accuracy of abnormal medical expenses as the basis for determining malicious use of medical insurance.
  • each virtual function module of the apparatus for determining abnormal medical expenses is stored in the memory 1005 of the apparatus for determining abnormal medical expenses shown in FIG. 3, and when the processor 1001 executes the program for determining the abnormal medical expenses, the embodiment shown in FIG. 2 is implemented The function of each module in
  • FIG. 3 is a schematic structural diagram of a device in a hardware operating environment involved in a method according to an embodiment of the present application.
  • the device for judging abnormal medical expenses in the embodiment of the present application may be a PC (personal computer, personal computer ), or terminal devices such as smart phones, tablet computers, e-book readers, and portable computers.
  • PC personal computer, personal computer
  • terminal devices such as smart phones, tablet computers, e-book readers, and portable computers.
  • the structure of the abnormal visit cost judgment device shown in FIG. 3 does not constitute a limitation on the abnormal visit cost judgment device, and may include more or fewer parts than shown, or a combination of certain Components, or different component arrangements.
  • the memory 1005 which is a computer storage medium, may include an operating system, a network communication module, and a judgment program for abnormal medical expenses.
  • the operating system is a program that manages and controls equipment hardware and software resources for judging abnormal medical expenses, and supports the operation of the abnormal medical expenses judgment program and other software and/or programs.
  • the network communication module is used to realize communication between the components inside the memory 1005, and to communicate with other hardware and software in the device for judging the cost of abnormal treatment.
  • the processor 1001 is used to execute a procedure for judging the abnormal visit fee stored in the memory 1005 to implement the steps in each embodiment of the above method for determining the abnormal visit fee.
  • the present application provides a computer storage medium.
  • the computer storage medium is preferably a computer-readable computer storage medium.
  • the computer-readable computer storage medium stores one or more programs.
  • the one or more programs may also be One or more processors execute the steps in the embodiments of the method for judging the abnormal medical expenses.
  • the methods in the above embodiments can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware, but in many cases the former is better Implementation.
  • the technical solution of the present application can be embodied in the form of a software product in essence or part that contributes to the existing technology, and the computer software product is stored in a computer storage medium (such as ROM/ The RAM, magnetic disk, and optical disk include several instructions to enable a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the embodiments of the present application.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Marketing (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Economics (AREA)
  • Theoretical Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Epidemiology (AREA)
  • Health & Medical Sciences (AREA)
  • Technology Law (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Primary Health Care (AREA)
  • Public Health (AREA)
  • Game Theory and Decision Science (AREA)
  • Medical Treatment And Welfare Office Work (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

La présente invention concerne un procédé, un appareil et un dispositif de détermination de frais de traitement anormaux, et un support de stockage informatique. Le procédé consiste à : obtenir de multiples éléments de données de traitement historiques, et transmettre chaque élément des données de traitement historiques à de multiples modèles de classification prédéfinis en vue d'une classification pour générer chaque résultat de classification ; détecter chaque résultat de classification pour déterminer un résultat de classification cible, et déterminer, comme modèle de classification cible, le modèle de classification prédéfini générant le résultat de classification cible ; et lors de la réception de données de traitement actuelles, transmettre les données de traitement actuelles au modèle de classification cible, et déterminer si les frais de traitement actuels figurant dans les données de traitement actuelles sont anormaux. Selon cette solution, le résultat de classification cible généré sur la base de la classification d'un grand volume de données de traitement dans le modèle de classification cible est la classification de correspondances parmi différentes informations de maladie, différents systèmes thérapeutiques et différentes informations de frais, et a une précision relativement élevée, si bien que la détermination d'anomalie pour les données de traitement actuelles par le modèle de classification cible est plus précise et efficace.
PCT/CN2019/097447 2018-11-30 2019-07-24 Procédé, appareil et dispositif de détermination de frais de traitement anormaux, et support de stockage informatique WO2020107909A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811462247.XA CN109670971A (zh) 2018-11-30 2018-11-30 异常就诊费用的判断方法、装置、设备及计算机存储介质
CN201811462247.X 2018-11-30

Publications (1)

Publication Number Publication Date
WO2020107909A1 true WO2020107909A1 (fr) 2020-06-04

Family

ID=66143507

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/097447 WO2020107909A1 (fr) 2018-11-30 2019-07-24 Procédé, appareil et dispositif de détermination de frais de traitement anormaux, et support de stockage informatique

Country Status (2)

Country Link
CN (1) CN109670971A (fr)
WO (1) WO2020107909A1 (fr)

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109524098A (zh) * 2018-10-27 2019-03-26 平安医疗健康管理股份有限公司 基于数据分析的就诊信息处理方法、装置、设备及介质
CN109670971A (zh) * 2018-11-30 2019-04-23 平安医疗健康管理股份有限公司 异常就诊费用的判断方法、装置、设备及计算机存储介质
CN110472114B (zh) * 2019-07-03 2024-01-26 平安科技（深圳）有限公司 异常数据预警方法、装置、计算机设备及存储介质
CN111046957B (zh) * 2019-12-13 2021-03-16 支付宝(杭州)信息技术有限公司 一种模型盗用的检测、模型的训练方法和装置
CN111127426B (zh) * 2019-12-23 2020-12-01 山东大学齐鲁医院 一种基于深度学习的胃黏膜清洁度评价方法及系统
CN111398686B (zh) * 2020-04-09 2022-04-12 国网山东省电力公司肥城市供电公司 接地电阻测量系统
CN112102098B (zh) * 2020-08-12 2023-10-27 泰康保险集团股份有限公司 数据处理方法、装置、电子设备及存储介质
CN113780855A (zh) * 2021-09-17 2021-12-10 平安医疗健康管理股份有限公司 医疗机构监管方法、装置、计算机设备及存储介质
CN113869387A (zh) * 2021-09-18 2021-12-31 平安科技（深圳）有限公司 基于人工智能技术的异常医保报销识别方法及系统
CN117316404A (zh) * 2023-09-11 2023-12-29 北京合源汇丰医药科技有限公司 一种基于ai算法的医疗信息异常检测方法及系统

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107609980A (zh) * 2017-09-07 2018-01-19 平安医疗健康管理股份有限公司 医疗数据处理方法、装置、计算机设备及存储介质
EP3358476A1 (fr) * 2016-06-14 2018-08-08 Ping An Technology (Shenzhen) Co., Ltd. Procédé et appareil pour construire un modèle de décision, dispositif informatique et dispositif de stockage
CN109670971A (zh) * 2018-11-30 2019-04-23 平安医疗健康管理股份有限公司 异常就诊费用的判断方法、装置、设备及计算机存储介质

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103488656B (zh) * 2012-06-14 2018-11-13 深圳市世纪光速信息技术有限公司 一种数据处理方法及装置
CN104182517B (zh) * 2014-08-22 2017-10-27 北京羽乐创新科技有限公司 数据处理的方法及装置
CN108229507A (zh) * 2016-12-14 2018-06-29 中国电信股份有限公司 数据分类方法以及装置
CN106875402B (zh) * 2017-01-11 2019-05-21 齐鲁工业大学 一种基于选取合适聚类数目的聚类算法的数字图像处理方法
CN107169518A (zh) * 2017-05-18 2017-09-15 北京京东金融科技控股有限公司 数据分类方法、装置、电子装置及计算机可读介质
CN107290304A (zh) * 2017-07-10 2017-10-24 天津工业大学 一种对三七及其伪品的快速鉴别方法
CN108876636B (zh) * 2018-06-19 2023-10-27 平安健康保险股份有限公司 理赔智能风控方法、系统、计算机设备及存储介质

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3358476A1 (fr) * 2016-06-14 2018-08-08 Ping An Technology (Shenzhen) Co., Ltd. Procédé et appareil pour construire un modèle de décision, dispositif informatique et dispositif de stockage
CN107609980A (zh) * 2017-09-07 2018-01-19 平安医疗健康管理股份有限公司 医疗数据处理方法、装置、计算机设备及存储介质
CN109670971A (zh) * 2018-11-30 2019-04-23 平安医疗健康管理股份有限公司 异常就诊费用的判断方法、装置、设备及计算机存储介质

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WANG, LIZHEN ET AL: "Detecting Abnormal Use of Drugs Based on Coupled Relationships", COMPUTER SCIENCE AND APPLICATION, vol. 07, no. 1, 26 January 2017 (2017-01-26), pages 88 - 99, XP009521403, DOI: 10.12677/CSA.2017.71011 *

Also Published As

Publication number Publication date
CN109670971A (zh) 2019-04-23

Similar Documents

Publication Publication Date Title
WO2020107909A1 (fr) Procédé, appareil et dispositif de détermination de frais de traitement anormaux, et support de stockage informatique
CN108899070B (zh) 处方推荐生成方法、装置、计算机设备和存储介质
WO2022145782A2 (fr) Système de prise de décision de soins médicaux d'urgence et de transport de patient en urgence par intelligence artificielle basés sur des mégadonnées et un système en nuage et procédé associé
WO2020107899A1 (fr) Procédé, dispositif et équipement de prédiction de coût médical, et support d'informations lisible par ordinateur
WO2019182297A1 (fr) Appareil et procédé de prédiction de résultat d'essai clinique
WO2020119176A1 (fr) Procédé de vérification de données de remboursement, serveur d'identification et support de stockage
Cole et al. Profiling risk factors for chronic uveitis in juvenile idiopathic arthritis: a new model for EHR-based research
CN103955608A (zh) 一种智能医疗信息远程处理系统及处理方法
WO2017065359A1 (fr) Système et procédé de soins de santé à distance utilisant un réseau de communication
WO2020108111A1 (fr) Procédé d'identification de comportement de fraude d'assurance médicale, appareil, dispositif et support de mémoire lisible
WO2020119131A1 (fr) Procédé et dispositif d'identification d'anomalies de régime médicamenteux, terminal et support de stockage lisible
CN112289437A (zh) 一种基于边缘计算架构的糖尿病辅助治疗云平台系统
KR100737382B1 (ko) 의료 진단 데이터 연산방법
CN108986873A (zh) 一种可追溯的诊疗数据处理方法及系统
Palandri et al. Telemedicine in patients with haematological diseases during the coronavirus disease 2019 (COVID‐19) pandemic: selection criteria and patients’ satisfaction
Kuo et al. Assessing association between team structure and health outcome and cost by social network analysis
WO2018169257A1 (fr) Procédé et système de gestion de données d'informations médicales personnelles
Legler et al. Evaluation of an intrahospital telemedicine program for patients admitted with COVID-19: mixed methods study
Jordan et al. Association of voice and mental health diagnoses with differences in voice‐related care utilization
CN111933283A (zh) 一种健康监测方法及装置
CN112509637A (zh) 用于交换关于基因组变异的临床意义的信息的方法和设备
US20140149140A1 (en) System and method for patient identification index
US20220036978A1 (en) Systems And Methods For Management Of Clinical Trial Electronic Health Records And Machine Learning Systems Therefor
WO2021075703A2 (fr) Méthode et système de gestion d'un symptôme de patient et de soulagement d'un symptôme sur la base d'un réseau social
Singh et al. A concept‐wide association study to identify potential risk factors for nonadherence among prevalent users of antihypertensives

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19889455

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19889455

Country of ref document: EP

Kind code of ref document: A1